1 |
=head1 NAME |
2 |
|
3 |
Coro - the only real threads in perl |
4 |
|
5 |
=head1 SYNOPSIS |
6 |
|
7 |
use Coro; |
8 |
|
9 |
async { |
10 |
# some asynchronous thread of execution |
11 |
print "2\n"; |
12 |
cede; # yield back to main |
13 |
print "4\n"; |
14 |
}; |
15 |
print "1\n"; |
16 |
cede; # yield to coro |
17 |
print "3\n"; |
18 |
cede; # and again |
19 |
|
20 |
# use locking |
21 |
use Coro::Semaphore; |
22 |
my $lock = new Coro::Semaphore; |
23 |
my $locked; |
24 |
|
25 |
$lock->down; |
26 |
$locked = 1; |
27 |
$lock->up; |
28 |
|
29 |
=head1 DESCRIPTION |
30 |
|
31 |
For a tutorial-style introduction, please read the L<Coro::Intro> |
32 |
manpage. This manpage mainly contains reference information. |
33 |
|
34 |
This module collection manages continuations in general, most often in |
35 |
the form of cooperative threads (also called coros, or simply "coro" |
36 |
in the documentation). They are similar to kernel threads but don't (in |
37 |
general) run in parallel at the same time even on SMP machines. The |
38 |
specific flavor of thread offered by this module also guarantees you that |
39 |
it will not switch between threads unless necessary, at easily-identified |
40 |
points in your program, so locking and parallel access are rarely an |
41 |
issue, making thread programming much safer and easier than using other |
42 |
thread models. |
43 |
|
44 |
Unlike the so-called "Perl threads" (which are not actually real threads |
45 |
but only the windows process emulation ported to unix, and as such act |
46 |
as processes), Coro provides a full shared address space, which makes |
47 |
communication between threads very easy. And Coro's threads are fast, |
48 |
too: disabling the Windows process emulation code in your perl and using |
49 |
Coro can easily result in a two to four times speed increase for your |
50 |
programs. A parallel matrix multiplication benchmark runs over 300 times |
51 |
faster on a single core than perl's pseudo-threads on a quad core using |
52 |
all four cores. |
53 |
|
54 |
Coro achieves that by supporting multiple running interpreters that share |
55 |
data, which is especially useful to code pseudo-parallel processes and |
56 |
for event-based programming, such as multiple HTTP-GET requests running |
57 |
concurrently. See L<Coro::AnyEvent> to learn more on how to integrate Coro |
58 |
into an event-based environment. |
59 |
|
60 |
In this module, a thread is defined as "callchain + lexical variables + |
61 |
some package variables + C stack), that is, a thread has its own callchain, |
62 |
its own set of lexicals and its own set of perls most important global |
63 |
variables (see L<Coro::State> for more configuration and background info). |
64 |
|
65 |
See also the C<SEE ALSO> section at the end of this document - the Coro |
66 |
module family is quite large. |
67 |
|
68 |
=cut |
69 |
|
70 |
package Coro; |
71 |
|
72 |
use strict qw(vars subs); |
73 |
no warnings "uninitialized"; |
74 |
|
75 |
use Guard (); |
76 |
|
77 |
use Coro::State; |
78 |
|
79 |
use base qw(Coro::State Exporter); |
80 |
|
81 |
our $idle; # idle handler |
82 |
our $main; # main coro |
83 |
our $current; # current coro |
84 |
|
85 |
our $VERSION = 5.13; |
86 |
|
87 |
our @EXPORT = qw(async async_pool cede schedule terminate current unblock_sub); |
88 |
our %EXPORT_TAGS = ( |
89 |
prio => [qw(PRIO_MAX PRIO_HIGH PRIO_NORMAL PRIO_LOW PRIO_IDLE PRIO_MIN)], |
90 |
); |
91 |
our @EXPORT_OK = (@{$EXPORT_TAGS{prio}}, qw(nready)); |
92 |
|
93 |
=head1 GLOBAL VARIABLES |
94 |
|
95 |
=over 4 |
96 |
|
97 |
=item $Coro::main |
98 |
|
99 |
This variable stores the Coro object that represents the main |
100 |
program. While you cna C<ready> it and do most other things you can do to |
101 |
coro, it is mainly useful to compare again C<$Coro::current>, to see |
102 |
whether you are running in the main program or not. |
103 |
|
104 |
=cut |
105 |
|
106 |
# $main is now being initialised by Coro::State |
107 |
|
108 |
=item $Coro::current |
109 |
|
110 |
The Coro object representing the current coro (the last |
111 |
coro that the Coro scheduler switched to). The initial value is |
112 |
C<$Coro::main> (of course). |
113 |
|
114 |
This variable is B<strictly> I<read-only>. You can take copies of the |
115 |
value stored in it and use it as any other Coro object, but you must |
116 |
not otherwise modify the variable itself. |
117 |
|
118 |
=cut |
119 |
|
120 |
sub current() { $current } # [DEPRECATED] |
121 |
|
122 |
=item $Coro::idle |
123 |
|
124 |
This variable is mainly useful to integrate Coro into event loops. It is |
125 |
usually better to rely on L<Coro::AnyEvent> or L<Coro::EV>, as this is |
126 |
pretty low-level functionality. |
127 |
|
128 |
This variable stores either a Coro object or a callback. |
129 |
|
130 |
If it is a callback, the it is called whenever the scheduler finds no |
131 |
ready coros to run. The default implementation prints "FATAL: |
132 |
deadlock detected" and exits, because the program has no other way to |
133 |
continue. |
134 |
|
135 |
If it is a coro object, then this object will be readied (without |
136 |
invoking any ready hooks, however) when the scheduler finds no other ready |
137 |
coros to run. |
138 |
|
139 |
This hook is overwritten by modules such as C<Coro::EV> and |
140 |
C<Coro::AnyEvent> to wait on an external event that hopefully wake up a |
141 |
coro so the scheduler can run it. |
142 |
|
143 |
Note that the callback I<must not>, under any circumstances, block |
144 |
the current coro. Normally, this is achieved by having an "idle |
145 |
coro" that calls the event loop and then blocks again, and then |
146 |
readying that coro in the idle handler, or by simply placing the idle |
147 |
coro in this variable. |
148 |
|
149 |
See L<Coro::Event> or L<Coro::AnyEvent> for examples of using this |
150 |
technique. |
151 |
|
152 |
Please note that if your callback recursively invokes perl (e.g. for event |
153 |
handlers), then it must be prepared to be called recursively itself. |
154 |
|
155 |
=cut |
156 |
|
157 |
$idle = sub { |
158 |
require Carp; |
159 |
Carp::croak ("FATAL: deadlock detected"); |
160 |
}; |
161 |
|
162 |
# this coro is necessary because a coro |
163 |
# cannot destroy itself. |
164 |
our @destroy; |
165 |
our $manager; |
166 |
|
167 |
$manager = new Coro sub { |
168 |
while () { |
169 |
Coro::State::cancel shift @destroy |
170 |
while @destroy; |
171 |
|
172 |
&schedule; |
173 |
} |
174 |
}; |
175 |
$manager->{desc} = "[coro manager]"; |
176 |
$manager->prio (PRIO_MAX); |
177 |
|
178 |
=back |
179 |
|
180 |
=head1 SIMPLE CORO CREATION |
181 |
|
182 |
=over 4 |
183 |
|
184 |
=item async { ... } [@args...] |
185 |
|
186 |
Create a new coro and return its Coro object (usually |
187 |
unused). The coro will be put into the ready queue, so |
188 |
it will start running automatically on the next scheduler run. |
189 |
|
190 |
The first argument is a codeblock/closure that should be executed in the |
191 |
coro. When it returns argument returns the coro is automatically |
192 |
terminated. |
193 |
|
194 |
The remaining arguments are passed as arguments to the closure. |
195 |
|
196 |
See the C<Coro::State::new> constructor for info about the coro |
197 |
environment in which coro are executed. |
198 |
|
199 |
Calling C<exit> in a coro will do the same as calling exit outside |
200 |
the coro. Likewise, when the coro dies, the program will exit, |
201 |
just as it would in the main program. |
202 |
|
203 |
If you do not want that, you can provide a default C<die> handler, or |
204 |
simply avoid dieing (by use of C<eval>). |
205 |
|
206 |
Example: Create a new coro that just prints its arguments. |
207 |
|
208 |
async { |
209 |
print "@_\n"; |
210 |
} 1,2,3,4; |
211 |
|
212 |
=cut |
213 |
|
214 |
sub async(&@) { |
215 |
my $coro = new Coro @_; |
216 |
$coro->ready; |
217 |
$coro |
218 |
} |
219 |
|
220 |
=item async_pool { ... } [@args...] |
221 |
|
222 |
Similar to C<async>, but uses a coro pool, so you should not call |
223 |
terminate or join on it (although you are allowed to), and you get a |
224 |
coro that might have executed other code already (which can be good |
225 |
or bad :). |
226 |
|
227 |
On the plus side, this function is about twice as fast as creating (and |
228 |
destroying) a completely new coro, so if you need a lot of generic |
229 |
coros in quick successsion, use C<async_pool>, not C<async>. |
230 |
|
231 |
The code block is executed in an C<eval> context and a warning will be |
232 |
issued in case of an exception instead of terminating the program, as |
233 |
C<async> does. As the coro is being reused, stuff like C<on_destroy> |
234 |
will not work in the expected way, unless you call terminate or cancel, |
235 |
which somehow defeats the purpose of pooling (but is fine in the |
236 |
exceptional case). |
237 |
|
238 |
The priority will be reset to C<0> after each run, tracing will be |
239 |
disabled, the description will be reset and the default output filehandle |
240 |
gets restored, so you can change all these. Otherwise the coro will |
241 |
be re-used "as-is": most notably if you change other per-coro global |
242 |
stuff such as C<$/> you I<must needs> revert that change, which is most |
243 |
simply done by using local as in: C<< local $/ >>. |
244 |
|
245 |
The idle pool size is limited to C<8> idle coros (this can be |
246 |
adjusted by changing $Coro::POOL_SIZE), but there can be as many non-idle |
247 |
coros as required. |
248 |
|
249 |
If you are concerned about pooled coros growing a lot because a |
250 |
single C<async_pool> used a lot of stackspace you can e.g. C<async_pool |
251 |
{ terminate }> once per second or so to slowly replenish the pool. In |
252 |
addition to that, when the stacks used by a handler grows larger than 32kb |
253 |
(adjustable via $Coro::POOL_RSS) it will also be destroyed. |
254 |
|
255 |
=cut |
256 |
|
257 |
our $POOL_SIZE = 8; |
258 |
our $POOL_RSS = 32 * 1024; |
259 |
our @async_pool; |
260 |
|
261 |
sub pool_handler { |
262 |
while () { |
263 |
eval { |
264 |
&{&_pool_handler} while 1; |
265 |
}; |
266 |
|
267 |
warn $@ if $@; |
268 |
} |
269 |
} |
270 |
|
271 |
=back |
272 |
|
273 |
=head1 STATIC METHODS |
274 |
|
275 |
Static methods are actually functions that implicitly operate on the |
276 |
current coro. |
277 |
|
278 |
=over 4 |
279 |
|
280 |
=item schedule |
281 |
|
282 |
Calls the scheduler. The scheduler will find the next coro that is |
283 |
to be run from the ready queue and switches to it. The next coro |
284 |
to be run is simply the one with the highest priority that is longest |
285 |
in its ready queue. If there is no coro ready, it will clal the |
286 |
C<$Coro::idle> hook. |
287 |
|
288 |
Please note that the current coro will I<not> be put into the ready |
289 |
queue, so calling this function usually means you will never be called |
290 |
again unless something else (e.g. an event handler) calls C<< ->ready >>, |
291 |
thus waking you up. |
292 |
|
293 |
This makes C<schedule> I<the> generic method to use to block the current |
294 |
coro and wait for events: first you remember the current coro in |
295 |
a variable, then arrange for some callback of yours to call C<< ->ready |
296 |
>> on that once some event happens, and last you call C<schedule> to put |
297 |
yourself to sleep. Note that a lot of things can wake your coro up, |
298 |
so you need to check whether the event indeed happened, e.g. by storing the |
299 |
status in a variable. |
300 |
|
301 |
See B<HOW TO WAIT FOR A CALLBACK>, below, for some ways to wait for callbacks. |
302 |
|
303 |
=item cede |
304 |
|
305 |
"Cede" to other coros. This function puts the current coro into |
306 |
the ready queue and calls C<schedule>, which has the effect of giving |
307 |
up the current "timeslice" to other coros of the same or higher |
308 |
priority. Once your coro gets its turn again it will automatically be |
309 |
resumed. |
310 |
|
311 |
This function is often called C<yield> in other languages. |
312 |
|
313 |
=item Coro::cede_notself |
314 |
|
315 |
Works like cede, but is not exported by default and will cede to I<any> |
316 |
coro, regardless of priority. This is useful sometimes to ensure |
317 |
progress is made. |
318 |
|
319 |
=item terminate [arg...] |
320 |
|
321 |
Terminates the current coro with the given status values (see L<cancel>). |
322 |
|
323 |
=item Coro::on_enter BLOCK, Coro::on_leave BLOCK |
324 |
|
325 |
These function install enter and leave winders in the current scope. The |
326 |
enter block will be executed when on_enter is called and whenever the |
327 |
current coro is re-entered by the scheduler, while the leave block is |
328 |
executed whenever the current coro is blocked by the scheduler, and |
329 |
also when the containing scope is exited (by whatever means, be it exit, |
330 |
die, last etc.). |
331 |
|
332 |
I<Neither invoking the scheduler, nor exceptions, are allowed within those |
333 |
BLOCKs>. That means: do not even think about calling C<die> without an |
334 |
eval, and do not even think of entering the scheduler in any way. |
335 |
|
336 |
Since both BLOCKs are tied to the current scope, they will automatically |
337 |
be removed when the current scope exits. |
338 |
|
339 |
These functions implement the same concept as C<dynamic-wind> in scheme |
340 |
does, and are useful when you want to localise some resource to a specific |
341 |
coro. |
342 |
|
343 |
They slow down coro switching considerably for coros that use |
344 |
them (But coro switching is still reasonably fast if the handlers are |
345 |
fast). |
346 |
|
347 |
These functions are best understood by an example: The following function |
348 |
will change the current timezone to "Antarctica/South_Pole", which |
349 |
requires a call to C<tzset>, but by using C<on_enter> and C<on_leave>, |
350 |
which remember/change the current timezone and restore the previous |
351 |
value, respectively, the timezone is only changes for the coro that |
352 |
installed 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 |
364 |
}; |
365 |
|
366 |
Coro::on_leave { |
367 |
$ENV{TZ} = $old_tz; |
368 |
tzset; # restore old value |
369 |
}; |
370 |
|
371 |
# at this place, the timezone is Antarctica/South_Pole, |
372 |
# without disturbing the TZ of any other coro. |
373 |
}; |
374 |
|
375 |
This can be used to localise about any resource (locale, uid, current |
376 |
working directory etc.) to a block, despite the existance of other |
377 |
coros. |
378 |
|
379 |
=item killall |
380 |
|
381 |
Kills/terminates/cancels all coros except the currently running one. |
382 |
|
383 |
Note that while this will try to free some of the main interpreter |
384 |
resources if the calling coro isn't the main coro, but one |
385 |
cannot free all of them, so if a coro that is not the main coro |
386 |
calls this function, there will be some one-time resource leak. |
387 |
|
388 |
=cut |
389 |
|
390 |
sub killall { |
391 |
for (Coro::State::list) { |
392 |
$_->cancel |
393 |
if $_ != $current && UNIVERSAL::isa $_, "Coro"; |
394 |
} |
395 |
} |
396 |
|
397 |
=back |
398 |
|
399 |
=head1 CORO OBJECT METHODS |
400 |
|
401 |
These are the methods you can call on coro objects (or to create |
402 |
them). |
403 |
|
404 |
=over 4 |
405 |
|
406 |
=item new Coro \&sub [, @args...] |
407 |
|
408 |
Create a new coro and return it. When the sub returns, the coro |
409 |
automatically terminates as if C<terminate> with the returned values were |
410 |
called. To make the coro run you must first put it into the ready |
411 |
queue by calling the ready method. |
412 |
|
413 |
See C<async> and C<Coro::State::new> for additional info about the |
414 |
coro environment. |
415 |
|
416 |
=cut |
417 |
|
418 |
sub _coro_run { |
419 |
terminate &{+shift}; |
420 |
} |
421 |
|
422 |
=item $success = $coro->ready |
423 |
|
424 |
Put the given coro into the end of its ready queue (there is one |
425 |
queue for each priority) and return true. If the coro is already in |
426 |
the ready queue, do nothing and return false. |
427 |
|
428 |
This ensures that the scheduler will resume this coro automatically |
429 |
once all the coro of higher priority and all coro of the same |
430 |
priority that were put into the ready queue earlier have been resumed. |
431 |
|
432 |
=item $is_ready = $coro->is_ready |
433 |
|
434 |
Returns true iff the Coro object is in the ready queue. Unless the Coro |
435 |
object gets destroyed, it will eventually be scheduled by the scheduler. |
436 |
|
437 |
=item $is_running = $coro->is_running |
438 |
|
439 |
Returns true iff the Coro object is currently running. Only one Coro object |
440 |
can ever be in the running state (but it currently is possible to have |
441 |
multiple running Coro::States). |
442 |
|
443 |
=item $is_suspended = $coro->is_suspended |
444 |
|
445 |
Returns true iff this Coro object has been suspended. Suspended Coros will |
446 |
not ever be scheduled. |
447 |
|
448 |
=item $coro->cancel (arg...) |
449 |
|
450 |
Terminates the given Coro and makes it return the given arguments as |
451 |
status (default: the empty list). Never returns if the Coro is the |
452 |
current Coro. |
453 |
|
454 |
=cut |
455 |
|
456 |
sub cancel { |
457 |
my $self = shift; |
458 |
|
459 |
if ($current == $self) { |
460 |
terminate @_; |
461 |
} else { |
462 |
$self->{_status} = [@_]; |
463 |
Coro::State::cancel $self; |
464 |
} |
465 |
} |
466 |
|
467 |
=item $coro->schedule_to |
468 |
|
469 |
Puts the current coro to sleep (like C<Coro::schedule>), but instead |
470 |
of continuing with the next coro from the ready queue, always switch to |
471 |
the given coro object (regardless of priority etc.). The readyness |
472 |
state of that coro isn't changed. |
473 |
|
474 |
This is an advanced method for special cases - I'd love to hear about any |
475 |
uses for this one. |
476 |
|
477 |
=item $coro->cede_to |
478 |
|
479 |
Like C<schedule_to>, but puts the current coro into the ready |
480 |
queue. This has the effect of temporarily switching to the given |
481 |
coro, and continuing some time later. |
482 |
|
483 |
This is an advanced method for special cases - I'd love to hear about any |
484 |
uses for this one. |
485 |
|
486 |
=item $coro->throw ([$scalar]) |
487 |
|
488 |
If C<$throw> is specified and defined, it will be thrown as an exception |
489 |
inside the coro at the next convenient point in time. Otherwise |
490 |
clears the exception object. |
491 |
|
492 |
Coro will check for the exception each time a schedule-like-function |
493 |
returns, i.e. after each C<schedule>, C<cede>, C<< Coro::Semaphore->down |
494 |
>>, C<< Coro::Handle->readable >> and so on. Most of these functions |
495 |
detect this case and return early in case an exception is pending. |
496 |
|
497 |
The exception object will be thrown "as is" with the specified scalar in |
498 |
C<$@>, i.e. if it is a string, no line number or newline will be appended |
499 |
(unlike with C<die>). |
500 |
|
501 |
This can be used as a softer means than C<cancel> to ask a coro to |
502 |
end itself, although there is no guarantee that the exception will lead to |
503 |
termination, and if the exception isn't caught it might well end the whole |
504 |
program. |
505 |
|
506 |
You might also think of C<throw> as being the moral equivalent of |
507 |
C<kill>ing a coro with a signal (in this case, a scalar). |
508 |
|
509 |
=item $coro->join |
510 |
|
511 |
Wait until the coro terminates and return any values given to the |
512 |
C<terminate> or C<cancel> functions. C<join> can be called concurrently |
513 |
from multiple coro, and all will be resumed and given the status |
514 |
return once the C<$coro> terminates. |
515 |
|
516 |
=cut |
517 |
|
518 |
sub join { |
519 |
my $self = shift; |
520 |
|
521 |
unless ($self->{_status}) { |
522 |
my $current = $current; |
523 |
|
524 |
push @{$self->{_on_destroy}}, sub { |
525 |
$current->ready; |
526 |
undef $current; |
527 |
}; |
528 |
|
529 |
&schedule while $current; |
530 |
} |
531 |
|
532 |
wantarray ? @{$self->{_status}} : $self->{_status}[0]; |
533 |
} |
534 |
|
535 |
=item $coro->on_destroy (\&cb) |
536 |
|
537 |
Registers a callback that is called when this coro gets destroyed, |
538 |
but before it is joined. The callback gets passed the terminate arguments, |
539 |
if any, and I<must not> die, under any circumstances. |
540 |
|
541 |
=cut |
542 |
|
543 |
sub on_destroy { |
544 |
my ($self, $cb) = @_; |
545 |
|
546 |
push @{ $self->{_on_destroy} }, $cb; |
547 |
} |
548 |
|
549 |
=item $oldprio = $coro->prio ($newprio) |
550 |
|
551 |
Sets (or gets, if the argument is missing) the priority of the |
552 |
coro. Higher priority coro get run before lower priority |
553 |
coro. Priorities are small signed integers (currently -4 .. +3), |
554 |
that you can refer to using PRIO_xxx constants (use the import tag :prio |
555 |
to get then): |
556 |
|
557 |
PRIO_MAX > PRIO_HIGH > PRIO_NORMAL > PRIO_LOW > PRIO_IDLE > PRIO_MIN |
558 |
3 > 1 > 0 > -1 > -3 > -4 |
559 |
|
560 |
# set priority to HIGH |
561 |
current->prio (PRIO_HIGH); |
562 |
|
563 |
The idle coro ($Coro::idle) always has a lower priority than any |
564 |
existing coro. |
565 |
|
566 |
Changing the priority of the current coro will take effect immediately, |
567 |
but changing the priority of coro in the ready queue (but not |
568 |
running) will only take effect after the next schedule (of that |
569 |
coro). This is a bug that will be fixed in some future version. |
570 |
|
571 |
=item $newprio = $coro->nice ($change) |
572 |
|
573 |
Similar to C<prio>, but subtract the given value from the priority (i.e. |
574 |
higher values mean lower priority, just as in unix). |
575 |
|
576 |
=item $olddesc = $coro->desc ($newdesc) |
577 |
|
578 |
Sets (or gets in case the argument is missing) the description for this |
579 |
coro. This is just a free-form string you can associate with a |
580 |
coro. |
581 |
|
582 |
This method simply sets the C<< $coro->{desc} >> member to the given |
583 |
string. You can modify this member directly if you wish. |
584 |
|
585 |
=cut |
586 |
|
587 |
sub desc { |
588 |
my $old = $_[0]{desc}; |
589 |
$_[0]{desc} = $_[1] if @_ > 1; |
590 |
$old; |
591 |
} |
592 |
|
593 |
sub transfer { |
594 |
require Carp; |
595 |
Carp::croak ("You must not call ->transfer on Coro objects. Use Coro::State objects or the ->schedule_to method. Caught"); |
596 |
} |
597 |
|
598 |
=back |
599 |
|
600 |
=head1 GLOBAL FUNCTIONS |
601 |
|
602 |
=over 4 |
603 |
|
604 |
=item Coro::nready |
605 |
|
606 |
Returns the number of coro that are currently in the ready state, |
607 |
i.e. that can be switched to by calling C<schedule> directory or |
608 |
indirectly. The value C<0> means that the only runnable coro is the |
609 |
currently running one, so C<cede> would have no effect, and C<schedule> |
610 |
would cause a deadlock unless there is an idle handler that wakes up some |
611 |
coro. |
612 |
|
613 |
=item my $guard = Coro::guard { ... } |
614 |
|
615 |
This function still exists, but is deprecated. Please use the |
616 |
C<Guard::guard> function instead. |
617 |
|
618 |
=cut |
619 |
|
620 |
BEGIN { *guard = \&Guard::guard } |
621 |
|
622 |
=item unblock_sub { ... } |
623 |
|
624 |
This utility function takes a BLOCK or code reference and "unblocks" it, |
625 |
returning a new coderef. Unblocking means that calling the new coderef |
626 |
will return immediately without blocking, returning nothing, while the |
627 |
original code ref will be called (with parameters) from within another |
628 |
coro. |
629 |
|
630 |
The reason this function exists is that many event libraries (such as the |
631 |
venerable L<Event|Event> module) are not thread-safe (a weaker form |
632 |
of reentrancy). This means you must not block within event callbacks, |
633 |
otherwise you might suffer from crashes or worse. The only event library |
634 |
currently known that is safe to use without C<unblock_sub> is L<EV>. |
635 |
|
636 |
This function allows your callbacks to block by executing them in another |
637 |
coro where it is safe to block. One example where blocking is handy |
638 |
is when you use the L<Coro::AIO|Coro::AIO> functions to save results to |
639 |
disk, for example. |
640 |
|
641 |
In short: simply use C<unblock_sub { ... }> instead of C<sub { ... }> when |
642 |
creating event callbacks that want to block. |
643 |
|
644 |
If your handler does not plan to block (e.g. simply sends a message to |
645 |
another coro, or puts some other coro into the ready queue), there is |
646 |
no reason to use C<unblock_sub>. |
647 |
|
648 |
Note that you also need to use C<unblock_sub> for any other callbacks that |
649 |
are indirectly executed by any C-based event loop. For example, when you |
650 |
use a module that uses L<AnyEvent> (and you use L<Coro::AnyEvent>) and it |
651 |
provides callbacks that are the result of some event callback, then you |
652 |
must not block either, or use C<unblock_sub>. |
653 |
|
654 |
=cut |
655 |
|
656 |
our @unblock_queue; |
657 |
|
658 |
# we create a special coro because we want to cede, |
659 |
# to reduce pressure on the coro pool (because most callbacks |
660 |
# return immediately and can be reused) and because we cannot cede |
661 |
# inside an event callback. |
662 |
our $unblock_scheduler = new Coro sub { |
663 |
while () { |
664 |
while (my $cb = pop @unblock_queue) { |
665 |
&async_pool (@$cb); |
666 |
|
667 |
# for short-lived callbacks, this reduces pressure on the coro pool |
668 |
# as the chance is very high that the async_poll coro will be back |
669 |
# in the idle state when cede returns |
670 |
cede; |
671 |
} |
672 |
schedule; # sleep well |
673 |
} |
674 |
}; |
675 |
$unblock_scheduler->{desc} = "[unblock_sub scheduler]"; |
676 |
|
677 |
sub unblock_sub(&) { |
678 |
my $cb = shift; |
679 |
|
680 |
sub { |
681 |
unshift @unblock_queue, [$cb, @_]; |
682 |
$unblock_scheduler->ready; |
683 |
} |
684 |
} |
685 |
|
686 |
=item $cb = Coro::rouse_cb |
687 |
|
688 |
Create and return a "rouse callback". That's a code reference that, |
689 |
when called, will remember a copy of its arguments and notify the owner |
690 |
coro of the callback. |
691 |
|
692 |
See the next function. |
693 |
|
694 |
=item @args = Coro::rouse_wait [$cb] |
695 |
|
696 |
Wait for the specified rouse callback (or the last one that was created in |
697 |
this coro). |
698 |
|
699 |
As soon as the callback is invoked (or when the callback was invoked |
700 |
before C<rouse_wait>), it will return the arguments originally passed to |
701 |
the rouse callback. |
702 |
|
703 |
See the section B<HOW TO WAIT FOR A CALLBACK> for an actual usage example. |
704 |
|
705 |
=back |
706 |
|
707 |
=cut |
708 |
|
709 |
1; |
710 |
|
711 |
=head1 HOW TO WAIT FOR A CALLBACK |
712 |
|
713 |
It is very common for a coro to wait for some callback to be |
714 |
called. This occurs naturally when you use coro in an otherwise |
715 |
event-based program, or when you use event-based libraries. |
716 |
|
717 |
These typically register a callback for some event, and call that callback |
718 |
when the event occured. In a coro, however, you typically want to |
719 |
just wait for the event, simplyifying things. |
720 |
|
721 |
For example C<< AnyEvent->child >> registers a callback to be called when |
722 |
a specific child has exited: |
723 |
|
724 |
my $child_watcher = AnyEvent->child (pid => $pid, cb => sub { ... }); |
725 |
|
726 |
But from within a coro, you often just want to write this: |
727 |
|
728 |
my $status = wait_for_child $pid; |
729 |
|
730 |
Coro offers two functions specifically designed to make this easy, |
731 |
C<Coro::rouse_cb> and C<Coro::rouse_wait>. |
732 |
|
733 |
The first function, C<rouse_cb>, generates and returns a callback that, |
734 |
when invoked, will save its arguments and notify the coro that |
735 |
created the callback. |
736 |
|
737 |
The second function, C<rouse_wait>, waits for the callback to be called |
738 |
(by calling C<schedule> to go to sleep) and returns the arguments |
739 |
originally passed to the callback. |
740 |
|
741 |
Using these functions, it becomes easy to write the C<wait_for_child> |
742 |
function mentioned above: |
743 |
|
744 |
sub wait_for_child($) { |
745 |
my ($pid) = @_; |
746 |
|
747 |
my $watcher = AnyEvent->child (pid => $pid, cb => Coro::rouse_cb); |
748 |
|
749 |
my ($rpid, $rstatus) = Coro::rouse_wait; |
750 |
$rstatus |
751 |
} |
752 |
|
753 |
In the case where C<rouse_cb> and C<rouse_wait> are not flexible enough, |
754 |
you can roll your own, using C<schedule>: |
755 |
|
756 |
sub wait_for_child($) { |
757 |
my ($pid) = @_; |
758 |
|
759 |
# store the current coro in $current, |
760 |
# and provide result variables for the closure passed to ->child |
761 |
my $current = $Coro::current; |
762 |
my ($done, $rstatus); |
763 |
|
764 |
# pass a closure to ->child |
765 |
my $watcher = AnyEvent->child (pid => $pid, cb => sub { |
766 |
$rstatus = $_[1]; # remember rstatus |
767 |
$done = 1; # mark $rstatus as valud |
768 |
}); |
769 |
|
770 |
# wait until the closure has been called |
771 |
schedule while !$done; |
772 |
|
773 |
$rstatus |
774 |
} |
775 |
|
776 |
|
777 |
=head1 BUGS/LIMITATIONS |
778 |
|
779 |
=over 4 |
780 |
|
781 |
=item fork with pthread backend |
782 |
|
783 |
When Coro is compiled using the pthread backend (which isn't recommended |
784 |
but required on many BSDs as their libcs are completely broken), then |
785 |
coro will not survive a fork. There is no known workaround except to |
786 |
fix your libc and use a saner backend. |
787 |
|
788 |
=item perl process emulation ("threads") |
789 |
|
790 |
This module is not perl-pseudo-thread-safe. You should only ever use this |
791 |
module from the first thread (this requirement might be removed in the |
792 |
future to allow per-thread schedulers, but Coro::State does not yet allow |
793 |
this). I recommend disabling thread support and using processes, as having |
794 |
the windows process emulation enabled under unix roughly halves perl |
795 |
performance, even when not used. |
796 |
|
797 |
=item coro switching is not signal safe |
798 |
|
799 |
You must not switch to another coro from within a signal handler |
800 |
(only relevant with %SIG - most event libraries provide safe signals). |
801 |
|
802 |
That means you I<MUST NOT> call any function that might "block" the |
803 |
current coro - C<cede>, C<schedule> C<< Coro::Semaphore->down >> or |
804 |
anything that calls those. Everything else, including calling C<ready>, |
805 |
works. |
806 |
|
807 |
=back |
808 |
|
809 |
|
810 |
=head1 SEE ALSO |
811 |
|
812 |
Event-Loop integration: L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>. |
813 |
|
814 |
Debugging: L<Coro::Debug>. |
815 |
|
816 |
Support/Utility: L<Coro::Specific>, L<Coro::Util>. |
817 |
|
818 |
Locking and IPC: L<Coro::Signal>, L<Coro::Channel>, L<Coro::Semaphore>, |
819 |
L<Coro::SemaphoreSet>, L<Coro::RWLock>. |
820 |
|
821 |
I/O and Timers: L<Coro::Timer>, L<Coro::Handle>, L<Coro::Socket>, L<Coro::AIO>. |
822 |
|
823 |
Compatibility with other modules: L<Coro::LWP> (but see also L<AnyEvent::HTTP> for |
824 |
a better-working alternative), L<Coro::BDB>, L<Coro::Storable>, |
825 |
L<Coro::Select>. |
826 |
|
827 |
XS API: L<Coro::MakeMaker>. |
828 |
|
829 |
Low level Configuration, Thread Environment, Continuations: L<Coro::State>. |
830 |
|
831 |
=head1 AUTHOR |
832 |
|
833 |
Marc Lehmann <schmorp@schmorp.de> |
834 |
http://home.schmorp.de/ |
835 |
|
836 |
=cut |
837 |
|