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

Comparing Coro/Coro.pm (file contents):
Revision 1.198 by root, Sun Sep 21 01:23:26 2008 UTC vs.
Revision 1.259 by root, Tue Jun 30 08:28:55 2009 UTC

1=head1 NAME 1=head1 NAME
2 2
3Coro - coroutine process abstraction 3Coro - the only real threads in perl
4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 use Coro; 7 use Coro;
8 8
… …
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 coro
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 use Coro::Semaphore;
… …
26 $locked = 1; 26 $locked = 1;
27 $lock->up; 27 $lock->up;
28 28
29=head1 DESCRIPTION 29=head1 DESCRIPTION
30 30
31This module collection manages coroutines. Coroutines are similar to 31For a tutorial-style introduction, please read the L<Coro::Intro>
32threads but don't (in general) run in parallel at the same time even 32manpage. This manpage mainly contains reference information.
33on SMP machines. The specific flavor of coroutine used in this module
34also guarantees you that it will not switch between coroutines unless
35necessary, at easily-identified points in your program, so locking and
36parallel access are rarely an issue, making coroutine programming much
37safer and easier than threads programming.
38 33
39Unlike a normal perl program, however, coroutines allow you to have 34This module collection manages continuations in general, most often in
40multiple running interpreters that share data, which is especially useful 35the form of cooperative threads (also called coros, or simply "coro"
41to code pseudo-parallel processes and for event-based programming, such as 36in the documentation). They are similar to kernel threads but don't (in
42multiple HTTP-GET requests running concurrently. See L<Coro::AnyEvent> to 37general) run in parallel at the same time even on SMP machines. The
43learn more. 38specific flavor of thread offered by this module also guarantees you that
39it will not switch between threads unless necessary, at easily-identified
40points in your program, so locking and parallel access are rarely an
41issue, making thread programming much safer and easier than using other
42thread models.
44 43
45Coroutines are also useful because Perl has no support for threads (the so 44Unlike the so-called "Perl threads" (which are not actually real threads
46called "threads" that perl offers are nothing more than the (bad) process 45but only the windows process emulation ported to unix, and as such act
47emulation coming from the Windows platform: On standard operating systems 46as processes), Coro provides a full shared address space, which makes
48they serve no purpose whatsoever, except by making your programs slow and 47communication between threads very easy. And Coro's threads are fast,
49making them use a lot of memory. Best disable them when building perl, or 48too: disabling the Windows process emulation code in your perl and using
50aks your software vendor/distributor to do it for you). 49Coro can easily result in a two to four times speed increase for your
50programs. A parallel matrix multiplication benchmark runs over 300 times
51faster on a single core than perl's pseudo-threads on a quad core using
52all four cores.
51 53
54Coro achieves that by supporting multiple running interpreters that share
55data, which is especially useful to code pseudo-parallel processes and
56for event-based programming, such as multiple HTTP-GET requests running
57concurrently. See L<Coro::AnyEvent> to learn more on how to integrate Coro
58into an event-based environment.
59
52In this module, coroutines are defined as "callchain + lexical variables + 60In this module, a thread is defined as "callchain + lexical variables +
53@_ + $_ + $@ + $/ + C stack), that is, a coroutine has its own callchain, 61some package variables + C stack), that is, a thread has its own callchain,
54its own set of lexicals and its own set of perls most important global 62its own set of lexicals and its own set of perls most important global
55variables (see L<Coro::State> for more configuration). 63variables (see L<Coro::State> for more configuration and background info).
64
65See also the C<SEE ALSO> section at the end of this document - the Coro
66module family is quite large.
56 67
57=cut 68=cut
58 69
59package Coro; 70package Coro;
60 71
61use strict; 72use strict qw(vars subs);
62no warnings "uninitialized"; 73no warnings "uninitialized";
63 74
75use Guard ();
76
64use Coro::State; 77use Coro::State;
65 78
66use base qw(Coro::State Exporter); 79use base qw(Coro::State Exporter);
67 80
68our $idle; # idle handler 81our $idle; # idle handler
69our $main; # main coroutine 82our $main; # main coro
70our $current; # current coroutine 83our $current; # current coro
71 84
72our $VERSION = 4.746; 85our $VERSION = 5.15;
73 86
74our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub); 87our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub);
75our %EXPORT_TAGS = ( 88our %EXPORT_TAGS = (
76 prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)], 89 prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)],
77); 90);
78our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready)); 91our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready));
79 92
93=head1 GLOBAL VARIABLES
94
80=over 4 95=over 4
81 96
82=item $Coro::main 97=item $Coro::main
83 98
84This variable stores the coroutine object that represents the main 99This variable stores the Coro object that represents the main
85program. While you cna C<ready> it and do most other things you can do to 100program. While you cna C<ready> it and do most other things you can do to
86coroutines, it is mainly useful to compare again C<$Coro::current>, to see 101coro, it is mainly useful to compare again C<$Coro::current>, to see
87whether you are running in the main program or not. 102whether you are running in the main program or not.
88 103
89=cut 104=cut
90 105
91$main = new Coro; 106# $main is now being initialised by Coro::State
92 107
93=item $Coro::current 108=item $Coro::current
94 109
95The coroutine object representing the current coroutine (the last 110The Coro object representing the current coro (the last
96coroutine that the Coro scheduler switched to). The initial value is 111coro that the Coro scheduler switched to). The initial value is
97C<$main> (of course). 112C<$Coro::main> (of course).
98 113
99This variable is B<strictly> I<read-only>. You can take copies of the 114This variable is B<strictly> I<read-only>. You can take copies of the
100value stored in it and use it as any other coroutine object, but you must 115value stored in it and use it as any other Coro object, but you must
101not otherwise modify the variable itself. 116not otherwise modify the variable itself.
102 117
103=cut 118=cut
104 119
105$main->{desc} = "[main::]";
106
107# maybe some other module used Coro::Specific before...
108$main->{_specific} = $current->{_specific}
109 if $current;
110
111_set_current $main;
112
113sub current() { $current } # [DEPRECATED] 120sub current() { $current } # [DEPRECATED]
114 121
115=item $Coro::idle 122=item $Coro::idle
116 123
117This variable is mainly useful to integrate Coro into event loops. It is 124This variable is mainly useful to integrate Coro into event loops. It is
118usually better to rely on L<Coro::AnyEvent> or LC<Coro::EV>, as this is 125usually better to rely on L<Coro::AnyEvent> or L<Coro::EV>, as this is
119pretty low-level functionality. 126pretty low-level functionality.
120 127
121This variable stores a callback that is called whenever the scheduler 128This variable stores either a Coro object or a callback.
129
130If it is a callback, the it is called whenever the scheduler finds no
122finds no ready coroutines to run. The default implementation prints 131ready coros to run. The default implementation prints "FATAL:
123"FATAL: deadlock detected" and exits, because the program has no other way 132deadlock detected" and exits, because the program has no other way to
124to continue. 133continue.
125 134
135If it is a coro object, then this object will be readied (without
136invoking any ready hooks, however) when the scheduler finds no other ready
137coros to run.
138
126This hook is overwritten by modules such as C<Coro::Timer> and 139This hook is overwritten by modules such as C<Coro::EV> and
127C<Coro::AnyEvent> to wait on an external event that hopefully wake up a 140C<Coro::AnyEvent> to wait on an external event that hopefully wake up a
128coroutine so the scheduler can run it. 141coro so the scheduler can run it.
129 142
130Note that the callback I<must not>, under any circumstances, block 143Note that the callback I<must not>, under any circumstances, block
131the current coroutine. Normally, this is achieved by having an "idle 144the current coro. Normally, this is achieved by having an "idle
132coroutine" that calls the event loop and then blocks again, and then 145coro" that calls the event loop and then blocks again, and then
133readying that coroutine in the idle handler. 146readying that coro in the idle handler, or by simply placing the idle
147coro in this variable.
134 148
135See L<Coro::Event> or L<Coro::AnyEvent> for examples of using this 149See L<Coro::Event> or L<Coro::AnyEvent> for examples of using this
136technique. 150technique.
137 151
138Please note that if your callback recursively invokes perl (e.g. for event 152Please note that if your callback recursively invokes perl (e.g. for event
… …
143$idle = sub { 157$idle = sub {
144 require Carp; 158 require Carp;
145 Carp::croak ("FATAL: deadlock detected"); 159 Carp::croak ("FATAL: deadlock detected");
146}; 160};
147 161
148sub _cancel {
149 my ($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->{_on_destroy}) || []};
158}
159
160# this coroutine is necessary because a coroutine 162# this coro is necessary because a coro
161# cannot destroy itself. 163# cannot destroy itself.
162my @destroy; 164our @destroy;
163my $manager; 165our $manager;
164 166
165$manager = new Coro sub { 167$manager = new Coro sub {
166 while () { 168 while () {
167 (shift @destroy)->_cancel 169 Coro::State::cancel shift @destroy
168 while @destroy; 170 while @destroy;
169 171
170 &schedule; 172 &schedule;
171 } 173 }
172}; 174};
173$manager->desc ("[coro manager]"); 175$manager->{desc} = "[coro manager]";
174$manager->prio (PRIO_MAX); 176$manager->prio (PRIO_MAX);
175 177
176=back 178=back
177 179
178=head2 SIMPLE COROUTINE CREATION 180=head1 SIMPLE CORO CREATION
179 181
180=over 4 182=over 4
181 183
182=item async { ... } [@args...] 184=item async { ... } [@args...]
183 185
184Create a new coroutine and return it's coroutine object (usually 186Create a new coro and return its Coro object (usually
185unused). The coroutine will be put into the ready queue, so 187unused). The coro will be put into the ready queue, so
186it will start running automatically on the next scheduler run. 188it will start running automatically on the next scheduler run.
187 189
188The first argument is a codeblock/closure that should be executed in the 190The first argument is a codeblock/closure that should be executed in the
189coroutine. When it returns argument returns the coroutine is automatically 191coro. When it returns argument returns the coro is automatically
190terminated. 192terminated.
191 193
192The remaining arguments are passed as arguments to the closure. 194The remaining arguments are passed as arguments to the closure.
193 195
194See the C<Coro::State::new> constructor for info about the coroutine 196See the C<Coro::State::new> constructor for info about the coro
195environment in which coroutines are executed. 197environment in which coro are executed.
196 198
197Calling C<exit> in a coroutine will do the same as calling exit outside 199Calling C<exit> in a coro will do the same as calling exit outside
198the coroutine. Likewise, when the coroutine dies, the program will exit, 200the coro. Likewise, when the coro dies, the program will exit,
199just as it would in the main program. 201just as it would in the main program.
200 202
201If you do not want that, you can provide a default C<die> handler, or 203If you do not want that, you can provide a default C<die> handler, or
202simply avoid dieing (by use of C<eval>). 204simply avoid dieing (by use of C<eval>).
203 205
204Example: Create a new coroutine that just prints its arguments. 206Example: Create a new coro that just prints its arguments.
205 207
206 async { 208 async {
207 print "@_\n"; 209 print "@_\n";
208 } 1,2,3,4; 210 } 1,2,3,4;
209 211
… …
215 $coro 217 $coro
216} 218}
217 219
218=item async_pool { ... } [@args...] 220=item async_pool { ... } [@args...]
219 221
220Similar to C<async>, but uses a coroutine pool, so you should not call 222Similar to C<async>, but uses a coro pool, so you should not call
221terminate or join on it (although you are allowed to), and you get a 223terminate or join on it (although you are allowed to), and you get a
222coroutine that might have executed other code already (which can be good 224coro that might have executed other code already (which can be good
223or bad :). 225or bad :).
224 226
225On the plus side, this function is faster than creating (and destroying) 227On the plus side, this function is about twice as fast as creating (and
226a completely new coroutine, so if you need a lot of generic coroutines in 228destroying) a completely new coro, so if you need a lot of generic
227quick successsion, use C<async_pool>, not C<async>. 229coros in quick successsion, use C<async_pool>, not C<async>.
228 230
229The code block is executed in an C<eval> context and a warning will be 231The code block is executed in an C<eval> context and a warning will be
230issued in case of an exception instead of terminating the program, as 232issued in case of an exception instead of terminating the program, as
231C<async> does. As the coroutine is being reused, stuff like C<on_destroy> 233C<async> does. As the coro is being reused, stuff like C<on_destroy>
232will not work in the expected way, unless you call terminate or cancel, 234will not work in the expected way, unless you call terminate or cancel,
233which somehow defeats the purpose of pooling (but is fine in the 235which somehow defeats the purpose of pooling (but is fine in the
234exceptional case). 236exceptional case).
235 237
236The priority will be reset to C<0> after each run, tracing will be 238The priority will be reset to C<0> after each run, tracing will be
237disabled, the description will be reset and the default output filehandle 239disabled, the description will be reset and the default output filehandle
238gets restored, so you can change all these. Otherwise the coroutine will 240gets restored, so you can change all these. Otherwise the coro will
239be re-used "as-is": most notably if you change other per-coroutine global 241be re-used "as-is": most notably if you change other per-coro global
240stuff such as C<$/> you I<must needs> to revert that change, which is most 242stuff such as C<$/> you I<must needs> revert that change, which is most
241simply done by using local as in: C< local $/ >. 243simply done by using local as in: C<< local $/ >>.
242 244
243The pool size is limited to C<8> idle coroutines (this can be adjusted by 245The idle pool size is limited to C<8> idle coros (this can be
244changing $Coro::POOL_SIZE), and there can be as many non-idle coros as 246adjusted by changing $Coro::POOL_SIZE), but there can be as many non-idle
245required. 247coros as required.
246 248
247If you are concerned about pooled coroutines growing a lot because a 249If you are concerned about pooled coros growing a lot because a
248single C<async_pool> used a lot of stackspace you can e.g. C<async_pool 250single C<async_pool> used a lot of stackspace you can e.g. C<async_pool
249{ terminate }> once per second or so to slowly replenish the pool. In 251{ terminate }> once per second or so to slowly replenish the pool. In
250addition to that, when the stacks used by a handler grows larger than 16kb 252addition to that, when the stacks used by a handler grows larger than 32kb
251(adjustable via $Coro::POOL_RSS) it will also be destroyed. 253(adjustable via $Coro::POOL_RSS) it will also be destroyed.
252 254
253=cut 255=cut
254 256
255our $POOL_SIZE = 8; 257our $POOL_SIZE = 8;
256our $POOL_RSS = 16 * 1024; 258our $POOL_RSS = 32 * 1024;
257our @async_pool; 259our @async_pool;
258 260
259sub pool_handler { 261sub pool_handler {
260 my $cb;
261
262 while () { 262 while () {
263 eval { 263 eval {
264 while () { 264 &{&_pool_handler} while 1;
265 _pool_1 $cb;
266 &$cb;
267 _pool_2 $cb;
268 &schedule;
269 }
270 }; 265 };
271 266
272 if ($@) {
273 last if $@ eq "\3async_pool terminate\2\n";
274 warn $@; 267 warn $@ if $@;
275 }
276 } 268 }
277} 269}
278 270
279sub async_pool(&@) {
280 # this is also inlined into the unlock_scheduler
281 my $coro = (pop @async_pool) || new Coro \&pool_handler;
282
283 $coro->{_invoke} = [@_];
284 $coro->ready;
285
286 $coro
287}
288
289=back 271=back
290 272
291=head2 STATIC METHODS 273=head1 STATIC METHODS
292 274
293Static methods are actually functions that operate on the current coroutine. 275Static methods are actually functions that implicitly operate on the
276current coro.
294 277
295=over 4 278=over 4
296 279
297=item schedule 280=item schedule
298 281
299Calls the scheduler. The scheduler will find the next coroutine that is 282Calls the scheduler. The scheduler will find the next coro that is
300to be run from the ready queue and switches to it. The next coroutine 283to be run from the ready queue and switches to it. The next coro
301to be run is simply the one with the highest priority that is longest 284to be run is simply the one with the highest priority that is longest
302in its ready queue. If there is no coroutine ready, it will clal the 285in its ready queue. If there is no coro ready, it will clal the
303C<$Coro::idle> hook. 286C<$Coro::idle> hook.
304 287
305Please note that the current coroutine will I<not> be put into the ready 288Please note that the current coro will I<not> be put into the ready
306queue, so calling this function usually means you will never be called 289queue, so calling this function usually means you will never be called
307again unless something else (e.g. an event handler) calls C<< ->ready >>, 290again unless something else (e.g. an event handler) calls C<< ->ready >>,
308thus waking you up. 291thus waking you up.
309 292
310This makes C<schedule> I<the> generic method to use to block the current 293This makes C<schedule> I<the> generic method to use to block the current
311coroutine and wait for events: first you remember the current coroutine in 294coro and wait for events: first you remember the current coro in
312a variable, then arrange for some callback of yours to call C<< ->ready 295a variable, then arrange for some callback of yours to call C<< ->ready
313>> on that once some event happens, and last you call C<schedule> to put 296>> on that once some event happens, and last you call C<schedule> to put
314yourself to sleep. Note that a lot of things can wake your coroutine up, 297yourself to sleep. Note that a lot of things can wake your coro up,
315so you need to check whether the event indeed happened, e.g. by storing the 298so you need to check whether the event indeed happened, e.g. by storing the
316status in a variable. 299status in a variable.
317 300
318The canonical way to wait on external events is this: 301See B<HOW TO WAIT FOR A CALLBACK>, below, for some ways to wait for callbacks.
319 302
320 { 303=item cede
321 # remember current coroutine
322 my $current = $Coro::current;
323 304
324 # register a hypothetical event handler 305"Cede" to other coros. This function puts the current coro into
325 on_event_invoke sub { 306the ready queue and calls C<schedule>, which has the effect of giving
326 # wake up sleeping coroutine 307up the current "timeslice" to other coros of the same or higher
327 $current->ready; 308priority. Once your coro gets its turn again it will automatically be
328 undef $current; 309resumed.
310
311This function is often called C<yield> in other languages.
312
313=item Coro::cede_notself
314
315Works like cede, but is not exported by default and will cede to I<any>
316coro, regardless of priority. This is useful sometimes to ensure
317progress is made.
318
319=item terminate [arg...]
320
321Terminates the current coro with the given status values (see L<cancel>).
322
323=item Coro::on_enter BLOCK, Coro::on_leave BLOCK
324
325These function install enter and leave winders in the current scope. The
326enter block will be executed when on_enter is called and whenever the
327current coro is re-entered by the scheduler, while the leave block is
328executed whenever the current coro is blocked by the scheduler, and
329also when the containing scope is exited (by whatever means, be it exit,
330die, last etc.).
331
332I<Neither invoking the scheduler, nor exceptions, are allowed within those
333BLOCKs>. That means: do not even think about calling C<die> without an
334eval, and do not even think of entering the scheduler in any way.
335
336Since both BLOCKs are tied to the current scope, they will automatically
337be removed when the current scope exits.
338
339These functions implement the same concept as C<dynamic-wind> in scheme
340does, and are useful when you want to localise some resource to a specific
341coro.
342
343They slow down thread switching considerably for coros that use them
344(about 40% for a BLOCK with a single assignment, so thread switching is
345still reasonably fast if the handlers are fast).
346
347These functions are best understood by an example: The following function
348will change the current timezone to "Antarctica/South_Pole", which
349requires a call to C<tzset>, but by using C<on_enter> and C<on_leave>,
350which remember/change the current timezone and restore the previous
351value, respectively, the timezone is only changed for the coro that
352installed those handlers.
353
354 use POSIX qw(tzset);
355
356 async {
357 my $old_tz; # store outside TZ value here
358
359 Coro::on_enter {
360 $old_tz = $ENV{TZ}; # remember the old value
361
362 $ENV{TZ} = "Antarctica/South_Pole";
363 tzset; # enable new value
329 }; 364 };
330 365
331 # call schedule until event occurred. 366 Coro::on_leave {
332 # in case we are woken up for other reasons 367 $ENV{TZ} = $old_tz;
333 # (current still defined), loop. 368 tzset; # restore old value
334 Coro::schedule while $current; 369 };
370
371 # at this place, the timezone is Antarctica/South_Pole,
372 # without disturbing the TZ of any other coro.
335 } 373 };
336 374
337=item cede 375This can be used to localise about any resource (locale, uid, current
376working directory etc.) to a block, despite the existance of other
377coros.
338 378
339"Cede" to other coroutines. This function puts the current coroutine into 379Another interesting example implements time-sliced multitasking using
340the ready queue and calls C<schedule>, which has the effect of giving 380interval timers (this could obviously be optimised, but does the job):
341up the current "timeslice" to other coroutines of the same or higher
342priority. Once your coroutine gets its turn again it will automatically be
343resumed.
344 381
345This function is often called C<yield> in other languages. 382 # "timeslice" the given block
383 sub timeslice(&) {
384 use Time::HiRes ();
346 385
347=item Coro::cede_notself 386 Coro::on_enter {
387 # on entering the thread, we set an VTALRM handler to cede
388 $SIG{VTALRM} = sub { cede };
389 # and then start the interval timer
390 Time::HiRes::setitimer &Time::HiRes::ITIMER_VIRTUAL, 0.01, 0.01;
391 };
392 Coro::on_leave {
393 # on leaving the thread, we stop the interval timer again
394 Time::HiRes::setitimer &Time::HiRes::ITIMER_VIRTUAL, 0, 0;
395 };
348 396
349Works like cede, but is not exported by default and will cede to I<any> 397 &{+shift};
350coroutine, regardless of priority. This is useful sometimes to ensure 398 }
351progress is made.
352 399
353=item terminate [arg...] 400 # use like this:
401 timeslice {
402 # The following is an endless loop that would normally
403 # monopolise the process. Since it runs in a timesliced
404 # environment, it will regularly cede to other threads.
405 while () { }
406 };
354 407
355Terminates the current coroutine with the given status values (see L<cancel>).
356 408
357=item killall 409=item killall
358 410
359Kills/terminates/cancels all coroutines except the currently running 411Kills/terminates/cancels all coros except the currently running one.
360one. This is useful after a fork, either in the child or the parent, as
361usually only one of them should inherit the running coroutines.
362 412
363Note that while this will try to free some of the main programs resources, 413Note that while this will try to free some of the main interpreter
414resources if the calling coro isn't the main coro, but one
364you cannot free all of them, so if a coroutine that is not the main 415cannot free all of them, so if a coro that is not the main coro
365program calls this function, there will be some one-time resource leak. 416calls this function, there will be some one-time resource leak.
366 417
367=cut 418=cut
368
369sub terminate {
370 $current->cancel (@_);
371}
372 419
373sub killall { 420sub killall {
374 for (Coro::State::list) { 421 for (Coro::State::list) {
375 $_->cancel 422 $_->cancel
376 if $_ != $current && UNIVERSAL::isa $_, "Coro"; 423 if $_ != $current && UNIVERSAL::isa $_, "Coro";
377 } 424 }
378} 425}
379 426
380=back 427=back
381 428
382=head2 COROUTINE METHODS 429=head1 CORO OBJECT METHODS
383 430
384These are the methods you can call on coroutine objects (or to create 431These are the methods you can call on coro objects (or to create
385them). 432them).
386 433
387=over 4 434=over 4
388 435
389=item new Coro \&sub [, @args...] 436=item new Coro \&sub [, @args...]
390 437
391Create a new coroutine and return it. When the sub returns, the coroutine 438Create a new coro and return it. When the sub returns, the coro
392automatically terminates as if C<terminate> with the returned values were 439automatically terminates as if C<terminate> with the returned values were
393called. To make the coroutine run you must first put it into the ready 440called. To make the coro run you must first put it into the ready
394queue by calling the ready method. 441queue by calling the ready method.
395 442
396See C<async> and C<Coro::State::new> for additional info about the 443See C<async> and C<Coro::State::new> for additional info about the
397coroutine environment. 444coro environment.
398 445
399=cut 446=cut
400 447
401sub _run_coro { 448sub _coro_run {
402 terminate &{+shift}; 449 terminate &{+shift};
403} 450}
404 451
405sub new {
406 my $class = shift;
407
408 $class->SUPER::new (\&_run_coro, @_)
409}
410
411=item $success = $coroutine->ready 452=item $success = $coro->ready
412 453
413Put the given coroutine into the end of its ready queue (there is one 454Put the given coro into the end of its ready queue (there is one
414queue for each priority) and return true. If the coroutine is already in 455queue for each priority) and return true. If the coro is already in
415the ready queue, do nothing and return false. 456the ready queue, do nothing and return false.
416 457
417This ensures that the scheduler will resume this coroutine automatically 458This ensures that the scheduler will resume this coro automatically
418once all the coroutines of higher priority and all coroutines of the same 459once all the coro of higher priority and all coro of the same
419priority that were put into the ready queue earlier have been resumed. 460priority that were put into the ready queue earlier have been resumed.
420 461
462=item $coro->suspend
463
464Suspends the specified coro. A suspended coro works just like any other
465coro, except that the scheduler will not select a suspended coro for
466execution.
467
468Suspending a coro can be useful when you want to keep the coro from
469running, but you don't want to destroy it, or when you want to temporarily
470freeze a coro (e.g. for debugging) to resume it later.
471
472A scenario for the former would be to suspend all (other) coros after a
473fork and keep them alive, so their destructors aren't called, but new
474coros can be created.
475
476=item $coro->resume
477
478If the specified coro was suspended, it will be resumed. Note that when
479the coro was in the ready queue when it was suspended, it might have been
480unreadied by the scheduler, so an activation might have been lost.
481
482To avoid this, it is best to put a suspended coro into the ready queue
483unconditionally, as every synchronisation mechanism must protect itself
484against spurious wakeups, and the one in the Coro family certainly do
485that.
486
421=item $is_ready = $coroutine->is_ready 487=item $is_ready = $coro->is_ready
422 488
423Return whether the coroutine is currently the ready queue or not, 489Returns true iff the Coro object is in the ready queue. Unless the Coro
490object gets destroyed, it will eventually be scheduled by the scheduler.
424 491
492=item $is_running = $coro->is_running
493
494Returns true iff the Coro object is currently running. Only one Coro object
495can ever be in the running state (but it currently is possible to have
496multiple running Coro::States).
497
498=item $is_suspended = $coro->is_suspended
499
500Returns true iff this Coro object has been suspended. Suspended Coros will
501not ever be scheduled.
502
425=item $coroutine->cancel (arg...) 503=item $coro->cancel (arg...)
426 504
427Terminates the given coroutine and makes it return the given arguments as 505Terminates the given Coro and makes it return the given arguments as
428status (default: the empty list). Never returns if the coroutine is the 506status (default: the empty list). Never returns if the Coro is the
429current coroutine. 507current Coro.
430 508
431=cut 509=cut
432 510
433sub cancel { 511sub cancel {
434 my $self = shift; 512 my $self = shift;
435 $self->{_status} = [@_];
436 513
437 if ($current == $self) { 514 if ($current == $self) {
438 push @destroy, $self; 515 terminate @_;
439 $manager->ready;
440 &schedule while 1;
441 } else { 516 } else {
442 $self->_cancel; 517 $self->{_status} = [@_];
518 Coro::State::cancel $self;
443 } 519 }
444} 520}
445 521
522=item $coro->schedule_to
523
524Puts the current coro to sleep (like C<Coro::schedule>), but instead
525of continuing with the next coro from the ready queue, always switch to
526the given coro object (regardless of priority etc.). The readyness
527state of that coro isn't changed.
528
529This is an advanced method for special cases - I'd love to hear about any
530uses for this one.
531
532=item $coro->cede_to
533
534Like C<schedule_to>, but puts the current coro into the ready
535queue. This has the effect of temporarily switching to the given
536coro, and continuing some time later.
537
538This is an advanced method for special cases - I'd love to hear about any
539uses for this one.
540
541=item $coro->throw ([$scalar])
542
543If C<$throw> is specified and defined, it will be thrown as an exception
544inside the coro at the next convenient point in time. Otherwise
545clears the exception object.
546
547Coro will check for the exception each time a schedule-like-function
548returns, i.e. after each C<schedule>, C<cede>, C<< Coro::Semaphore->down
549>>, C<< Coro::Handle->readable >> and so on. Most of these functions
550detect this case and return early in case an exception is pending.
551
552The exception object will be thrown "as is" with the specified scalar in
553C<$@>, i.e. if it is a string, no line number or newline will be appended
554(unlike with C<die>).
555
556This can be used as a softer means than C<cancel> to ask a coro to
557end itself, although there is no guarantee that the exception will lead to
558termination, and if the exception isn't caught it might well end the whole
559program.
560
561You might also think of C<throw> as being the moral equivalent of
562C<kill>ing a coro with a signal (in this case, a scalar).
563
446=item $coroutine->join 564=item $coro->join
447 565
448Wait until the coroutine terminates and return any values given to the 566Wait until the coro terminates and return any values given to the
449C<terminate> or C<cancel> functions. C<join> can be called concurrently 567C<terminate> or C<cancel> functions. C<join> can be called concurrently
450from multiple coroutines, and all will be resumed and given the status 568from multiple coro, and all will be resumed and given the status
451return once the C<$coroutine> terminates. 569return once the C<$coro> terminates.
452 570
453=cut 571=cut
454 572
455sub join { 573sub join {
456 my $self = shift; 574 my $self = shift;
… …
467 } 585 }
468 586
469 wantarray ? @{$self->{_status}} : $self->{_status}[0]; 587 wantarray ? @{$self->{_status}} : $self->{_status}[0];
470} 588}
471 589
472=item $coroutine->on_destroy (\&cb) 590=item $coro->on_destroy (\&cb)
473 591
474Registers a callback that is called when this coroutine gets destroyed, 592Registers a callback that is called when this coro gets destroyed,
475but before it is joined. The callback gets passed the terminate arguments, 593but before it is joined. The callback gets passed the terminate arguments,
476if any, and I<must not> die, under any circumstances. 594if any, and I<must not> die, under any circumstances.
477 595
478=cut 596=cut
479 597
… …
481 my ($self, $cb) = @_; 599 my ($self, $cb) = @_;
482 600
483 push @{ $self->{_on_destroy} }, $cb; 601 push @{ $self->{_on_destroy} }, $cb;
484} 602}
485 603
486=item $oldprio = $coroutine->prio ($newprio) 604=item $oldprio = $coro->prio ($newprio)
487 605
488Sets (or gets, if the argument is missing) the priority of the 606Sets (or gets, if the argument is missing) the priority of the
489coroutine. Higher priority coroutines get run before lower priority 607coro. Higher priority coro get run before lower priority
490coroutines. Priorities are small signed integers (currently -4 .. +3), 608coro. Priorities are small signed integers (currently -4 .. +3),
491that you can refer to using PRIO_xxx constants (use the import tag :prio 609that you can refer to using PRIO_xxx constants (use the import tag :prio
492to get then): 610to get then):
493 611
494 PRIO_MAX > PRIO_HIGH > PRIO_NORMAL > PRIO_LOW > PRIO_IDLE > PRIO_MIN 612 PRIO_MAX > PRIO_HIGH > PRIO_NORMAL > PRIO_LOW > PRIO_IDLE > PRIO_MIN
495 3 > 1 > 0 > -1 > -3 > -4 613 3 > 1 > 0 > -1 > -3 > -4
496 614
497 # set priority to HIGH 615 # set priority to HIGH
498 current->prio(PRIO_HIGH); 616 current->prio (PRIO_HIGH);
499 617
500The idle coroutine ($Coro::idle) always has a lower priority than any 618The idle coro ($Coro::idle) always has a lower priority than any
501existing coroutine. 619existing coro.
502 620
503Changing the priority of the current coroutine will take effect immediately, 621Changing the priority of the current coro will take effect immediately,
504but changing the priority of coroutines in the ready queue (but not 622but changing the priority of coro in the ready queue (but not
505running) will only take effect after the next schedule (of that 623running) will only take effect after the next schedule (of that
506coroutine). This is a bug that will be fixed in some future version. 624coro). This is a bug that will be fixed in some future version.
507 625
508=item $newprio = $coroutine->nice ($change) 626=item $newprio = $coro->nice ($change)
509 627
510Similar to C<prio>, but subtract the given value from the priority (i.e. 628Similar to C<prio>, but subtract the given value from the priority (i.e.
511higher values mean lower priority, just as in unix). 629higher values mean lower priority, just as in unix).
512 630
513=item $olddesc = $coroutine->desc ($newdesc) 631=item $olddesc = $coro->desc ($newdesc)
514 632
515Sets (or gets in case the argument is missing) the description for this 633Sets (or gets in case the argument is missing) the description for this
516coroutine. This is just a free-form string you can associate with a coroutine. 634coro. This is just a free-form string you can associate with a
635coro.
517 636
518This method simply sets the C<< $coroutine->{desc} >> member to the given string. You 637This method simply sets the C<< $coro->{desc} >> member to the given
519can modify this member directly if you wish. 638string. You can modify this member directly if you wish.
520
521=item $coroutine->throw ([$scalar])
522
523If C<$throw> is specified and defined, it will be thrown as an exception
524inside the coroutine at the next convinient point in time (usually after
525it gains control at the next schedule/transfer/cede). Otherwise clears the
526exception object.
527
528The exception object will be thrown "as is" with the specified scalar in
529C<$@>, i.e. if it is a string, no line number or newline will be appended
530(unlike with C<die>).
531
532This can be used as a softer means than C<cancel> to ask a coroutine to
533end itself, although there is no guarentee that the exception will lead to
534termination, and if the exception isn't caught it might well end the whole
535program.
536 639
537=cut 640=cut
538 641
539sub desc { 642sub desc {
540 my $old = $_[0]{desc}; 643 my $old = $_[0]{desc};
541 $_[0]{desc} = $_[1] if @_ > 1; 644 $_[0]{desc} = $_[1] if @_ > 1;
542 $old; 645 $old;
543} 646}
544 647
648sub transfer {
649 require Carp;
650 Carp::croak ("You must not call ->transfer on Coro objects. Use Coro::State objects or the ->schedule_to method. Caught");
651}
652
545=back 653=back
546 654
547=head2 GLOBAL FUNCTIONS 655=head1 GLOBAL FUNCTIONS
548 656
549=over 4 657=over 4
550 658
551=item Coro::nready 659=item Coro::nready
552 660
553Returns the number of coroutines that are currently in the ready state, 661Returns the number of coro that are currently in the ready state,
554i.e. that can be switched to by calling C<schedule> directory or 662i.e. that can be switched to by calling C<schedule> directory or
555indirectly. The value C<0> means that the only runnable coroutine is the 663indirectly. The value C<0> means that the only runnable coro is the
556currently running one, so C<cede> would have no effect, and C<schedule> 664currently running one, so C<cede> would have no effect, and C<schedule>
557would cause a deadlock unless there is an idle handler that wakes up some 665would cause a deadlock unless there is an idle handler that wakes up some
558coroutines. 666coro.
559 667
560=item my $guard = Coro::guard { ... } 668=item my $guard = Coro::guard { ... }
561 669
562This creates and returns a guard object. Nothing happens until the object 670This function still exists, but is deprecated. Please use the
563gets destroyed, in which case the codeblock given as argument will be 671C<Guard::guard> function instead.
564executed. This is useful to free locks or other resources in case of a
565runtime error or when the coroutine gets canceled, as in both cases the
566guard block will be executed. The guard object supports only one method,
567C<< ->cancel >>, which will keep the codeblock from being executed.
568 672
569Example: set some flag and clear it again when the coroutine gets canceled
570or the function returns:
571
572 sub do_something {
573 my $guard = Coro::guard { $busy = 0 };
574 $busy = 1;
575
576 # do something that requires $busy to be true
577 }
578
579=cut 673=cut
580 674
581sub guard(&) { 675BEGIN { *guard = \&Guard::guard }
582 bless \(my $cb = $_[0]), "Coro::guard"
583}
584
585sub Coro::guard::cancel {
586 ${$_[0]} = sub { };
587}
588
589sub Coro::guard::DESTROY {
590 ${$_[0]}->();
591}
592
593 676
594=item unblock_sub { ... } 677=item unblock_sub { ... }
595 678
596This utility function takes a BLOCK or code reference and "unblocks" it, 679This utility function takes a BLOCK or code reference and "unblocks" it,
597returning a new coderef. Unblocking means that calling the new coderef 680returning a new coderef. Unblocking means that calling the new coderef
598will return immediately without blocking, returning nothing, while the 681will return immediately without blocking, returning nothing, while the
599original code ref will be called (with parameters) from within another 682original code ref will be called (with parameters) from within another
600coroutine. 683coro.
601 684
602The reason this function exists is that many event libraries (such as the 685The reason this function exists is that many event libraries (such as the
603venerable L<Event|Event> module) are not coroutine-safe (a weaker form 686venerable L<Event|Event> module) are not thread-safe (a weaker form
604of thread-safety). This means you must not block within event callbacks, 687of reentrancy). This means you must not block within event callbacks,
605otherwise you might suffer from crashes or worse. The only event library 688otherwise you might suffer from crashes or worse. The only event library
606currently known that is safe to use without C<unblock_sub> is L<EV>. 689currently known that is safe to use without C<unblock_sub> is L<EV>.
607 690
608This function allows your callbacks to block by executing them in another 691This function allows your callbacks to block by executing them in another
609coroutine where it is safe to block. One example where blocking is handy 692coro where it is safe to block. One example where blocking is handy
610is when you use the L<Coro::AIO|Coro::AIO> functions to save results to 693is when you use the L<Coro::AIO|Coro::AIO> functions to save results to
611disk, for example. 694disk, for example.
612 695
613In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when 696In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when
614creating event callbacks that want to block. 697creating event callbacks that want to block.
615 698
616If your handler does not plan to block (e.g. simply sends a message to 699If your handler does not plan to block (e.g. simply sends a message to
617another coroutine, or puts some other coroutine into the ready queue), 700another coro, or puts some other coro into the ready queue), there is
618there is no reason to use C<unblock_sub>. 701no reason to use C<unblock_sub>.
619 702
620Note that you also need to use C<unblock_sub> for any other callbacks that 703Note that you also need to use C<unblock_sub> for any other callbacks that
621are indirectly executed by any C-based event loop. For example, when you 704are indirectly executed by any C-based event loop. For example, when you
622use a module that uses L<AnyEvent> (and you use L<Coro::AnyEvent>) and it 705use a module that uses L<AnyEvent> (and you use L<Coro::AnyEvent>) and it
623provides callbacks that are the result of some event callback, then you 706provides callbacks that are the result of some event callback, then you
… …
632# return immediately and can be reused) and because we cannot cede 715# return immediately and can be reused) and because we cannot cede
633# inside an event callback. 716# inside an event callback.
634our $unblock_scheduler = new Coro sub { 717our $unblock_scheduler = new Coro sub {
635 while () { 718 while () {
636 while (my $cb = pop @unblock_queue) { 719 while (my $cb = pop @unblock_queue) {
637 # this is an inlined copy of async_pool 720 &async_pool (@$cb);
638 my $coro = (pop @async_pool) || new Coro \&pool_handler;
639 721
640 $coro->{_invoke} = $cb;
641 $coro->ready;
642 cede; # for short-lived callbacks, this reduces pressure on the coro pool 722 # for short-lived callbacks, this reduces pressure on the coro pool
723 # as the chance is very high that the async_poll coro will be back
724 # in the idle state when cede returns
725 cede;
643 } 726 }
644 schedule; # sleep well 727 schedule; # sleep well
645 } 728 }
646}; 729};
647$unblock_scheduler->desc ("[unblock_sub scheduler]"); 730$unblock_scheduler->{desc} = "[unblock_sub scheduler]";
648 731
649sub unblock_sub(&) { 732sub unblock_sub(&) {
650 my $cb = shift; 733 my $cb = shift;
651 734
652 sub { 735 sub {
653 unshift @unblock_queue, [$cb, @_]; 736 unshift @unblock_queue, [$cb, @_];
654 $unblock_scheduler->ready; 737 $unblock_scheduler->ready;
655 } 738 }
656} 739}
657 740
741=item $cb = Coro::rouse_cb
742
743Create and return a "rouse callback". That's a code reference that,
744when called, will remember a copy of its arguments and notify the owner
745coro of the callback.
746
747See the next function.
748
749=item @args = Coro::rouse_wait [$cb]
750
751Wait for the specified rouse callback (or the last one that was created in
752this coro).
753
754As soon as the callback is invoked (or when the callback was invoked
755before C<rouse_wait>), it will return the arguments originally passed to
756the rouse callback. In scalar context, that means you get the I<last>
757argument, just as if C<rouse_wait> had a C<return ($a1, $a2, $a3...)>
758statement at the end.
759
760See the section B<HOW TO WAIT FOR A CALLBACK> for an actual usage example.
761
658=back 762=back
659 763
660=cut 764=cut
661 765
6621; 7661;
663 767
768=head1 HOW TO WAIT FOR A CALLBACK
769
770It is very common for a coro to wait for some callback to be
771called. This occurs naturally when you use coro in an otherwise
772event-based program, or when you use event-based libraries.
773
774These typically register a callback for some event, and call that callback
775when the event occured. In a coro, however, you typically want to
776just wait for the event, simplyifying things.
777
778For example C<< AnyEvent->child >> registers a callback to be called when
779a specific child has exited:
780
781 my $child_watcher = AnyEvent->child (pid => $pid, cb => sub { ... });
782
783But from within a coro, you often just want to write this:
784
785 my $status = wait_for_child $pid;
786
787Coro offers two functions specifically designed to make this easy,
788C<Coro::rouse_cb> and C<Coro::rouse_wait>.
789
790The first function, C<rouse_cb>, generates and returns a callback that,
791when invoked, will save its arguments and notify the coro that
792created the callback.
793
794The second function, C<rouse_wait>, waits for the callback to be called
795(by calling C<schedule> to go to sleep) and returns the arguments
796originally passed to the callback.
797
798Using these functions, it becomes easy to write the C<wait_for_child>
799function mentioned above:
800
801 sub wait_for_child($) {
802 my ($pid) = @_;
803
804 my $watcher = AnyEvent->child (pid => $pid, cb => Coro::rouse_cb);
805
806 my ($rpid, $rstatus) = Coro::rouse_wait;
807 $rstatus
808 }
809
810In the case where C<rouse_cb> and C<rouse_wait> are not flexible enough,
811you can roll your own, using C<schedule>:
812
813 sub wait_for_child($) {
814 my ($pid) = @_;
815
816 # store the current coro in $current,
817 # and provide result variables for the closure passed to ->child
818 my $current = $Coro::current;
819 my ($done, $rstatus);
820
821 # pass a closure to ->child
822 my $watcher = AnyEvent->child (pid => $pid, cb => sub {
823 $rstatus = $_[1]; # remember rstatus
824 $done = 1; # mark $rstatus as valud
825 });
826
827 # wait until the closure has been called
828 schedule while !$done;
829
830 $rstatus
831 }
832
833
664=head1 BUGS/LIMITATIONS 834=head1 BUGS/LIMITATIONS
665 835
836=over 4
837
838=item fork with pthread backend
839
840When Coro is compiled using the pthread backend (which isn't recommended
841but required on many BSDs as their libcs are completely broken), then
842coro will not survive a fork. There is no known workaround except to
843fix your libc and use a saner backend.
844
845=item perl process emulation ("threads")
846
666This module is not perl-pseudo-thread-safe. You should only ever use this 847This module is not perl-pseudo-thread-safe. You should only ever use this
667module from the same thread (this requirement might be removed in the 848module from the first thread (this requirement might be removed in the
668future to allow per-thread schedulers, but Coro::State does not yet allow 849future to allow per-thread schedulers, but Coro::State does not yet allow
669this). I recommend disabling thread support and using processes, as this 850this). I recommend disabling thread support and using processes, as having
670is much faster and uses less memory. 851the windows process emulation enabled under unix roughly halves perl
852performance, even when not used.
853
854=item coro switching is not signal safe
855
856You must not switch to another coro from within a signal handler
857(only relevant with %SIG - most event libraries provide safe signals).
858
859That means you I<MUST NOT> call any function that might "block" the
860current coro - C<cede>, C<schedule> C<< Coro::Semaphore->down >> or
861anything that calls those. Everything else, including calling C<ready>,
862works.
863
864=back
865
671 866
672=head1 SEE ALSO 867=head1 SEE ALSO
673 868
674Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>. 869Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>.
675 870
676Debugging: L<Coro::Debug>. 871Debugging: L<Coro::Debug>.
677 872
678Support/Utility: L<Coro::Specific>, L<Coro::Util>. 873Support/Utility: L<Coro::Specific>, L<Coro::Util>.
679 874
680Locking/IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>, L<Coro::SemaphoreSet>, L<Coro::RWLock>. 875Locking and IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>,
876L<Coro::SemaphoreSet>, L<Coro::RWLock>.
681 877
682IO/Timers: L<Coro::Timer>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::AIO>. 878I/O and Timers: L<Coro::Timer>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::AIO>.
683 879
684Compatibility: L<Coro::LWP>, L<Coro::BDB>, L<Coro::Storable>, L<Coro::Select>. 880Compatibility with other modules: L<Coro::LWP> (but see also L<AnyEvent::HTTP> for
881a better-working alternative), L<Coro::BDB>, L<Coro::Storable>,
882L<Coro::Select>.
685 883
686XS API: L<Coro::MakeMaker>. 884XS API: L<Coro::MakeMaker>.
687 885
688Low level Configuration, Coroutine Environment: L<Coro::State>. 886Low level Configuration, Thread Environment, Continuations: L<Coro::State>.
689 887
690=head1 AUTHOR 888=head1 AUTHOR
691 889
692 Marc Lehmann <schmorp@schmorp.de> 890 Marc Lehmann <schmorp@schmorp.de>
693 http://home.schmorp.de/ 891 http://home.schmorp.de/

Diff Legend

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