1 |
NAME |
2 |
EV - perl interface to libev, a high performance full-featured event |
3 |
loop |
4 |
|
5 |
SYNOPSIS |
6 |
use EV; |
7 |
|
8 |
# TIMERS |
9 |
|
10 |
my $w = EV::timer 2, 0, sub { |
11 |
warn "is called after 2s"; |
12 |
}; |
13 |
|
14 |
my $w = EV::timer 2, 2, sub { |
15 |
warn "is called roughly every 2s (repeat = 2)"; |
16 |
}; |
17 |
|
18 |
undef $w; # destroy event watcher again |
19 |
|
20 |
my $w = EV::periodic 0, 60, 0, sub { |
21 |
warn "is called every minute, on the minute, exactly"; |
22 |
}; |
23 |
|
24 |
# IO |
25 |
|
26 |
my $w = EV::io *STDIN, EV::READ, sub { |
27 |
my ($w, $revents) = @_; # all callbacks receive the watcher and event mask |
28 |
warn "stdin is readable, you entered: ", <STDIN>; |
29 |
}; |
30 |
|
31 |
# SIGNALS |
32 |
|
33 |
my $w = EV::signal 'QUIT', sub { |
34 |
warn "sigquit received\n"; |
35 |
}; |
36 |
|
37 |
# CHILD/PID STATUS CHANGES |
38 |
|
39 |
my $w = EV::child 666, 0, sub { |
40 |
my ($w, $revents) = @_; |
41 |
my $status = $w->rstatus; |
42 |
}; |
43 |
|
44 |
# STAT CHANGES |
45 |
my $w = EV::stat "/etc/passwd", 10, sub { |
46 |
my ($w, $revents) = @_; |
47 |
warn $w->path, " has changed somehow.\n"; |
48 |
}; |
49 |
|
50 |
# MAINLOOP |
51 |
EV::run; # loop until EV::break is called or all watchers stop |
52 |
EV::run EV::RUN_ONCE; # block until at least one event could be handled |
53 |
EV::run EV::RUN_NOWAIT; # try to handle same events, but do not block |
54 |
|
55 |
BEFORE YOU START USING THIS MODULE |
56 |
If you only need timer, I/O, signal, child and idle watchers and not the |
57 |
advanced functionality of this module, consider using AnyEvent instead, |
58 |
specifically the simplified API described in AE. |
59 |
|
60 |
When used with EV as backend, the AE API is as fast as the native EV |
61 |
API, but your programs/modules will still run with many other event |
62 |
loops. |
63 |
|
64 |
DESCRIPTION |
65 |
This module provides an interface to libev |
66 |
(<http://software.schmorp.de/pkg/libev.html>). While the documentation |
67 |
below is comprehensive, one might also consult the documentation of |
68 |
libev itself (<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod> or |
69 |
perldoc EV::libev) for more subtle details on watcher semantics or some |
70 |
discussion on the available backends, or how to force a specific backend |
71 |
with "LIBEV_FLAGS", or just about in any case because it has much more |
72 |
detailed information. |
73 |
|
74 |
This module is very fast and scalable. It is actually so fast that you |
75 |
can use it through the AnyEvent module, stay portable to other event |
76 |
loops (if you don't rely on any watcher types not available through it) |
77 |
and still be faster than with any other event loop currently supported |
78 |
in Perl. |
79 |
|
80 |
PORTING FROM EV 3.X to 4.X |
81 |
EV version 4 introduces a number of incompatible changes summarised |
82 |
here. According to the depreciation strategy used by libev, there is a |
83 |
compatibility layer in place so programs should continue to run |
84 |
unchanged (the XS interface lacks this layer, so programs using that one |
85 |
need to be updated). |
86 |
|
87 |
This compatibility layer will be switched off in some future release. |
88 |
|
89 |
All changes relevant to Perl are renames of symbols, functions and |
90 |
methods: |
91 |
|
92 |
EV::loop => EV::run |
93 |
EV::LOOP_NONBLOCK => EV::RUN_NOWAIT |
94 |
EV::LOOP_ONESHOT => EV::RUN_ONCE |
95 |
|
96 |
EV::unloop => EV::break |
97 |
EV::UNLOOP_CANCEL => EV::BREAK_CANCEL |
98 |
EV::UNLOOP_ONE => EV::BREAK_ONE |
99 |
EV::UNLOOP_ALL => EV::BREAK_ALL |
100 |
|
101 |
EV::TIMEOUT => EV::TIMER |
102 |
|
103 |
EV::loop_count => EV::iteration |
104 |
EV::loop_depth => EV::depth |
105 |
EV::loop_verify => EV::verify |
106 |
|
107 |
The loop object methods corresponding to the functions above have been |
108 |
similarly renamed. |
109 |
|
110 |
MODULE EXPORTS |
111 |
This module does not export any symbols. |
112 |
|
113 |
EVENT LOOPS |
114 |
EV supports multiple event loops: There is a single "default event loop" |
115 |
that can handle everything including signals and child watchers, and any |
116 |
number of "dynamic event loops" that can use different backends (with |
117 |
various limitations), but no child and signal watchers. |
118 |
|
119 |
You do not have to do anything to create the default event loop: When |
120 |
the module is loaded a suitable backend is selected on the premise of |
121 |
selecting a working backend (which for example rules out kqueue on most |
122 |
BSDs). Modules should, unless they have "special needs" always use the |
123 |
default loop as this is fastest (perl-wise), best supported by other |
124 |
modules (e.g. AnyEvent or Coro) and most portable event loop. |
125 |
|
126 |
For specific programs you can create additional event loops dynamically. |
127 |
|
128 |
If you want to take advantage of kqueue (which often works properly for |
129 |
sockets only) even though the default loop doesn't enable it, you can |
130 |
*embed* a kqueue loop into the default loop: running the default loop |
131 |
will then also service the kqueue loop to some extent. See the example |
132 |
in the section about embed watchers for an example on how to achieve |
133 |
that. |
134 |
|
135 |
$loop = new EV::Loop [$flags] |
136 |
Create a new event loop as per the specified flags. Please refer to |
137 |
the "ev_loop_new ()" function description in the libev documentation |
138 |
(<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#GLOBAL_FUNCTI |
139 |
ONS>, or locally-installed as EV::libev manpage) for more info. |
140 |
|
141 |
The loop will automatically be destroyed when it is no longer |
142 |
referenced by any watcher and the loop object goes out of scope. |
143 |
|
144 |
If you are not embedding the loop, then Using "EV::FLAG_FORKCHECK" |
145 |
is recommended, as only the default event loop is protected by this |
146 |
module. If you *are* embedding this loop in the default loop, this |
147 |
is not necessary, as "EV::embed" automatically does the right thing |
148 |
on fork. |
149 |
|
150 |
$loop->loop_fork |
151 |
Must be called after a fork in the child, before entering or |
152 |
continuing the event loop. An alternative is to use |
153 |
"EV::FLAG_FORKCHECK" which calls this function automatically, at |
154 |
some performance loss (refer to the libev documentation). |
155 |
|
156 |
$loop->verify |
157 |
Calls "ev_verify" to make internal consistency checks (for debugging |
158 |
libev) and abort the program if any data structures were found to be |
159 |
corrupted. |
160 |
|
161 |
$loop = EV::default_loop [$flags] |
162 |
Return the default loop (which is a singleton object). Since this |
163 |
module already creates the default loop with default flags, |
164 |
specifying flags here will not have any effect unless you destroy |
165 |
the default loop first, which isn't supported. So in short: don't do |
166 |
it, and if you break it, you get to keep the pieces. |
167 |
|
168 |
BASIC INTERFACE |
169 |
$EV::DIED |
170 |
Must contain a reference to a function that is called when a |
171 |
callback throws an exception (with $@ containing the error). The |
172 |
default prints an informative message and continues. |
173 |
|
174 |
If this callback throws an exception it will be silently ignored. |
175 |
|
176 |
$flags = EV::supported_backends |
177 |
$flags = EV::recommended_backends |
178 |
$flags = EV::embeddable_backends |
179 |
Returns the set (see "EV::BACKEND_*" flags) of backends supported by |
180 |
this instance of EV, the set of recommended backends (supposed to be |
181 |
good) for this platform and the set of embeddable backends (see |
182 |
EMBED WATCHERS). |
183 |
|
184 |
EV::sleep $seconds |
185 |
Block the process for the given number of (fractional) seconds. |
186 |
|
187 |
$time = EV::time |
188 |
Returns the current time in (fractional) seconds since the epoch. |
189 |
|
190 |
$time = EV::now |
191 |
$time = $loop->now |
192 |
Returns the time the last event loop iteration has been started. |
193 |
This is the time that (relative) timers are based on, and referring |
194 |
to it is usually faster then calling EV::time. |
195 |
|
196 |
EV::now_update |
197 |
$loop->now_update |
198 |
Establishes the current time by querying the kernel, updating the |
199 |
time returned by "EV::now" in the progress. This is a costly |
200 |
operation and is usually done automatically within "EV::run". |
201 |
|
202 |
This function is rarely useful, but when some event callback runs |
203 |
for a very long time without entering the event loop, updating |
204 |
libev's idea of the current time is a good idea. |
205 |
|
206 |
EV::suspend |
207 |
$loop->suspend |
208 |
EV::resume |
209 |
$loop->resume |
210 |
These two functions suspend and resume a loop, for use when the loop |
211 |
is not used for a while and timeouts should not be processed. |
212 |
|
213 |
A typical use case would be an interactive program such as a game: |
214 |
When the user presses "^Z" to suspend the game and resumes it an |
215 |
hour later it would be best to handle timeouts as if no time had |
216 |
actually passed while the program was suspended. This can be |
217 |
achieved by calling "suspend" in your "SIGTSTP" handler, sending |
218 |
yourself a "SIGSTOP" and calling "resume" directly afterwards to |
219 |
resume timer processing. |
220 |
|
221 |
Effectively, all "timer" watchers will be delayed by the time spend |
222 |
between "suspend" and "resume", and all "periodic" watchers will be |
223 |
rescheduled (that is, they will lose any events that would have |
224 |
occured while suspended). |
225 |
|
226 |
After calling "suspend" you must not call *any* function on the |
227 |
given loop other than "resume", and you must not call "resume" |
228 |
without a previous call to "suspend". |
229 |
|
230 |
Calling "suspend"/"resume" has the side effect of updating the event |
231 |
loop time (see "now_update"). |
232 |
|
233 |
$backend = EV::backend |
234 |
$backend = $loop->backend |
235 |
Returns an integer describing the backend used by libev |
236 |
(EV::BACKEND_SELECT or EV::BACKEND_EPOLL). |
237 |
|
238 |
$active = EV::run [$flags] |
239 |
$active = $loop->run ([$flags]) |
240 |
Begin checking for events and calling callbacks. It returns when a |
241 |
callback calls EV::break or the flags are nonzero (in which case the |
242 |
return value is true) or when there are no active watchers which |
243 |
reference the loop (keepalive is true), in which case the return |
244 |
value will be false. The return value can generally be interpreted |
245 |
as "if true, there is more work left to do". |
246 |
|
247 |
The $flags argument can be one of the following: |
248 |
|
249 |
0 as above |
250 |
EV::RUN_ONCE block at most once (wait, but do not loop) |
251 |
EV::RUN_NOWAIT do not block at all (fetch/handle events but do not wait) |
252 |
|
253 |
EV::break [$how] |
254 |
$loop->break ([$how]) |
255 |
When called with no arguments or an argument of EV::BREAK_ONE, makes |
256 |
the innermost call to EV::run return. |
257 |
|
258 |
When called with an argument of EV::BREAK_ALL, all calls to EV::run |
259 |
will return as fast as possible. |
260 |
|
261 |
When called with an argument of EV::BREAK_CANCEL, any pending break |
262 |
will be cancelled. |
263 |
|
264 |
$count = EV::iteration |
265 |
$count = $loop->iteration |
266 |
Return the number of times the event loop has polled for new events. |
267 |
Sometimes useful as a generation counter. |
268 |
|
269 |
EV::once $fh_or_undef, $events, $timeout, $cb->($revents) |
270 |
$loop->once ($fh_or_undef, $events, $timeout, $cb->($revents)) |
271 |
This function rolls together an I/O and a timer watcher for a single |
272 |
one-shot event without the need for managing a watcher object. |
273 |
|
274 |
If $fh_or_undef is a filehandle or file descriptor, then $events |
275 |
must be a bitset containing either "EV::READ", "EV::WRITE" or |
276 |
"EV::READ | EV::WRITE", indicating the type of I/O event you want to |
277 |
wait for. If you do not want to wait for some I/O event, specify |
278 |
"undef" for $fh_or_undef and 0 for $events). |
279 |
|
280 |
If timeout is "undef" or negative, then there will be no timeout. |
281 |
Otherwise an "EV::timer" with this value will be started. |
282 |
|
283 |
When an error occurs or either the timeout or I/O watcher triggers, |
284 |
then the callback will be called with the received event set (in |
285 |
general you can expect it to be a combination of "EV::ERROR", |
286 |
"EV::READ", "EV::WRITE" and "EV::TIMER"). |
287 |
|
288 |
EV::once doesn't return anything: the watchers stay active till |
289 |
either of them triggers, then they will be stopped and freed, and |
290 |
the callback invoked. |
291 |
|
292 |
EV::feed_fd_event $fd, $revents |
293 |
$loop->feed_fd_event ($fd, $revents) |
294 |
Feed an event on a file descriptor into EV. EV will react to this |
295 |
call as if the readyness notifications specified by $revents (a |
296 |
combination of "EV::READ" and "EV::WRITE") happened on the file |
297 |
descriptor $fd. |
298 |
|
299 |
EV::feed_signal_event $signal |
300 |
Feed a signal event into the default loop. EV will react to this |
301 |
call as if the signal specified by $signal had occured. |
302 |
|
303 |
EV::feed_signal $signal |
304 |
Feed a signal event into EV - unlike "EV::feed_signal_event", this |
305 |
works regardless of which loop has registered the signal, and is |
306 |
mainly useful fro custom signal implementations. |
307 |
|
308 |
EV::set_io_collect_interval $time |
309 |
$loop->set_io_collect_interval ($time) |
310 |
EV::set_timeout_collect_interval $time |
311 |
$loop->set_timeout_collect_interval ($time) |
312 |
These advanced functions set the minimum block interval when polling |
313 |
for I/O events and the minimum wait interval for timer events. See |
314 |
the libev documentation at |
315 |
<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#FUNCTIONS_CONT |
316 |
ROLLING_THE_EVENT_LOOP> (locally installed as EV::libev) for a more |
317 |
detailed discussion. |
318 |
|
319 |
$count = EV::pending_count |
320 |
$count = $loop->pending_count |
321 |
Returns the number of currently pending watchers. |
322 |
|
323 |
EV::invoke_pending |
324 |
$loop->invoke_pending |
325 |
Invoke all currently pending watchers. |
326 |
|
327 |
WATCHER OBJECTS |
328 |
A watcher is an object that gets created to record your interest in some |
329 |
event. For instance, if you want to wait for STDIN to become readable, |
330 |
you would create an EV::io watcher for that: |
331 |
|
332 |
my $watcher = EV::io *STDIN, EV::READ, sub { |
333 |
my ($watcher, $revents) = @_; |
334 |
warn "yeah, STDIN should now be readable without blocking!\n" |
335 |
}; |
336 |
|
337 |
All watchers can be active (waiting for events) or inactive (paused). |
338 |
Only active watchers will have their callbacks invoked. All callbacks |
339 |
will be called with at least two arguments: the watcher and a bitmask of |
340 |
received events. |
341 |
|
342 |
Each watcher type has its associated bit in revents, so you can use the |
343 |
same callback for multiple watchers. The event mask is named after the |
344 |
type, i.e. EV::child sets EV::CHILD, EV::prepare sets EV::PREPARE, |
345 |
EV::periodic sets EV::PERIODIC and so on, with the exception of I/O |
346 |
events (which can set both EV::READ and EV::WRITE bits). |
347 |
|
348 |
In the rare case where one wants to create a watcher but not start it at |
349 |
the same time, each constructor has a variant with a trailing "_ns" in |
350 |
its name, e.g. EV::io has a non-starting variant EV::io_ns and so on. |
351 |
|
352 |
Please note that a watcher will automatically be stopped when the |
353 |
watcher object is destroyed, so you *need* to keep the watcher objects |
354 |
returned by the constructors. |
355 |
|
356 |
Also, all methods changing some aspect of a watcher (->set, ->priority, |
357 |
->fh and so on) automatically stop and start it again if it is active, |
358 |
which means pending events get lost. |
359 |
|
360 |
COMMON WATCHER METHODS |
361 |
This section lists methods common to all watchers. |
362 |
|
363 |
$w->start |
364 |
Starts a watcher if it isn't active already. Does nothing to an |
365 |
already active watcher. By default, all watchers start out in the |
366 |
active state (see the description of the "_ns" variants if you need |
367 |
stopped watchers). |
368 |
|
369 |
$w->stop |
370 |
Stop a watcher if it is active. Also clear any pending events |
371 |
(events that have been received but that didn't yet result in a |
372 |
callback invocation), regardless of whether the watcher was active |
373 |
or not. |
374 |
|
375 |
$bool = $w->is_active |
376 |
Returns true if the watcher is active, false otherwise. |
377 |
|
378 |
$current_data = $w->data |
379 |
$old_data = $w->data ($new_data) |
380 |
Queries a freely usable data scalar on the watcher and optionally |
381 |
changes it. This is a way to associate custom data with a watcher: |
382 |
|
383 |
my $w = EV::timer 60, 0, sub { |
384 |
warn $_[0]->data; |
385 |
}; |
386 |
$w->data ("print me!"); |
387 |
|
388 |
$current_cb = $w->cb |
389 |
$old_cb = $w->cb ($new_cb) |
390 |
Queries the callback on the watcher and optionally changes it. You |
391 |
can do this at any time without the watcher restarting. |
392 |
|
393 |
$current_priority = $w->priority |
394 |
$old_priority = $w->priority ($new_priority) |
395 |
Queries the priority on the watcher and optionally changes it. |
396 |
Pending watchers with higher priority will be invoked first. The |
397 |
valid range of priorities lies between EV::MAXPRI (default 2) and |
398 |
EV::MINPRI (default -2). If the priority is outside this range it |
399 |
will automatically be normalised to the nearest valid priority. |
400 |
|
401 |
The default priority of any newly-created watcher is 0. |
402 |
|
403 |
Note that the priority semantics have not yet been fleshed out and |
404 |
are subject to almost certain change. |
405 |
|
406 |
$w->invoke ($revents) |
407 |
Call the callback *now* with the given event mask. |
408 |
|
409 |
$w->feed_event ($revents) |
410 |
Feed some events on this watcher into EV. EV will react to this call |
411 |
as if the watcher had received the given $revents mask. |
412 |
|
413 |
$revents = $w->clear_pending |
414 |
If the watcher is pending, this function clears its pending status |
415 |
and returns its $revents bitset (as if its callback was invoked). If |
416 |
the watcher isn't pending it does nothing and returns 0. |
417 |
|
418 |
$previous_state = $w->keepalive ($bool) |
419 |
Normally, "EV::run" will return when there are no active watchers |
420 |
(which is a "deadlock" because no progress can be made anymore). |
421 |
This is convenient because it allows you to start your watchers (and |
422 |
your jobs), call "EV::run" once and when it returns you know that |
423 |
all your jobs are finished (or they forgot to register some watchers |
424 |
for their task :). |
425 |
|
426 |
Sometimes, however, this gets in your way, for example when the |
427 |
module that calls "EV::run" (usually the main program) is not the |
428 |
same module as a long-living watcher (for example a DNS client |
429 |
module written by somebody else even). Then you might want any |
430 |
outstanding requests to be handled, but you would not want to keep |
431 |
"EV::run" from returning just because you happen to have this |
432 |
long-running UDP port watcher. |
433 |
|
434 |
In this case you can clear the keepalive status, which means that |
435 |
even though your watcher is active, it won't keep "EV::run" from |
436 |
returning. |
437 |
|
438 |
The initial value for keepalive is true (enabled), and you can |
439 |
change it any time. |
440 |
|
441 |
Example: Register an I/O watcher for some UDP socket but do not keep |
442 |
the event loop from running just because of that watcher. |
443 |
|
444 |
my $udp_socket = ... |
445 |
my $udp_watcher = EV::io $udp_socket, EV::READ, sub { ... }; |
446 |
$udp_watcher->keepalive (0); |
447 |
|
448 |
$loop = $w->loop |
449 |
Return the loop that this watcher is attached to. |
450 |
|
451 |
WATCHER TYPES |
452 |
Each of the following subsections describes a single watcher type. |
453 |
|
454 |
I/O WATCHERS - is this file descriptor readable or writable? |
455 |
$w = EV::io $fileno_or_fh, $eventmask, $callback |
456 |
$w = EV::io_ns $fileno_or_fh, $eventmask, $callback |
457 |
$w = $loop->io ($fileno_or_fh, $eventmask, $callback) |
458 |
$w = $loop->io_ns ($fileno_or_fh, $eventmask, $callback) |
459 |
As long as the returned watcher object is alive, call the $callback |
460 |
when at least one of events specified in $eventmask occurs. |
461 |
|
462 |
The $eventmask can be one or more of these constants ORed together: |
463 |
|
464 |
EV::READ wait until read() wouldn't block anymore |
465 |
EV::WRITE wait until write() wouldn't block anymore |
466 |
|
467 |
The "io_ns" variant doesn't start (activate) the newly created |
468 |
watcher. |
469 |
|
470 |
$w->set ($fileno_or_fh, $eventmask) |
471 |
Reconfigures the watcher, see the constructor above for details. Can |
472 |
be called at any time. |
473 |
|
474 |
$current_fh = $w->fh |
475 |
$old_fh = $w->fh ($new_fh) |
476 |
Returns the previously set filehandle and optionally set a new one. |
477 |
|
478 |
$current_eventmask = $w->events |
479 |
$old_eventmask = $w->events ($new_eventmask) |
480 |
Returns the previously set event mask and optionally set a new one. |
481 |
|
482 |
TIMER WATCHERS - relative and optionally repeating timeouts |
483 |
$w = EV::timer $after, $repeat, $callback |
484 |
$w = EV::timer_ns $after, $repeat, $callback |
485 |
$w = $loop->timer ($after, $repeat, $callback) |
486 |
$w = $loop->timer_ns ($after, $repeat, $callback) |
487 |
Calls the callback after $after seconds (which may be fractional or |
488 |
negative). If $repeat is non-zero, the timer will be restarted (with |
489 |
the $repeat value as $after) after the callback returns. |
490 |
|
491 |
This means that the callback would be called roughly after $after |
492 |
seconds, and then every $repeat seconds. The timer does his best not |
493 |
to drift, but it will not invoke the timer more often then once per |
494 |
event loop iteration, and might drift in other cases. If that isn't |
495 |
acceptable, look at EV::periodic, which can provide long-term stable |
496 |
timers. |
497 |
|
498 |
The timer is based on a monotonic clock, that is, if somebody is |
499 |
sitting in front of the machine while the timer is running and |
500 |
changes the system clock, the timer will nevertheless run (roughly) |
501 |
the same time. |
502 |
|
503 |
The "timer_ns" variant doesn't start (activate) the newly created |
504 |
watcher. |
505 |
|
506 |
$w->set ($after, $repeat = 0) |
507 |
Reconfigures the watcher, see the constructor above for details. Can |
508 |
be called at any time. |
509 |
|
510 |
$w->again |
511 |
$w->again ($repeat) |
512 |
Similar to the "start" method, but has special semantics for |
513 |
repeating timers: |
514 |
|
515 |
If the timer is active and non-repeating, it will be stopped. |
516 |
|
517 |
If the timer is active and repeating, reset the timeout to occur |
518 |
$repeat seconds after now. |
519 |
|
520 |
If the timer is inactive and repeating, start it using the repeat |
521 |
value. |
522 |
|
523 |
Otherwise do nothing. |
524 |
|
525 |
This behaviour is useful when you have a timeout for some IO |
526 |
operation. You create a timer object with the same value for $after |
527 |
and $repeat, and then, in the read/write watcher, run the "again" |
528 |
method on the timeout. |
529 |
|
530 |
If called with a $repeat argument, then it uses this a timer repeat |
531 |
value. |
532 |
|
533 |
$after = $w->remaining |
534 |
Calculates and returns the remaining time till the timer will fire. |
535 |
|
536 |
PERIODIC WATCHERS - to cron or not to cron? |
537 |
$w = EV::periodic $at, $interval, $reschedule_cb, $callback |
538 |
$w = EV::periodic_ns $at, $interval, $reschedule_cb, $callback |
539 |
$w = $loop->periodic ($at, $interval, $reschedule_cb, $callback) |
540 |
$w = $loop->periodic_ns ($at, $interval, $reschedule_cb, $callback) |
541 |
Similar to EV::timer, but is not based on relative timeouts but on |
542 |
absolute times. Apart from creating "simple" timers that trigger |
543 |
"at" the specified time, it can also be used for non-drifting |
544 |
absolute timers and more complex, cron-like, setups that are not |
545 |
adversely affected by time jumps (i.e. when the system clock is |
546 |
changed by explicit date -s or other means such as ntpd). It is also |
547 |
the most complex watcher type in EV. |
548 |
|
549 |
It has three distinct "modes": |
550 |
|
551 |
* absolute timer ($interval = $reschedule_cb = 0) |
552 |
|
553 |
This time simply fires at the wallclock time $at and doesn't |
554 |
repeat. It will not adjust when a time jump occurs, that is, if |
555 |
it is to be run at January 1st 2011 then it will run when the |
556 |
system time reaches or surpasses this time. |
557 |
|
558 |
* repeating interval timer ($interval > 0, $reschedule_cb = 0) |
559 |
|
560 |
In this mode the watcher will always be scheduled to time out at |
561 |
the next "$at + N * $interval" time (for the lowest integer N) |
562 |
and then repeat, regardless of any time jumps. Note that, since |
563 |
"N" can be negative, the first trigger can happen before $at. |
564 |
|
565 |
This can be used to create timers that do not drift with respect |
566 |
to system time: |
567 |
|
568 |
my $hourly = EV::periodic 0, 3600, 0, sub { print "once/hour\n" }; |
569 |
|
570 |
That doesn't mean there will always be 3600 seconds in between |
571 |
triggers, but only that the the callback will be called when the |
572 |
system time shows a full hour (UTC). |
573 |
|
574 |
Another way to think about it (for the mathematically inclined) |
575 |
is that EV::periodic will try to run the callback in this mode |
576 |
at the next possible time where "$time = $at (mod $interval)", |
577 |
regardless of any time jumps. |
578 |
|
579 |
* manual reschedule mode ($reschedule_cb = coderef) |
580 |
|
581 |
In this mode $interval and $at are both being ignored. Instead, |
582 |
each time the periodic watcher gets scheduled, the reschedule |
583 |
callback ($reschedule_cb) will be called with the watcher as |
584 |
first, and the current time as second argument. |
585 |
|
586 |
*This callback MUST NOT stop or destroy this or any other |
587 |
periodic watcher, ever, and MUST NOT call any event loop |
588 |
functions or methods*. If you need to stop it, return 1e30 and |
589 |
stop it afterwards. You may create and start an "EV::prepare" |
590 |
watcher for this task. |
591 |
|
592 |
It must return the next time to trigger, based on the passed |
593 |
time value (that is, the lowest time value larger than or equal |
594 |
to to the second argument). It will usually be called just |
595 |
before the callback will be triggered, but might be called at |
596 |
other times, too. |
597 |
|
598 |
This can be used to create very complex timers, such as a timer |
599 |
that triggers on each midnight, local time (actually one day |
600 |
after the last midnight, to keep the example simple): |
601 |
|
602 |
my $daily = EV::periodic 0, 0, sub { |
603 |
my ($w, $now) = @_; |
604 |
|
605 |
use Time::Local (); |
606 |
my (undef, undef, undef, $d, $m, $y) = localtime $now; |
607 |
Time::Local::timelocal_nocheck 0, 0, 0, $d + 1, $m, $y |
608 |
}, sub { |
609 |
print "it's midnight or likely shortly after, now\n"; |
610 |
}; |
611 |
|
612 |
The "periodic_ns" variant doesn't start (activate) the newly created |
613 |
watcher. |
614 |
|
615 |
$w->set ($at, $interval, $reschedule_cb) |
616 |
Reconfigures the watcher, see the constructor above for details. Can |
617 |
be called at any time. |
618 |
|
619 |
$w->again |
620 |
Simply stops and starts the watcher again. |
621 |
|
622 |
$time = $w->at |
623 |
Return the time that the watcher is expected to trigger next. |
624 |
|
625 |
SIGNAL WATCHERS - signal me when a signal gets signalled! |
626 |
$w = EV::signal $signal, $callback |
627 |
$w = EV::signal_ns $signal, $callback |
628 |
$w = $loop->signal ($signal, $callback) |
629 |
$w = $loop->signal_ns ($signal, $callback) |
630 |
Call the callback when $signal is received (the signal can be |
631 |
specified by number or by name, just as with "kill" or %SIG). |
632 |
|
633 |
Only one event loop can grab a given signal - attempting to grab the |
634 |
same signal from two EV loops will crash the program immediately or |
635 |
cause data corruption. |
636 |
|
637 |
EV will grab the signal for the process (the kernel only allows one |
638 |
component to receive a signal at a time) when you start a signal |
639 |
watcher, and removes it again when you stop it. Perl does the same |
640 |
when you add/remove callbacks to %SIG, so watch out. |
641 |
|
642 |
You can have as many signal watchers per signal as you want. |
643 |
|
644 |
The "signal_ns" variant doesn't start (activate) the newly created |
645 |
watcher. |
646 |
|
647 |
$w->set ($signal) |
648 |
Reconfigures the watcher, see the constructor above for details. Can |
649 |
be called at any time. |
650 |
|
651 |
$current_signum = $w->signal |
652 |
$old_signum = $w->signal ($new_signal) |
653 |
Returns the previously set signal (always as a number not name) and |
654 |
optionally set a new one. |
655 |
|
656 |
CHILD WATCHERS - watch out for process status changes |
657 |
$w = EV::child $pid, $trace, $callback |
658 |
$w = EV::child_ns $pid, $trace, $callback |
659 |
$w = $loop->child ($pid, $trace, $callback) |
660 |
$w = $loop->child_ns ($pid, $trace, $callback) |
661 |
Call the callback when a status change for pid $pid (or any pid if |
662 |
$pid is 0) has been received (a status change happens when the |
663 |
process terminates or is killed, or, when trace is true, |
664 |
additionally when it is stopped or continued). More precisely: when |
665 |
the process receives a "SIGCHLD", EV will fetch the outstanding |
666 |
exit/wait status for all changed/zombie children and call the |
667 |
callback. |
668 |
|
669 |
It is valid (and fully supported) to install a child watcher after a |
670 |
child has exited but before the event loop has started its next |
671 |
iteration (for example, first you "fork", then the new child process |
672 |
might exit, and only then do you install a child watcher in the |
673 |
parent for the new pid). |
674 |
|
675 |
You can access both exit (or tracing) status and pid by using the |
676 |
"rstatus" and "rpid" methods on the watcher object. |
677 |
|
678 |
You can have as many pid watchers per pid as you want, they will all |
679 |
be called. |
680 |
|
681 |
The "child_ns" variant doesn't start (activate) the newly created |
682 |
watcher. |
683 |
|
684 |
$w->set ($pid, $trace) |
685 |
Reconfigures the watcher, see the constructor above for details. Can |
686 |
be called at any time. |
687 |
|
688 |
$current_pid = $w->pid |
689 |
Returns the previously set process id and optionally set a new one. |
690 |
|
691 |
$exit_status = $w->rstatus |
692 |
Return the exit/wait status (as returned by waitpid, see the waitpid |
693 |
entry in perlfunc). |
694 |
|
695 |
$pid = $w->rpid |
696 |
Return the pid of the awaited child (useful when you have installed |
697 |
a watcher for all pids). |
698 |
|
699 |
STAT WATCHERS - did the file attributes just change? |
700 |
$w = EV::stat $path, $interval, $callback |
701 |
$w = EV::stat_ns $path, $interval, $callback |
702 |
$w = $loop->stat ($path, $interval, $callback) |
703 |
$w = $loop->stat_ns ($path, $interval, $callback) |
704 |
Call the callback when a file status change has been detected on |
705 |
$path. The $path does not need to exist, changing from "path exists" |
706 |
to "path does not exist" is a status change like any other. |
707 |
|
708 |
The $interval is a recommended polling interval for systems where |
709 |
OS-supported change notifications don't exist or are not supported. |
710 |
If you use 0 then an unspecified default is used (which is highly |
711 |
recommended!), which is to be expected to be around five seconds |
712 |
usually. |
713 |
|
714 |
This watcher type is not meant for massive numbers of stat watchers, |
715 |
as even with OS-supported change notifications, this can be |
716 |
resource-intensive. |
717 |
|
718 |
The "stat_ns" variant doesn't start (activate) the newly created |
719 |
watcher. |
720 |
|
721 |
... = $w->stat |
722 |
This call is very similar to the perl "stat" built-in: It stats |
723 |
(using "lstat") the path specified in the watcher and sets perls |
724 |
stat cache (as well as EV's idea of the current stat values) to the |
725 |
values found. |
726 |
|
727 |
In scalar context, a boolean is return indicating success or failure |
728 |
of the stat. In list context, the same 13-value list as with stat is |
729 |
returned (except that the blksize and blocks fields are not |
730 |
reliable). |
731 |
|
732 |
In the case of an error, errno is set to "ENOENT" (regardless of the |
733 |
actual error value) and the "nlink" value is forced to zero (if the |
734 |
stat was successful then nlink is guaranteed to be non-zero). |
735 |
|
736 |
See also the next two entries for more info. |
737 |
|
738 |
... = $w->attr |
739 |
Just like "$w->stat", but without the initial stat'ing: this returns |
740 |
the values most recently detected by EV. See the next entry for more |
741 |
info. |
742 |
|
743 |
... = $w->prev |
744 |
Just like "$w->stat", but without the initial stat'ing: this returns |
745 |
the previous set of values, before the change. |
746 |
|
747 |
That is, when the watcher callback is invoked, "$w->prev" will be |
748 |
set to the values found *before* a change was detected, while |
749 |
"$w->attr" returns the values found leading to the change detection. |
750 |
The difference (if any) between "prev" and "attr" is what triggered |
751 |
the callback. |
752 |
|
753 |
If you did something to the filesystem object and do not want to |
754 |
trigger yet another change, you can call "stat" to update EV's idea |
755 |
of what the current attributes are. |
756 |
|
757 |
$w->set ($path, $interval) |
758 |
Reconfigures the watcher, see the constructor above for details. Can |
759 |
be called at any time. |
760 |
|
761 |
$current_path = $w->path |
762 |
$old_path = $w->path ($new_path) |
763 |
Returns the previously set path and optionally set a new one. |
764 |
|
765 |
$current_interval = $w->interval |
766 |
$old_interval = $w->interval ($new_interval) |
767 |
Returns the previously set interval and optionally set a new one. |
768 |
Can be used to query the actual interval used. |
769 |
|
770 |
IDLE WATCHERS - when you've got nothing better to do... |
771 |
$w = EV::idle $callback |
772 |
$w = EV::idle_ns $callback |
773 |
$w = $loop->idle ($callback) |
774 |
$w = $loop->idle_ns ($callback) |
775 |
Call the callback when there are no other pending watchers of the |
776 |
same or higher priority (excluding check, prepare and other idle |
777 |
watchers of the same or lower priority, of course). They are called |
778 |
idle watchers because when the watcher is the highest priority |
779 |
pending event in the process, the process is considered to be idle |
780 |
at that priority. |
781 |
|
782 |
If you want a watcher that is only ever called when *no* other |
783 |
events are outstanding you have to set the priority to "EV::MINPRI". |
784 |
|
785 |
The process will not block as long as any idle watchers are active, |
786 |
and they will be called repeatedly until stopped. |
787 |
|
788 |
For example, if you have idle watchers at priority 0 and 1, and an |
789 |
I/O watcher at priority 0, then the idle watcher at priority 1 and |
790 |
the I/O watcher will always run when ready. Only when the idle |
791 |
watcher at priority 1 is stopped and the I/O watcher at priority 0 |
792 |
is not pending with the 0-priority idle watcher be invoked. |
793 |
|
794 |
The "idle_ns" variant doesn't start (activate) the newly created |
795 |
watcher. |
796 |
|
797 |
PREPARE WATCHERS - customise your event loop! |
798 |
$w = EV::prepare $callback |
799 |
$w = EV::prepare_ns $callback |
800 |
$w = $loop->prepare ($callback) |
801 |
$w = $loop->prepare_ns ($callback) |
802 |
Call the callback just before the process would block. You can still |
803 |
create/modify any watchers at this point. |
804 |
|
805 |
See the EV::check watcher, below, for explanations and an example. |
806 |
|
807 |
The "prepare_ns" variant doesn't start (activate) the newly created |
808 |
watcher. |
809 |
|
810 |
CHECK WATCHERS - customise your event loop even more! |
811 |
$w = EV::check $callback |
812 |
$w = EV::check_ns $callback |
813 |
$w = $loop->check ($callback) |
814 |
$w = $loop->check_ns ($callback) |
815 |
Call the callback just after the process wakes up again (after it |
816 |
has gathered events), but before any other callbacks have been |
817 |
invoked. |
818 |
|
819 |
This can be used to integrate other event-based software into the EV |
820 |
mainloop: You register a prepare callback and in there, you create |
821 |
io and timer watchers as required by the other software. Here is a |
822 |
real-world example of integrating Net::SNMP (with some details left |
823 |
out): |
824 |
|
825 |
our @snmp_watcher; |
826 |
|
827 |
our $snmp_prepare = EV::prepare sub { |
828 |
# do nothing unless active |
829 |
$dispatcher->{_event_queue_h} |
830 |
or return; |
831 |
|
832 |
# make the dispatcher handle any outstanding stuff |
833 |
... not shown |
834 |
|
835 |
# create an I/O watcher for each and every socket |
836 |
@snmp_watcher = ( |
837 |
(map { EV::io $_, EV::READ, sub { } } |
838 |
keys %{ $dispatcher->{_descriptors} }), |
839 |
|
840 |
EV::timer +($event->[Net::SNMP::Dispatcher::_ACTIVE] |
841 |
? $event->[Net::SNMP::Dispatcher::_TIME] - EV::now : 0), |
842 |
0, sub { }, |
843 |
); |
844 |
}; |
845 |
|
846 |
The callbacks are irrelevant (and are not even being called), the |
847 |
only purpose of those watchers is to wake up the process as soon as |
848 |
one of those events occurs (socket readable, or timer timed out). |
849 |
The corresponding EV::check watcher will then clean up: |
850 |
|
851 |
our $snmp_check = EV::check sub { |
852 |
# destroy all watchers |
853 |
@snmp_watcher = (); |
854 |
|
855 |
# make the dispatcher handle any new stuff |
856 |
... not shown |
857 |
}; |
858 |
|
859 |
The callbacks of the created watchers will not be called as the |
860 |
watchers are destroyed before this can happen (remember EV::check |
861 |
gets called first). |
862 |
|
863 |
The "check_ns" variant doesn't start (activate) the newly created |
864 |
watcher. |
865 |
|
866 |
EV::CHECK constant issues |
867 |
Like all other watcher types, there is a bitmask constant for use in |
868 |
$revents and other places. The "EV::CHECK" is special as it has the |
869 |
same name as the "CHECK" sub called by Perl. This doesn't cause big |
870 |
issues on newer perls (beginning with 5.8.9), but it means thatthe |
871 |
constant must be *inlined*, i.e. runtime calls will not work. That |
872 |
means that as long as you always "use EV" and then "EV::CHECK" you |
873 |
are on the safe side. |
874 |
|
875 |
FORK WATCHERS - the audacity to resume the event loop after a fork |
876 |
Fork watchers are called when a "fork ()" was detected. The invocation |
877 |
is done before the event loop blocks next and before "check" watchers |
878 |
are being called, and only in the child after the fork. |
879 |
|
880 |
$w = EV::fork $callback |
881 |
$w = EV::fork_ns $callback |
882 |
$w = $loop->fork ($callback) |
883 |
$w = $loop->fork_ns ($callback) |
884 |
Call the callback before the event loop is resumed in the child |
885 |
process after a fork. |
886 |
|
887 |
The "fork_ns" variant doesn't start (activate) the newly created |
888 |
watcher. |
889 |
|
890 |
EMBED WATCHERS - when one backend isn't enough... |
891 |
This is a rather advanced watcher type that lets you embed one event |
892 |
loop into another (currently only IO events are supported in the |
893 |
embedded loop, other types of watchers might be handled in a delayed or |
894 |
incorrect fashion and must not be used). |
895 |
|
896 |
See the libev documentation at |
897 |
<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#code_ev_embed_code |
898 |
_when_one_backend_> (locally installed as EV::libev) for more details. |
899 |
|
900 |
In short, this watcher is most useful on BSD systems without working |
901 |
kqueue to still be able to handle a large number of sockets: |
902 |
|
903 |
my $socket_loop; |
904 |
|
905 |
# check wether we use SELECT or POLL _and_ KQUEUE is supported |
906 |
if ( |
907 |
(EV::backend & (EV::BACKEND_POLL | EV::BACKEND_SELECT)) |
908 |
&& (EV::supported_backends & EV::embeddable_backends & EV::BACKEND_KQUEUE) |
909 |
) { |
910 |
# use kqueue for sockets |
911 |
$socket_loop = new EV::Loop EV::BACKEND_KQUEUE | EV::FLAG_NOENV; |
912 |
} |
913 |
|
914 |
# use the default loop otherwise |
915 |
$socket_loop ||= EV::default_loop; |
916 |
|
917 |
$w = EV::embed $otherloop[, $callback] |
918 |
$w = EV::embed_ns $otherloop[, $callback] |
919 |
$w = $loop->embed ($otherloop[, $callback]) |
920 |
$w = $loop->embed_ns ($otherloop[, $callback]) |
921 |
Call the callback when the embedded event loop ($otherloop) has any |
922 |
I/O activity. The $callback is optional: if it is missing, then the |
923 |
embedded event loop will be managed automatically (which is |
924 |
recommended), otherwise you have to invoke "sweep" yourself. |
925 |
|
926 |
The "embed_ns" variant doesn't start (activate) the newly created |
927 |
watcher. |
928 |
|
929 |
ASYNC WATCHERS - how to wake up another event loop |
930 |
Async watchers are provided by EV, but have little use in perl directly, |
931 |
as perl neither supports threads running in parallel nor direct access |
932 |
to signal handlers or other contexts where they could be of value. |
933 |
|
934 |
It is, however, possible to use them from the XS level. |
935 |
|
936 |
Please see the libev documentation for further details. |
937 |
|
938 |
$w = EV::async $callback |
939 |
$w = EV::async_ns $callback |
940 |
$w = $loop->async ($callback) |
941 |
$w = $loop->async_ns ($callback) |
942 |
$w->send |
943 |
$bool = $w->async_pending |
944 |
|
945 |
CLEANUP WATCHERS - how to clean up when the event loop goes away |
946 |
Cleanup watchers are not supported on the Perl level, they can only be |
947 |
used via XS currently. |
948 |
|
949 |
PERL SIGNALS |
950 |
While Perl signal handling (%SIG) is not affected by EV, the behaviour |
951 |
with EV is as the same as any other C library: Perl-signals will only be |
952 |
handled when Perl runs, which means your signal handler might be invoked |
953 |
only the next time an event callback is invoked. |
954 |
|
955 |
The solution is to use EV signal watchers (see "EV::signal"), which will |
956 |
ensure proper operations with regards to other event watchers. |
957 |
|
958 |
If you cannot do this for whatever reason, you can also force a watcher |
959 |
to be called on every event loop iteration by installing a "EV::check" |
960 |
watcher: |
961 |
|
962 |
my $async_check = EV::check sub { }; |
963 |
|
964 |
This ensures that perl gets into control for a short time to handle any |
965 |
pending signals, and also ensures (slightly) slower overall operation. |
966 |
|
967 |
ITHREADS |
968 |
Ithreads are not supported by this module in any way. Perl |
969 |
pseudo-threads is evil stuff and must die. Real threads as provided by |
970 |
Coro are fully supported (and enhanced support is available via |
971 |
Coro::EV). |
972 |
|
973 |
FORK |
974 |
Most of the "improved" event delivering mechanisms of modern operating |
975 |
systems have quite a few problems with fork(2) (to put it bluntly: it is |
976 |
not supported and usually destructive). Libev makes it possible to work |
977 |
around this by having a function that recreates the kernel state after |
978 |
fork in the child. |
979 |
|
980 |
On non-win32 platforms, this module requires the pthread_atfork |
981 |
functionality to do this automatically for you. This function is quite |
982 |
buggy on most BSDs, though, so YMMV. The overhead for this is quite |
983 |
negligible, because everything the function currently does is set a flag |
984 |
that is checked only when the event loop gets used the next time, so |
985 |
when you do fork but not use EV, the overhead is minimal. |
986 |
|
987 |
On win32, there is no notion of fork so all this doesn't apply, of |
988 |
course. |
989 |
|
990 |
SEE ALSO |
991 |
EV::MakeMaker - MakeMaker interface to XS API, EV::ADNS (asynchronous |
992 |
DNS), Glib::EV (makes Glib/Gtk2 use EV as event loop), EV::Glib (embed |
993 |
Glib into EV), Coro::EV (efficient thread integration), Net::SNMP::EV |
994 |
(asynchronous SNMP), AnyEvent for event-loop agnostic and portable event |
995 |
driven programming. |
996 |
|
997 |
AUTHOR |
998 |
Marc Lehmann <schmorp@schmorp.de> |
999 |
http://home.schmorp.de/ |
1000 |
|