1 |
NAME |
2 |
Coro - coroutine process abstraction |
3 |
|
4 |
SYNOPSIS |
5 |
use Coro; |
6 |
|
7 |
async { |
8 |
# some asynchronous thread of execution |
9 |
print "2\n"; |
10 |
cede; # yield back to main |
11 |
print "4\n"; |
12 |
}; |
13 |
print "1\n"; |
14 |
cede; # yield to coroutine |
15 |
print "3\n"; |
16 |
cede; # and again |
17 |
|
18 |
# use locking |
19 |
use Coro::Semaphore; |
20 |
my $lock = new Coro::Semaphore; |
21 |
my $locked; |
22 |
|
23 |
$lock->down; |
24 |
$locked = 1; |
25 |
$lock->up; |
26 |
|
27 |
DESCRIPTION |
28 |
This module collection manages coroutines. Coroutines are similar to |
29 |
threads but don't (in general) run in parallel at the same time even on |
30 |
SMP machines. The specific flavor of coroutine used in this module also |
31 |
guarantees you that it will not switch between coroutines unless |
32 |
necessary, at easily-identified points in your program, so locking and |
33 |
parallel access are rarely an issue, making coroutine programming much |
34 |
safer and easier than threads programming. |
35 |
|
36 |
Unlike a normal perl program, however, coroutines allow you to have |
37 |
multiple running interpreters that share data, which is especially |
38 |
useful to code pseudo-parallel processes and for event-based |
39 |
programming, such as multiple HTTP-GET requests running concurrently. |
40 |
See Coro::AnyEvent to learn more. |
41 |
|
42 |
Coroutines are also useful because Perl has no support for threads (the |
43 |
so called "threads" that perl offers are nothing more than the (bad) |
44 |
process emulation coming from the Windows platform: On standard |
45 |
operating systems they serve no purpose whatsoever, except by making |
46 |
your programs slow and making them use a lot of memory. Best disable |
47 |
them when building perl, or aks your software vendor/distributor to do |
48 |
it for you). |
49 |
|
50 |
In this module, coroutines are defined as "callchain + lexical variables |
51 |
+ @_ + $_ + $@ + $/ + C stack), that is, a coroutine has its own |
52 |
callchain, its own set of lexicals and its own set of perls most |
53 |
important global variables (see Coro::State for more configuration). |
54 |
|
55 |
$Coro::main |
56 |
This variable stores the coroutine object that represents the main |
57 |
program. While you cna "ready" it and do most other things you can |
58 |
do to coroutines, it is mainly useful to compare again |
59 |
$Coro::current, to see whether you are running in the main program |
60 |
or not. |
61 |
|
62 |
$Coro::current |
63 |
The coroutine object representing the current coroutine (the last |
64 |
coroutine that the Coro scheduler switched to). The initial value is |
65 |
$main (of course). |
66 |
|
67 |
This variable is strictly *read-only*. You can take copies of the |
68 |
value stored in it and use it as any other coroutine object, but you |
69 |
must not otherwise modify the variable itself. |
70 |
|
71 |
$Coro::idle |
72 |
This variable is mainly useful to integrate Coro into event loops. |
73 |
It is usually better to rely on Coro::AnyEvent or L"Coro::EV", as |
74 |
this is pretty low-level functionality. |
75 |
|
76 |
This variable stores a callback that is called whenever the |
77 |
scheduler finds no ready coroutines to run. The default |
78 |
implementation prints "FATAL: deadlock detected" and exits, because |
79 |
the program has no other way to continue. |
80 |
|
81 |
This hook is overwritten by modules such as "Coro::Timer" and |
82 |
"Coro::AnyEvent" to wait on an external event that hopefully wake up |
83 |
a coroutine so the scheduler can run it. |
84 |
|
85 |
Note that the callback *must not*, under any circumstances, block |
86 |
the current coroutine. Normally, this is achieved by having an "idle |
87 |
coroutine" that calls the event loop and then blocks again, and then |
88 |
readying that coroutine in the idle handler. |
89 |
|
90 |
See Coro::Event or Coro::AnyEvent for examples of using this |
91 |
technique. |
92 |
|
93 |
Please note that if your callback recursively invokes perl (e.g. for |
94 |
event handlers), then it must be prepared to be called recursively |
95 |
itself. |
96 |
|
97 |
SIMPLE COROUTINE CREATION |
98 |
async { ... } [@args...] |
99 |
Create a new coroutine and return it's coroutine object (usually |
100 |
unused). The coroutine will be put into the ready queue, so it will |
101 |
start running automatically on the next scheduler run. |
102 |
|
103 |
The first argument is a codeblock/closure that should be executed in |
104 |
the coroutine. When it returns argument returns the coroutine is |
105 |
automatically terminated. |
106 |
|
107 |
The remaining arguments are passed as arguments to the closure. |
108 |
|
109 |
See the "Coro::State::new" constructor for info about the coroutine |
110 |
environment in which coroutines are executed. |
111 |
|
112 |
Calling "exit" in a coroutine will do the same as calling exit |
113 |
outside the coroutine. Likewise, when the coroutine dies, the |
114 |
program will exit, just as it would in the main program. |
115 |
|
116 |
If you do not want that, you can provide a default "die" handler, or |
117 |
simply avoid dieing (by use of "eval"). |
118 |
|
119 |
Example: Create a new coroutine that just prints its arguments. |
120 |
|
121 |
async { |
122 |
print "@_\n"; |
123 |
} 1,2,3,4; |
124 |
|
125 |
async_pool { ... } [@args...] |
126 |
Similar to "async", but uses a coroutine pool, so you should not |
127 |
call terminate or join on it (although you are allowed to), and you |
128 |
get a coroutine that might have executed other code already (which |
129 |
can be good or bad :). |
130 |
|
131 |
On the plus side, this function is faster than creating (and |
132 |
destroying) a completely new coroutine, so if you need a lot of |
133 |
generic coroutines in quick successsion, use "async_pool", not |
134 |
"async". |
135 |
|
136 |
The code block is executed in an "eval" context and a warning will |
137 |
be issued in case of an exception instead of terminating the |
138 |
program, as "async" does. As the coroutine is being reused, stuff |
139 |
like "on_destroy" will not work in the expected way, unless you call |
140 |
terminate or cancel, which somehow defeats the purpose of pooling |
141 |
(but is fine in the exceptional case). |
142 |
|
143 |
The priority will be reset to 0 after each run, tracing will be |
144 |
disabled, the description will be reset and the default output |
145 |
filehandle gets restored, so you can change all these. Otherwise the |
146 |
coroutine will be re-used "as-is": most notably if you change other |
147 |
per-coroutine global stuff such as $/ you *must needs* to revert |
148 |
that change, which is most simply done by using local as in: " local |
149 |
$/ ". |
150 |
|
151 |
The pool size is limited to 8 idle coroutines (this can be adjusted |
152 |
by changing $Coro::POOL_SIZE), and there can be as many non-idle |
153 |
coros as required. |
154 |
|
155 |
If you are concerned about pooled coroutines growing a lot because a |
156 |
single "async_pool" used a lot of stackspace you can e.g. |
157 |
"async_pool { terminate }" once per second or so to slowly replenish |
158 |
the pool. In addition to that, when the stacks used by a handler |
159 |
grows larger than 16kb (adjustable via $Coro::POOL_RSS) it will also |
160 |
be destroyed. |
161 |
|
162 |
STATIC METHODS |
163 |
Static methods are actually functions that operate on the current |
164 |
coroutine. |
165 |
|
166 |
schedule |
167 |
Calls the scheduler. The scheduler will find the next coroutine that |
168 |
is to be run from the ready queue and switches to it. The next |
169 |
coroutine to be run is simply the one with the highest priority that |
170 |
is longest in its ready queue. If there is no coroutine ready, it |
171 |
will clal the $Coro::idle hook. |
172 |
|
173 |
Please note that the current coroutine will *not* be put into the |
174 |
ready queue, so calling this function usually means you will never |
175 |
be called again unless something else (e.g. an event handler) calls |
176 |
"->ready", thus waking you up. |
177 |
|
178 |
This makes "schedule" *the* generic method to use to block the |
179 |
current coroutine and wait for events: first you remember the |
180 |
current coroutine in a variable, then arrange for some callback of |
181 |
yours to call "->ready" on that once some event happens, and last |
182 |
you call "schedule" to put yourself to sleep. Note that a lot of |
183 |
things can wake your coroutine up, so you need to check whether the |
184 |
event indeed happened, e.g. by storing the status in a variable. |
185 |
|
186 |
The canonical way to wait on external events is this: |
187 |
|
188 |
{ |
189 |
# remember current coroutine |
190 |
my $current = $Coro::current; |
191 |
|
192 |
# register a hypothetical event handler |
193 |
on_event_invoke sub { |
194 |
# wake up sleeping coroutine |
195 |
$current->ready; |
196 |
undef $current; |
197 |
}; |
198 |
|
199 |
# call schedule until event occurred. |
200 |
# in case we are woken up for other reasons |
201 |
# (current still defined), loop. |
202 |
Coro::schedule while $current; |
203 |
} |
204 |
|
205 |
cede |
206 |
"Cede" to other coroutines. This function puts the current coroutine |
207 |
into the ready queue and calls "schedule", which has the effect of |
208 |
giving up the current "timeslice" to other coroutines of the same or |
209 |
higher priority. Once your coroutine gets its turn again it will |
210 |
automatically be resumed. |
211 |
|
212 |
This function is often called "yield" in other languages. |
213 |
|
214 |
Coro::cede_notself |
215 |
Works like cede, but is not exported by default and will cede to |
216 |
*any* coroutine, regardless of priority. This is useful sometimes to |
217 |
ensure progress is made. |
218 |
|
219 |
terminate [arg...] |
220 |
Terminates the current coroutine with the given status values (see |
221 |
cancel). |
222 |
|
223 |
killall |
224 |
Kills/terminates/cancels all coroutines except the currently running |
225 |
one. This is useful after a fork, either in the child or the parent, |
226 |
as usually only one of them should inherit the running coroutines. |
227 |
|
228 |
Note that while this will try to free some of the main programs |
229 |
resources, you cannot free all of them, so if a coroutine that is |
230 |
not the main program calls this function, there will be some |
231 |
one-time resource leak. |
232 |
|
233 |
COROUTINE METHODS |
234 |
These are the methods you can call on coroutine objects (or to create |
235 |
them). |
236 |
|
237 |
new Coro \&sub [, @args...] |
238 |
Create a new coroutine and return it. When the sub returns, the |
239 |
coroutine automatically terminates as if "terminate" with the |
240 |
returned values were called. To make the coroutine run you must |
241 |
first put it into the ready queue by calling the ready method. |
242 |
|
243 |
See "async" and "Coro::State::new" for additional info about the |
244 |
coroutine environment. |
245 |
|
246 |
$success = $coroutine->ready |
247 |
Put the given coroutine into the end of its ready queue (there is |
248 |
one queue for each priority) and return true. If the coroutine is |
249 |
already in the ready queue, do nothing and return false. |
250 |
|
251 |
This ensures that the scheduler will resume this coroutine |
252 |
automatically once all the coroutines of higher priority and all |
253 |
coroutines of the same priority that were put into the ready queue |
254 |
earlier have been resumed. |
255 |
|
256 |
$is_ready = $coroutine->is_ready |
257 |
Return whether the coroutine is currently the ready queue or not, |
258 |
|
259 |
$coroutine->cancel (arg...) |
260 |
Terminates the given coroutine and makes it return the given |
261 |
arguments as status (default: the empty list). Never returns if the |
262 |
coroutine is the current coroutine. |
263 |
|
264 |
$coroutine->join |
265 |
Wait until the coroutine terminates and return any values given to |
266 |
the "terminate" or "cancel" functions. "join" can be called |
267 |
concurrently from multiple coroutines, and all will be resumed and |
268 |
given the status return once the $coroutine terminates. |
269 |
|
270 |
$coroutine->on_destroy (\&cb) |
271 |
Registers a callback that is called when this coroutine gets |
272 |
destroyed, but before it is joined. The callback gets passed the |
273 |
terminate arguments, if any, and *must not* die, under any |
274 |
circumstances. |
275 |
|
276 |
$oldprio = $coroutine->prio ($newprio) |
277 |
Sets (or gets, if the argument is missing) the priority of the |
278 |
coroutine. Higher priority coroutines get run before lower priority |
279 |
coroutines. Priorities are small signed integers (currently -4 .. |
280 |
+3), that you can refer to using PRIO_xxx constants (use the import |
281 |
tag :prio to get then): |
282 |
|
283 |
PRIO_MAX > PRIO_HIGH > PRIO_NORMAL > PRIO_LOW > PRIO_IDLE > PRIO_MIN |
284 |
3 > 1 > 0 > -1 > -3 > -4 |
285 |
|
286 |
# set priority to HIGH |
287 |
current->prio(PRIO_HIGH); |
288 |
|
289 |
The idle coroutine ($Coro::idle) always has a lower priority than |
290 |
any existing coroutine. |
291 |
|
292 |
Changing the priority of the current coroutine will take effect |
293 |
immediately, but changing the priority of coroutines in the ready |
294 |
queue (but not running) will only take effect after the next |
295 |
schedule (of that coroutine). This is a bug that will be fixed in |
296 |
some future version. |
297 |
|
298 |
$newprio = $coroutine->nice ($change) |
299 |
Similar to "prio", but subtract the given value from the priority |
300 |
(i.e. higher values mean lower priority, just as in unix). |
301 |
|
302 |
$olddesc = $coroutine->desc ($newdesc) |
303 |
Sets (or gets in case the argument is missing) the description for |
304 |
this coroutine. This is just a free-form string you can associate |
305 |
with a coroutine. |
306 |
|
307 |
This method simply sets the "$coroutine->{desc}" member to the given |
308 |
string. You can modify this member directly if you wish. |
309 |
|
310 |
$coroutine->throw ([$scalar]) |
311 |
If $throw is specified and defined, it will be thrown as an |
312 |
exception inside the coroutine at the next convinient point in time |
313 |
(usually after it gains control at the next schedule/transfer/cede). |
314 |
Otherwise clears the exception object. |
315 |
|
316 |
The exception object will be thrown "as is" with the specified |
317 |
scalar in $@, i.e. if it is a string, no line number or newline will |
318 |
be appended (unlike with "die"). |
319 |
|
320 |
This can be used as a softer means than "cancel" to ask a coroutine |
321 |
to end itself, although there is no guarentee that the exception |
322 |
will lead to termination, and if the exception isn't caught it might |
323 |
well end the whole program. |
324 |
|
325 |
GLOBAL FUNCTIONS |
326 |
Coro::nready |
327 |
Returns the number of coroutines that are currently in the ready |
328 |
state, i.e. that can be switched to by calling "schedule" directory |
329 |
or indirectly. The value 0 means that the only runnable coroutine is |
330 |
the currently running one, so "cede" would have no effect, and |
331 |
"schedule" would cause a deadlock unless there is an idle handler |
332 |
that wakes up some coroutines. |
333 |
|
334 |
my $guard = Coro::guard { ... } |
335 |
This creates and returns a guard object. Nothing happens until the |
336 |
object gets destroyed, in which case the codeblock given as argument |
337 |
will be executed. This is useful to free locks or other resources in |
338 |
case of a runtime error or when the coroutine gets canceled, as in |
339 |
both cases the guard block will be executed. The guard object |
340 |
supports only one method, "->cancel", which will keep the codeblock |
341 |
from being executed. |
342 |
|
343 |
Example: set some flag and clear it again when the coroutine gets |
344 |
canceled or the function returns: |
345 |
|
346 |
sub do_something { |
347 |
my $guard = Coro::guard { $busy = 0 }; |
348 |
$busy = 1; |
349 |
|
350 |
# do something that requires $busy to be true |
351 |
} |
352 |
|
353 |
unblock_sub { ... } |
354 |
This utility function takes a BLOCK or code reference and "unblocks" |
355 |
it, returning a new coderef. Unblocking means that calling the new |
356 |
coderef will return immediately without blocking, returning nothing, |
357 |
while the original code ref will be called (with parameters) from |
358 |
within another coroutine. |
359 |
|
360 |
The reason this function exists is that many event libraries (such |
361 |
as the venerable Event module) are not coroutine-safe (a weaker form |
362 |
of thread-safety). This means you must not block within event |
363 |
callbacks, otherwise you might suffer from crashes or worse. The |
364 |
only event library currently known that is safe to use without |
365 |
"unblock_sub" is EV. |
366 |
|
367 |
This function allows your callbacks to block by executing them in |
368 |
another coroutine where it is safe to block. One example where |
369 |
blocking is handy is when you use the Coro::AIO functions to save |
370 |
results to disk, for example. |
371 |
|
372 |
In short: simply use "unblock_sub { ... }" instead of "sub { ... }" |
373 |
when creating event callbacks that want to block. |
374 |
|
375 |
If your handler does not plan to block (e.g. simply sends a message |
376 |
to another coroutine, or puts some other coroutine into the ready |
377 |
queue), there is no reason to use "unblock_sub". |
378 |
|
379 |
Note that you also need to use "unblock_sub" for any other callbacks |
380 |
that are indirectly executed by any C-based event loop. For example, |
381 |
when you use a module that uses AnyEvent (and you use |
382 |
Coro::AnyEvent) and it provides callbacks that are the result of |
383 |
some event callback, then you must not block either, or use |
384 |
"unblock_sub". |
385 |
|
386 |
BUGS/LIMITATIONS |
387 |
This module is not perl-pseudo-thread-safe. You should only ever use |
388 |
this module from the same thread (this requirement might be removed in |
389 |
the future to allow per-thread schedulers, but Coro::State does not yet |
390 |
allow this). I recommend disabling thread support and using processes, |
391 |
as this is much faster and uses less memory. |
392 |
|
393 |
SEE ALSO |
394 |
Event-Loop integration: Coro::AnyEvent, Coro::EV, Coro::Event. |
395 |
|
396 |
Debugging: Coro::Debug. |
397 |
|
398 |
Support/Utility: Coro::Specific, Coro::Util. |
399 |
|
400 |
Locking/IPC: Coro::Signal, Coro::Channel, Coro::Semaphore, |
401 |
Coro::SemaphoreSet, Coro::RWLock. |
402 |
|
403 |
IO/Timers: Coro::Timer, Coro::Handle, Coro::Socket, Coro::AIO. |
404 |
|
405 |
Compatibility: Coro::LWP, Coro::BDB, Coro::Storable, Coro::Select. |
406 |
|
407 |
XS API: Coro::MakeMaker. |
408 |
|
409 |
Low level Configuration, Coroutine Environment: Coro::State. |
410 |
|
411 |
AUTHOR |
412 |
Marc Lehmann <schmorp@schmorp.de> |
413 |
http://home.schmorp.de/ |
414 |
|