1 |
=head1 NAME |
2 |
|
3 |
EV - perl interface to libev, a high performance full-featured event loop |
4 |
|
5 |
=head1 SYNOPSIS |
6 |
|
7 |
use EV; |
8 |
|
9 |
# TIMERS |
10 |
|
11 |
my $w = EV::timer 2, 0, sub { |
12 |
warn "is called after 2s"; |
13 |
}; |
14 |
|
15 |
my $w = EV::timer 2, 2, sub { |
16 |
warn "is called roughly every 2s (repeat = 2)"; |
17 |
}; |
18 |
|
19 |
undef $w; # destroy event watcher again |
20 |
|
21 |
my $w = EV::periodic 0, 60, 0, sub { |
22 |
warn "is called every minute, on the minute, exactly"; |
23 |
}; |
24 |
|
25 |
# IO |
26 |
|
27 |
my $w = EV::io *STDIN, EV::READ, sub { |
28 |
my ($w, $revents) = @_; # all callbacks receive the watcher and event mask |
29 |
warn "stdin is readable, you entered: ", <STDIN>; |
30 |
}; |
31 |
|
32 |
# SIGNALS |
33 |
|
34 |
my $w = EV::signal 'QUIT', sub { |
35 |
warn "sigquit received\n"; |
36 |
}; |
37 |
|
38 |
# CHILD/PID STATUS CHANGES |
39 |
|
40 |
my $w = EV::child 666, 0, sub { |
41 |
my ($w, $revents) = @_; |
42 |
my $status = $w->rstatus; |
43 |
}; |
44 |
|
45 |
# STAT CHANGES |
46 |
my $w = EV::stat "/etc/passwd", 10, sub { |
47 |
my ($w, $revents) = @_; |
48 |
warn $w->path, " has changed somehow.\n"; |
49 |
}; |
50 |
|
51 |
# MAINLOOP |
52 |
EV::loop; # loop until EV::unloop is called or all watchers stop |
53 |
EV::loop EV::LOOP_ONESHOT; # block until at least one event could be handled |
54 |
EV::loop EV::LOOP_NONBLOCK; # try to handle same events, but do not block |
55 |
|
56 |
=head1 DESCRIPTION |
57 |
|
58 |
This module provides an interface to libev |
59 |
(L<http://software.schmorp.de/pkg/libev.html>). While the documentation |
60 |
below is comprehensive, one might also consult the documentation of |
61 |
libev itself (L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod> or |
62 |
F<perldoc EV::libev>) for more subtle details on watcher semantics or some |
63 |
discussion on the available backends, or how to force a specific backend |
64 |
with C<LIBEV_FLAGS>, or just about in any case because it has much more |
65 |
detailed information. |
66 |
|
67 |
This module is very fast and scalable. It is actually so fast that you |
68 |
can use it through the L<AnyEvent> module, stay portable to other event |
69 |
loops (if you don't rely on any watcher types not available through it) |
70 |
and still be faster than with any other event loop currently supported in |
71 |
Perl. |
72 |
|
73 |
=head2 MODULE EXPORTS |
74 |
|
75 |
This module does not export any symbols. |
76 |
|
77 |
=cut |
78 |
|
79 |
package EV; |
80 |
|
81 |
no warnings; |
82 |
use strict; |
83 |
|
84 |
BEGIN { |
85 |
our $VERSION = '3.6'; |
86 |
use XSLoader; |
87 |
XSLoader::load "EV", $VERSION; |
88 |
} |
89 |
|
90 |
@EV::IO::ISA = |
91 |
@EV::Timer::ISA = |
92 |
@EV::Periodic::ISA = |
93 |
@EV::Signal::ISA = |
94 |
@EV::Child::ISA = |
95 |
@EV::Stat::ISA = |
96 |
@EV::Idle::ISA = |
97 |
@EV::Prepare::ISA = |
98 |
@EV::Check::ISA = |
99 |
@EV::Embed::ISA = |
100 |
@EV::Fork::ISA = |
101 |
@EV::Async::ISA = |
102 |
"EV::Watcher"; |
103 |
|
104 |
@EV::Loop::Default::ISA = "EV::Loop"; |
105 |
|
106 |
=head1 EVENT LOOPS |
107 |
|
108 |
EV supports multiple event loops: There is a single "default event loop" |
109 |
that can handle everything including signals and child watchers, and any |
110 |
number of "dynamic event loops" that can use different backends (with |
111 |
various limitations), but no child and signal watchers. |
112 |
|
113 |
You do not have to do anything to create the default event loop: When |
114 |
the module is loaded a suitable backend is selected on the premise of |
115 |
selecting a working backend (which for example rules out kqueue on most |
116 |
BSDs). Modules should, unless they have "special needs" always use the |
117 |
default loop as this is fastest (perl-wise), best supported by other |
118 |
modules (e.g. AnyEvent or Coro) and most portable event loop. |
119 |
|
120 |
For specific programs you can create additional event loops dynamically. |
121 |
|
122 |
If you want to take advantage of kqueue (which often works properly for |
123 |
sockets only) even though the default loop doesn't enable it, you can |
124 |
I<embed> a kqueue loop into the default loop: running the default loop |
125 |
will then also service the kqueue loop to some extent. See the example in |
126 |
the section about embed watchers for an example on how to achieve that. |
127 |
|
128 |
=over 4 |
129 |
|
130 |
=item $loop = new EV::Loop [$flags] |
131 |
|
132 |
Create a new event loop as per the specified flags. Please refer to |
133 |
the C<ev_loop_new ()> function description in the libev documentation |
134 |
(L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#GLOBAL_FUNCTIONS>, |
135 |
or locally-installed as F<EV::libev> manpage) for more info. |
136 |
|
137 |
The loop will automatically be destroyed when it is no longer referenced |
138 |
by any watcher and the loop object goes out of scope. |
139 |
|
140 |
If you are not embedding the loop, then Using C<EV::FLAG_FORKCHECK> |
141 |
is recommended, as only the default event loop is protected by this |
142 |
module. If you I<are> embedding this loop in the default loop, this is not |
143 |
necessary, as C<EV::embed> automatically does the right thing on fork. |
144 |
|
145 |
=item $loop->loop_fork |
146 |
|
147 |
Must be called after a fork in the child, before entering or continuing |
148 |
the event loop. An alternative is to use C<EV::FLAG_FORKCHECK> which calls |
149 |
this function automatically, at some performance loss (refer to the libev |
150 |
documentation). |
151 |
|
152 |
=item $loop->loop_verify |
153 |
|
154 |
Calls C<ev_verify> to make internal consistency checks (for debugging |
155 |
libev) and abort the program if any data structures were found to be |
156 |
corrupted. |
157 |
|
158 |
=item $loop = EV::default_loop [$flags] |
159 |
|
160 |
Return the default loop (which is a singleton object). Since this module |
161 |
already creates the default loop with default flags, specifying flags here |
162 |
will not have any effect unless you destroy the default loop first, which |
163 |
isn't supported. So in short: don't do it, and if you break it, you get to |
164 |
keep the pieces. |
165 |
|
166 |
=back |
167 |
|
168 |
|
169 |
=head1 BASIC INTERFACE |
170 |
|
171 |
=over 4 |
172 |
|
173 |
=item $EV::DIED |
174 |
|
175 |
Must contain a reference to a function that is called when a callback |
176 |
throws an exception (with $@ containing the error). The default prints an |
177 |
informative message and continues. |
178 |
|
179 |
If this callback throws an exception it will be silently ignored. |
180 |
|
181 |
=item $flags = EV::supported_backends |
182 |
|
183 |
=item $flags = EV::recommended_backends |
184 |
|
185 |
=item $flags = EV::embeddable_backends |
186 |
|
187 |
Returns the set (see C<EV::BACKEND_*> flags) of backends supported by this |
188 |
instance of EV, the set of recommended backends (supposed to be good) for |
189 |
this platform and the set of embeddable backends (see EMBED WATCHERS). |
190 |
|
191 |
=item EV::sleep $seconds |
192 |
|
193 |
Block the process for the given number of (fractional) seconds. |
194 |
|
195 |
=item $time = EV::time |
196 |
|
197 |
Returns the current time in (fractional) seconds since the epoch. |
198 |
|
199 |
=item $time = EV::now |
200 |
|
201 |
=item $time = $loop->now |
202 |
|
203 |
Returns the time the last event loop iteration has been started. This |
204 |
is the time that (relative) timers are based on, and referring to it is |
205 |
usually faster then calling EV::time. |
206 |
|
207 |
=item EV::now_update |
208 |
|
209 |
=item $loop->now_update |
210 |
|
211 |
Establishes the current time by querying the kernel, updating the time |
212 |
returned by C<EV::now> in the progress. This is a costly operation and |
213 |
is usually done automatically within C<EV::loop>. |
214 |
|
215 |
This function is rarely useful, but when some event callback runs for a |
216 |
very long time without entering the event loop, updating libev's idea of |
217 |
the current time is a good idea. |
218 |
|
219 |
=item EV::suspend |
220 |
|
221 |
=item $loop->suspend |
222 |
|
223 |
=item EV::resume |
224 |
|
225 |
=item $loop->resume |
226 |
|
227 |
These two functions suspend and resume a loop, for use when the loop is |
228 |
not used for a while and timeouts should not be processed. |
229 |
|
230 |
A typical use case would be an interactive program such as a game: When |
231 |
the user presses C<^Z> to suspend the game and resumes it an hour later it |
232 |
would be best to handle timeouts as if no time had actually passed while |
233 |
the program was suspended. This can be achieved by calling C<suspend> |
234 |
in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling |
235 |
C<resume> directly afterwards to resume timer processing. |
236 |
|
237 |
Effectively, all C<timer> watchers will be delayed by the time spend |
238 |
between C<suspend> and C<resume>, and all C<periodic> watchers |
239 |
will be rescheduled (that is, they will lose any events that would have |
240 |
occured while suspended). |
241 |
|
242 |
After calling C<suspend> you B<must not> call I<any> function on the given |
243 |
loop other than C<resume>, and you B<must not> call C<resume> |
244 |
without a previous call to C<suspend>. |
245 |
|
246 |
Calling C<suspend>/C<resume> has the side effect of updating the event |
247 |
loop time (see C<now_update>). |
248 |
|
249 |
=item $backend = EV::backend |
250 |
|
251 |
=item $backend = $loop->backend |
252 |
|
253 |
Returns an integer describing the backend used by libev (EV::BACKEND_SELECT |
254 |
or EV::BACKEND_EPOLL). |
255 |
|
256 |
=item EV::loop [$flags] |
257 |
|
258 |
=item $loop->loop ([$flags]) |
259 |
|
260 |
Begin checking for events and calling callbacks. It returns when a |
261 |
callback calls EV::unloop. |
262 |
|
263 |
The $flags argument can be one of the following: |
264 |
|
265 |
0 as above |
266 |
EV::LOOP_ONESHOT block at most once (wait, but do not loop) |
267 |
EV::LOOP_NONBLOCK do not block at all (fetch/handle events but do not wait) |
268 |
|
269 |
=item EV::unloop [$how] |
270 |
|
271 |
=item $loop->unloop ([$how]) |
272 |
|
273 |
When called with no arguments or an argument of EV::UNLOOP_ONE, makes the |
274 |
innermost call to EV::loop return. |
275 |
|
276 |
When called with an argument of EV::UNLOOP_ALL, all calls to EV::loop will return as |
277 |
fast as possible. |
278 |
|
279 |
=item $count = EV::loop_count |
280 |
|
281 |
=item $count = $loop->loop_count |
282 |
|
283 |
Return the number of times the event loop has polled for new |
284 |
events. Sometimes useful as a generation counter. |
285 |
|
286 |
=item EV::once $fh_or_undef, $events, $timeout, $cb->($revents) |
287 |
|
288 |
=item $loop->once ($fh_or_undef, $events, $timeout, $cb->($revents)) |
289 |
|
290 |
This function rolls together an I/O and a timer watcher for a single |
291 |
one-shot event without the need for managing a watcher object. |
292 |
|
293 |
If C<$fh_or_undef> is a filehandle or file descriptor, then C<$events> |
294 |
must be a bitset containing either C<EV::READ>, C<EV::WRITE> or C<EV::READ |
295 |
| EV::WRITE>, indicating the type of I/O event you want to wait for. If |
296 |
you do not want to wait for some I/O event, specify C<undef> for |
297 |
C<$fh_or_undef> and C<0> for C<$events>). |
298 |
|
299 |
If timeout is C<undef> or negative, then there will be no |
300 |
timeout. Otherwise a EV::timer with this value will be started. |
301 |
|
302 |
When an error occurs or either the timeout or I/O watcher triggers, then |
303 |
the callback will be called with the received event set (in general |
304 |
you can expect it to be a combination of C<EV::ERROR>, C<EV::READ>, |
305 |
C<EV::WRITE> and C<EV::TIMEOUT>). |
306 |
|
307 |
EV::once doesn't return anything: the watchers stay active till either |
308 |
of them triggers, then they will be stopped and freed, and the callback |
309 |
invoked. |
310 |
|
311 |
=item EV::feed_fd_event ($fd, $revents) |
312 |
|
313 |
=item $loop->feed_fd_event ($fd, $revents) |
314 |
|
315 |
Feed an event on a file descriptor into EV. EV will react to this call as |
316 |
if the readyness notifications specified by C<$revents> (a combination of |
317 |
C<EV::READ> and C<EV::WRITE>) happened on the file descriptor C<$fd>. |
318 |
|
319 |
=item EV::feed_signal_event ($signal) |
320 |
|
321 |
Feed a signal event into EV. EV will react to this call as if the signal |
322 |
specified by C<$signal> had occured. |
323 |
|
324 |
=item EV::set_io_collect_interval $time |
325 |
|
326 |
=item $loop->set_io_collect_interval ($time) |
327 |
|
328 |
=item EV::set_timeout_collect_interval $time |
329 |
|
330 |
=item $loop->set_timeout_collect_interval ($time) |
331 |
|
332 |
These advanced functions set the minimum block interval when polling for I/O events and the minimum |
333 |
wait interval for timer events. See the libev documentation at |
334 |
L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP> |
335 |
(locally installed as F<EV::libev>) for a more detailed discussion. |
336 |
|
337 |
=back |
338 |
|
339 |
|
340 |
=head1 WATCHER OBJECTS |
341 |
|
342 |
A watcher is an object that gets created to record your interest in some |
343 |
event. For instance, if you want to wait for STDIN to become readable, you |
344 |
would create an EV::io watcher for that: |
345 |
|
346 |
my $watcher = EV::io *STDIN, EV::READ, sub { |
347 |
my ($watcher, $revents) = @_; |
348 |
warn "yeah, STDIN should now be readable without blocking!\n" |
349 |
}; |
350 |
|
351 |
All watchers can be active (waiting for events) or inactive (paused). Only |
352 |
active watchers will have their callbacks invoked. All callbacks will be |
353 |
called with at least two arguments: the watcher and a bitmask of received |
354 |
events. |
355 |
|
356 |
Each watcher type has its associated bit in revents, so you can use the |
357 |
same callback for multiple watchers. The event mask is named after the |
358 |
type, i.e. EV::child sets EV::CHILD, EV::prepare sets EV::PREPARE, |
359 |
EV::periodic sets EV::PERIODIC and so on, with the exception of I/O events |
360 |
(which can set both EV::READ and EV::WRITE bits), and EV::timer (which |
361 |
uses EV::TIMEOUT). |
362 |
|
363 |
In the rare case where one wants to create a watcher but not start it at |
364 |
the same time, each constructor has a variant with a trailing C<_ns> in |
365 |
its name, e.g. EV::io has a non-starting variant EV::io_ns and so on. |
366 |
|
367 |
Please note that a watcher will automatically be stopped when the watcher |
368 |
object is destroyed, so you I<need> to keep the watcher objects returned by |
369 |
the constructors. |
370 |
|
371 |
Also, all methods changing some aspect of a watcher (->set, ->priority, |
372 |
->fh and so on) automatically stop and start it again if it is active, |
373 |
which means pending events get lost. |
374 |
|
375 |
=head2 COMMON WATCHER METHODS |
376 |
|
377 |
This section lists methods common to all watchers. |
378 |
|
379 |
=over 4 |
380 |
|
381 |
=item $w->start |
382 |
|
383 |
Starts a watcher if it isn't active already. Does nothing to an already |
384 |
active watcher. By default, all watchers start out in the active state |
385 |
(see the description of the C<_ns> variants if you need stopped watchers). |
386 |
|
387 |
=item $w->stop |
388 |
|
389 |
Stop a watcher if it is active. Also clear any pending events (events that |
390 |
have been received but that didn't yet result in a callback invocation), |
391 |
regardless of whether the watcher was active or not. |
392 |
|
393 |
=item $bool = $w->is_active |
394 |
|
395 |
Returns true if the watcher is active, false otherwise. |
396 |
|
397 |
=item $current_data = $w->data |
398 |
|
399 |
=item $old_data = $w->data ($new_data) |
400 |
|
401 |
Queries a freely usable data scalar on the watcher and optionally changes |
402 |
it. This is a way to associate custom data with a watcher: |
403 |
|
404 |
my $w = EV::timer 60, 0, sub { |
405 |
warn $_[0]->data; |
406 |
}; |
407 |
$w->data ("print me!"); |
408 |
|
409 |
=item $current_cb = $w->cb |
410 |
|
411 |
=item $old_cb = $w->cb ($new_cb) |
412 |
|
413 |
Queries the callback on the watcher and optionally changes it. You can do |
414 |
this at any time without the watcher restarting. |
415 |
|
416 |
=item $current_priority = $w->priority |
417 |
|
418 |
=item $old_priority = $w->priority ($new_priority) |
419 |
|
420 |
Queries the priority on the watcher and optionally changes it. Pending |
421 |
watchers with higher priority will be invoked first. The valid range of |
422 |
priorities lies between EV::MAXPRI (default 2) and EV::MINPRI (default |
423 |
-2). If the priority is outside this range it will automatically be |
424 |
normalised to the nearest valid priority. |
425 |
|
426 |
The default priority of any newly-created watcher is 0. |
427 |
|
428 |
Note that the priority semantics have not yet been fleshed out and are |
429 |
subject to almost certain change. |
430 |
|
431 |
=item $w->invoke ($revents) |
432 |
|
433 |
Call the callback *now* with the given event mask. |
434 |
|
435 |
=item $w->feed_event ($revents) |
436 |
|
437 |
Feed some events on this watcher into EV. EV will react to this call as if |
438 |
the watcher had received the given C<$revents> mask. |
439 |
|
440 |
=item $revents = $w->clear_pending |
441 |
|
442 |
If the watcher is pending, this function clears its pending status and |
443 |
returns its C<$revents> bitset (as if its callback was invoked). If the |
444 |
watcher isn't pending it does nothing and returns C<0>. |
445 |
|
446 |
=item $previous_state = $w->keepalive ($bool) |
447 |
|
448 |
Normally, C<EV::loop> will return when there are no active watchers |
449 |
(which is a "deadlock" because no progress can be made anymore). This is |
450 |
convinient because it allows you to start your watchers (and your jobs), |
451 |
call C<EV::loop> once and when it returns you know that all your jobs are |
452 |
finished (or they forgot to register some watchers for their task :). |
453 |
|
454 |
Sometimes, however, this gets in your way, for example when the module |
455 |
that calls C<EV::loop> (usually the main program) is not the same module |
456 |
as a long-living watcher (for example a DNS client module written by |
457 |
somebody else even). Then you might want any outstanding requests to be |
458 |
handled, but you would not want to keep C<EV::loop> from returning just |
459 |
because you happen to have this long-running UDP port watcher. |
460 |
|
461 |
In this case you can clear the keepalive status, which means that even |
462 |
though your watcher is active, it won't keep C<EV::loop> from returning. |
463 |
|
464 |
The initial value for keepalive is true (enabled), and you can change it |
465 |
any time. |
466 |
|
467 |
Example: Register an I/O watcher for some UDP socket but do not keep the |
468 |
event loop from running just because of that watcher. |
469 |
|
470 |
my $udp_socket = ... |
471 |
my $udp_watcher = EV::io $udp_socket, EV::READ, sub { ... }; |
472 |
$udp_watcher->keepalive (0); |
473 |
|
474 |
=item $loop = $w->loop |
475 |
|
476 |
Return the loop that this watcher is attached to. |
477 |
|
478 |
=back |
479 |
|
480 |
|
481 |
=head1 WATCHER TYPES |
482 |
|
483 |
Each of the following subsections describes a single watcher type. |
484 |
|
485 |
=head3 I/O WATCHERS - is this file descriptor readable or writable? |
486 |
|
487 |
=over 4 |
488 |
|
489 |
=item $w = EV::io $fileno_or_fh, $eventmask, $callback |
490 |
|
491 |
=item $w = EV::io_ns $fileno_or_fh, $eventmask, $callback |
492 |
|
493 |
=item $w = $loop->io ($fileno_or_fh, $eventmask, $callback) |
494 |
|
495 |
=item $w = $loop->io_ns ($fileno_or_fh, $eventmask, $callback) |
496 |
|
497 |
As long as the returned watcher object is alive, call the C<$callback> |
498 |
when at least one of events specified in C<$eventmask> occurs. |
499 |
|
500 |
The $eventmask can be one or more of these constants ORed together: |
501 |
|
502 |
EV::READ wait until read() wouldn't block anymore |
503 |
EV::WRITE wait until write() wouldn't block anymore |
504 |
|
505 |
The C<io_ns> variant doesn't start (activate) the newly created watcher. |
506 |
|
507 |
=item $w->set ($fileno_or_fh, $eventmask) |
508 |
|
509 |
Reconfigures the watcher, see the constructor above for details. Can be |
510 |
called at any time. |
511 |
|
512 |
=item $current_fh = $w->fh |
513 |
|
514 |
=item $old_fh = $w->fh ($new_fh) |
515 |
|
516 |
Returns the previously set filehandle and optionally set a new one. |
517 |
|
518 |
=item $current_eventmask = $w->events |
519 |
|
520 |
=item $old_eventmask = $w->events ($new_eventmask) |
521 |
|
522 |
Returns the previously set event mask and optionally set a new one. |
523 |
|
524 |
=back |
525 |
|
526 |
|
527 |
=head3 TIMER WATCHERS - relative and optionally repeating timeouts |
528 |
|
529 |
=over 4 |
530 |
|
531 |
=item $w = EV::timer $after, $repeat, $callback |
532 |
|
533 |
=item $w = EV::timer_ns $after, $repeat, $callback |
534 |
|
535 |
=item $w = $loop->timer ($after, $repeat, $callback) |
536 |
|
537 |
=item $w = $loop->timer_ns ($after, $repeat, $callback) |
538 |
|
539 |
Calls the callback after C<$after> seconds (which may be fractional). If |
540 |
C<$repeat> is non-zero, the timer will be restarted (with the $repeat |
541 |
value as $after) after the callback returns. |
542 |
|
543 |
This means that the callback would be called roughly after C<$after> |
544 |
seconds, and then every C<$repeat> seconds. The timer does his best not |
545 |
to drift, but it will not invoke the timer more often then once per event |
546 |
loop iteration, and might drift in other cases. If that isn't acceptable, |
547 |
look at EV::periodic, which can provide long-term stable timers. |
548 |
|
549 |
The timer is based on a monotonic clock, that is, if somebody is sitting |
550 |
in front of the machine while the timer is running and changes the system |
551 |
clock, the timer will nevertheless run (roughly) the same time. |
552 |
|
553 |
The C<timer_ns> variant doesn't start (activate) the newly created watcher. |
554 |
|
555 |
=item $w->set ($after, $repeat) |
556 |
|
557 |
Reconfigures the watcher, see the constructor above for details. Can be called at |
558 |
any time. |
559 |
|
560 |
=item $w->again |
561 |
|
562 |
Similar to the C<start> method, but has special semantics for repeating timers: |
563 |
|
564 |
If the timer is active and non-repeating, it will be stopped. |
565 |
|
566 |
If the timer is active and repeating, reset the timeout to occur |
567 |
C<$repeat> seconds after now. |
568 |
|
569 |
If the timer is inactive and repeating, start it using the repeat value. |
570 |
|
571 |
Otherwise do nothing. |
572 |
|
573 |
This behaviour is useful when you have a timeout for some IO |
574 |
operation. You create a timer object with the same value for C<$after> and |
575 |
C<$repeat>, and then, in the read/write watcher, run the C<again> method |
576 |
on the timeout. |
577 |
|
578 |
=back |
579 |
|
580 |
|
581 |
=head3 PERIODIC WATCHERS - to cron or not to cron? |
582 |
|
583 |
=over 4 |
584 |
|
585 |
=item $w = EV::periodic $at, $interval, $reschedule_cb, $callback |
586 |
|
587 |
=item $w = EV::periodic_ns $at, $interval, $reschedule_cb, $callback |
588 |
|
589 |
=item $w = $loop->periodic ($at, $interval, $reschedule_cb, $callback) |
590 |
|
591 |
=item $w = $loop->periodic_ns ($at, $interval, $reschedule_cb, $callback) |
592 |
|
593 |
Similar to EV::timer, but is not based on relative timeouts but on |
594 |
absolute times. Apart from creating "simple" timers that trigger "at" the |
595 |
specified time, it can also be used for non-drifting absolute timers and |
596 |
more complex, cron-like, setups that are not adversely affected by time |
597 |
jumps (i.e. when the system clock is changed by explicit date -s or other |
598 |
means such as ntpd). It is also the most complex watcher type in EV. |
599 |
|
600 |
It has three distinct "modes": |
601 |
|
602 |
=over 4 |
603 |
|
604 |
=item * absolute timer ($interval = $reschedule_cb = 0) |
605 |
|
606 |
This time simply fires at the wallclock time C<$at> and doesn't repeat. It |
607 |
will not adjust when a time jump occurs, that is, if it is to be run |
608 |
at January 1st 2011 then it will run when the system time reaches or |
609 |
surpasses this time. |
610 |
|
611 |
=item * repeating interval timer ($interval > 0, $reschedule_cb = 0) |
612 |
|
613 |
In this mode the watcher will always be scheduled to time out at the |
614 |
next C<$at + N * $interval> time (for some integer N) and then repeat, |
615 |
regardless of any time jumps. |
616 |
|
617 |
This can be used to create timers that do not drift with respect to system |
618 |
time: |
619 |
|
620 |
my $hourly = EV::periodic 0, 3600, 0, sub { print "once/hour\n" }; |
621 |
|
622 |
That doesn't mean there will always be 3600 seconds in between triggers, |
623 |
but only that the the clalback will be called when the system time shows a |
624 |
full hour (UTC). |
625 |
|
626 |
Another way to think about it (for the mathematically inclined) is that |
627 |
EV::periodic will try to run the callback in this mode at the next |
628 |
possible time where C<$time = $at (mod $interval)>, regardless of any time |
629 |
jumps. |
630 |
|
631 |
=item * manual reschedule mode ($reschedule_cb = coderef) |
632 |
|
633 |
In this mode $interval and $at are both being ignored. Instead, each |
634 |
time the periodic watcher gets scheduled, the reschedule callback |
635 |
($reschedule_cb) will be called with the watcher as first, and the current |
636 |
time as second argument. |
637 |
|
638 |
I<This callback MUST NOT stop or destroy this or any other periodic |
639 |
watcher, ever, and MUST NOT call any event loop functions or methods>. If |
640 |
you need to stop it, return 1e30 and stop it afterwards. You may create |
641 |
and start a C<EV::prepare> watcher for this task. |
642 |
|
643 |
It must return the next time to trigger, based on the passed time value |
644 |
(that is, the lowest time value larger than or equal to to the second |
645 |
argument). It will usually be called just before the callback will be |
646 |
triggered, but might be called at other times, too. |
647 |
|
648 |
This can be used to create very complex timers, such as a timer that |
649 |
triggers on each midnight, local time (actually 24 hours after the last |
650 |
midnight, to keep the example simple. If you know a way to do it correctly |
651 |
in about the same space (without requiring elaborate modules), drop me a |
652 |
note :): |
653 |
|
654 |
my $daily = EV::periodic 0, 0, sub { |
655 |
my ($w, $now) = @_; |
656 |
|
657 |
use Time::Local (); |
658 |
my (undef, undef, undef, $d, $m, $y) = localtime $now; |
659 |
86400 + Time::Local::timelocal 0, 0, 0, $d, $m, $y |
660 |
}, sub { |
661 |
print "it's midnight or likely shortly after, now\n"; |
662 |
}; |
663 |
|
664 |
=back |
665 |
|
666 |
The C<periodic_ns> variant doesn't start (activate) the newly created watcher. |
667 |
|
668 |
=item $w->set ($at, $interval, $reschedule_cb) |
669 |
|
670 |
Reconfigures the watcher, see the constructor above for details. Can be called at |
671 |
any time. |
672 |
|
673 |
=item $w->again |
674 |
|
675 |
Simply stops and starts the watcher again. |
676 |
|
677 |
=item $time = $w->at |
678 |
|
679 |
Return the time that the watcher is expected to trigger next. |
680 |
|
681 |
=back |
682 |
|
683 |
|
684 |
=head3 SIGNAL WATCHERS - signal me when a signal gets signalled! |
685 |
|
686 |
=over 4 |
687 |
|
688 |
=item $w = EV::signal $signal, $callback |
689 |
|
690 |
=item $w = EV::signal_ns $signal, $callback |
691 |
|
692 |
Call the callback when $signal is received (the signal can be specified by |
693 |
number or by name, just as with C<kill> or C<%SIG>). |
694 |
|
695 |
EV will grab the signal for the process (the kernel only allows one |
696 |
component to receive a signal at a time) when you start a signal watcher, |
697 |
and removes it again when you stop it. Perl does the same when you |
698 |
add/remove callbacks to C<%SIG>, so watch out. |
699 |
|
700 |
You can have as many signal watchers per signal as you want. |
701 |
|
702 |
The C<signal_ns> variant doesn't start (activate) the newly created watcher. |
703 |
|
704 |
=item $w->set ($signal) |
705 |
|
706 |
Reconfigures the watcher, see the constructor above for details. Can be |
707 |
called at any time. |
708 |
|
709 |
=item $current_signum = $w->signal |
710 |
|
711 |
=item $old_signum = $w->signal ($new_signal) |
712 |
|
713 |
Returns the previously set signal (always as a number not name) and |
714 |
optionally set a new one. |
715 |
|
716 |
=back |
717 |
|
718 |
|
719 |
=head3 CHILD WATCHERS - watch out for process status changes |
720 |
|
721 |
=over 4 |
722 |
|
723 |
=item $w = EV::child $pid, $trace, $callback |
724 |
|
725 |
=item $w = EV::child_ns $pid, $trace, $callback |
726 |
|
727 |
=item $w = $loop->child ($pid, $trace, $callback) |
728 |
|
729 |
=item $w = $loop->child_ns ($pid, $trace, $callback) |
730 |
|
731 |
Call the callback when a status change for pid C<$pid> (or any pid |
732 |
if C<$pid> is 0) has been received (a status change happens when the |
733 |
process terminates or is killed, or, when trace is true, additionally when |
734 |
it is stopped or continued). More precisely: when the process receives |
735 |
a C<SIGCHLD>, EV will fetch the outstanding exit/wait status for all |
736 |
changed/zombie children and call the callback. |
737 |
|
738 |
It is valid (and fully supported) to install a child watcher after a child |
739 |
has exited but before the event loop has started its next iteration (for |
740 |
example, first you C<fork>, then the new child process might exit, and |
741 |
only then do you install a child watcher in the parent for the new pid). |
742 |
|
743 |
You can access both exit (or tracing) status and pid by using the |
744 |
C<rstatus> and C<rpid> methods on the watcher object. |
745 |
|
746 |
You can have as many pid watchers per pid as you want, they will all be |
747 |
called. |
748 |
|
749 |
The C<child_ns> variant doesn't start (activate) the newly created watcher. |
750 |
|
751 |
=item $w->set ($pid, $trace) |
752 |
|
753 |
Reconfigures the watcher, see the constructor above for details. Can be called at |
754 |
any time. |
755 |
|
756 |
=item $current_pid = $w->pid |
757 |
|
758 |
Returns the previously set process id and optionally set a new one. |
759 |
|
760 |
=item $exit_status = $w->rstatus |
761 |
|
762 |
Return the exit/wait status (as returned by waitpid, see the waitpid entry |
763 |
in perlfunc). |
764 |
|
765 |
=item $pid = $w->rpid |
766 |
|
767 |
Return the pid of the awaited child (useful when you have installed a |
768 |
watcher for all pids). |
769 |
|
770 |
=back |
771 |
|
772 |
|
773 |
=head3 STAT WATCHERS - did the file attributes just change? |
774 |
|
775 |
=over 4 |
776 |
|
777 |
=item $w = EV::stat $path, $interval, $callback |
778 |
|
779 |
=item $w = EV::stat_ns $path, $interval, $callback |
780 |
|
781 |
=item $w = $loop->stat ($path, $interval, $callback) |
782 |
|
783 |
=item $w = $loop->stat_ns ($path, $interval, $callback) |
784 |
|
785 |
Call the callback when a file status change has been detected on |
786 |
C<$path>. The C<$path> does not need to exist, changing from "path exists" |
787 |
to "path does not exist" is a status change like any other. |
788 |
|
789 |
The C<$interval> is a recommended polling interval for systems where |
790 |
OS-supported change notifications don't exist or are not supported. If |
791 |
you use C<0> then an unspecified default is used (which is highly |
792 |
recommended!), which is to be expected to be around five seconds usually. |
793 |
|
794 |
This watcher type is not meant for massive numbers of stat watchers, |
795 |
as even with OS-supported change notifications, this can be |
796 |
resource-intensive. |
797 |
|
798 |
The C<stat_ns> variant doesn't start (activate) the newly created watcher. |
799 |
|
800 |
=item ... = $w->stat |
801 |
|
802 |
This call is very similar to the perl C<stat> built-in: It stats (using |
803 |
C<lstat>) the path specified in the watcher and sets perls stat cache (as |
804 |
well as EV's idea of the current stat values) to the values found. |
805 |
|
806 |
In scalar context, a boolean is return indicating success or failure of |
807 |
the stat. In list context, the same 13-value list as with stat is returned |
808 |
(except that the blksize and blocks fields are not reliable). |
809 |
|
810 |
In the case of an error, errno is set to C<ENOENT> (regardless of the |
811 |
actual error value) and the C<nlink> value is forced to zero (if the stat |
812 |
was successful then nlink is guaranteed to be non-zero). |
813 |
|
814 |
See also the next two entries for more info. |
815 |
|
816 |
=item ... = $w->attr |
817 |
|
818 |
Just like C<< $w->stat >>, but without the initial stat'ing: this returns |
819 |
the values most recently detected by EV. See the next entry for more info. |
820 |
|
821 |
=item ... = $w->prev |
822 |
|
823 |
Just like C<< $w->stat >>, but without the initial stat'ing: this returns |
824 |
the previous set of values, before the change. |
825 |
|
826 |
That is, when the watcher callback is invoked, C<< $w->prev >> will be set |
827 |
to the values found I<before> a change was detected, while C<< $w->attr >> |
828 |
returns the values found leading to the change detection. The difference (if any) |
829 |
between C<prev> and C<attr> is what triggered the callback. |
830 |
|
831 |
If you did something to the filesystem object and do not want to trigger |
832 |
yet another change, you can call C<stat> to update EV's idea of what the |
833 |
current attributes are. |
834 |
|
835 |
=item $w->set ($path, $interval) |
836 |
|
837 |
Reconfigures the watcher, see the constructor above for details. Can be |
838 |
called at any time. |
839 |
|
840 |
=item $current_path = $w->path |
841 |
|
842 |
=item $old_path = $w->path ($new_path) |
843 |
|
844 |
Returns the previously set path and optionally set a new one. |
845 |
|
846 |
=item $current_interval = $w->interval |
847 |
|
848 |
=item $old_interval = $w->interval ($new_interval) |
849 |
|
850 |
Returns the previously set interval and optionally set a new one. Can be |
851 |
used to query the actual interval used. |
852 |
|
853 |
=back |
854 |
|
855 |
|
856 |
=head3 IDLE WATCHERS - when you've got nothing better to do... |
857 |
|
858 |
=over 4 |
859 |
|
860 |
=item $w = EV::idle $callback |
861 |
|
862 |
=item $w = EV::idle_ns $callback |
863 |
|
864 |
=item $w = $loop->idle ($callback) |
865 |
|
866 |
=item $w = $loop->idle_ns ($callback) |
867 |
|
868 |
Call the callback when there are no other pending watchers of the same or |
869 |
higher priority (excluding check, prepare and other idle watchers of the |
870 |
same or lower priority, of course). They are called idle watchers because |
871 |
when the watcher is the highest priority pending event in the process, the |
872 |
process is considered to be idle at that priority. |
873 |
|
874 |
If you want a watcher that is only ever called when I<no> other events are |
875 |
outstanding you have to set the priority to C<EV::MINPRI>. |
876 |
|
877 |
The process will not block as long as any idle watchers are active, and |
878 |
they will be called repeatedly until stopped. |
879 |
|
880 |
For example, if you have idle watchers at priority C<0> and C<1>, and |
881 |
an I/O watcher at priority C<0>, then the idle watcher at priority C<1> |
882 |
and the I/O watcher will always run when ready. Only when the idle watcher |
883 |
at priority C<1> is stopped and the I/O watcher at priority C<0> is not |
884 |
pending with the C<0>-priority idle watcher be invoked. |
885 |
|
886 |
The C<idle_ns> variant doesn't start (activate) the newly created watcher. |
887 |
|
888 |
=back |
889 |
|
890 |
|
891 |
=head3 PREPARE WATCHERS - customise your event loop! |
892 |
|
893 |
=over 4 |
894 |
|
895 |
=item $w = EV::prepare $callback |
896 |
|
897 |
=item $w = EV::prepare_ns $callback |
898 |
|
899 |
=item $w = $loop->prepare ($callback) |
900 |
|
901 |
=item $w = $loop->prepare_ns ($callback) |
902 |
|
903 |
Call the callback just before the process would block. You can still |
904 |
create/modify any watchers at this point. |
905 |
|
906 |
See the EV::check watcher, below, for explanations and an example. |
907 |
|
908 |
The C<prepare_ns> variant doesn't start (activate) the newly created watcher. |
909 |
|
910 |
=back |
911 |
|
912 |
|
913 |
=head3 CHECK WATCHERS - customise your event loop even more! |
914 |
|
915 |
=over 4 |
916 |
|
917 |
=item $w = EV::check $callback |
918 |
|
919 |
=item $w = EV::check_ns $callback |
920 |
|
921 |
=item $w = $loop->check ($callback) |
922 |
|
923 |
=item $w = $loop->check_ns ($callback) |
924 |
|
925 |
Call the callback just after the process wakes up again (after it has |
926 |
gathered events), but before any other callbacks have been invoked. |
927 |
|
928 |
This is used to integrate other event-based software into the EV |
929 |
mainloop: You register a prepare callback and in there, you create io and |
930 |
timer watchers as required by the other software. Here is a real-world |
931 |
example of integrating Net::SNMP (with some details left out): |
932 |
|
933 |
our @snmp_watcher; |
934 |
|
935 |
our $snmp_prepare = EV::prepare sub { |
936 |
# do nothing unless active |
937 |
$dispatcher->{_event_queue_h} |
938 |
or return; |
939 |
|
940 |
# make the dispatcher handle any outstanding stuff |
941 |
... not shown |
942 |
|
943 |
# create an I/O watcher for each and every socket |
944 |
@snmp_watcher = ( |
945 |
(map { EV::io $_, EV::READ, sub { } } |
946 |
keys %{ $dispatcher->{_descriptors} }), |
947 |
|
948 |
EV::timer +($event->[Net::SNMP::Dispatcher::_ACTIVE] |
949 |
? $event->[Net::SNMP::Dispatcher::_TIME] - EV::now : 0), |
950 |
0, sub { }, |
951 |
); |
952 |
}; |
953 |
|
954 |
The callbacks are irrelevant (and are not even being called), the |
955 |
only purpose of those watchers is to wake up the process as soon as |
956 |
one of those events occurs (socket readable, or timer timed out). The |
957 |
corresponding EV::check watcher will then clean up: |
958 |
|
959 |
our $snmp_check = EV::check sub { |
960 |
# destroy all watchers |
961 |
@snmp_watcher = (); |
962 |
|
963 |
# make the dispatcher handle any new stuff |
964 |
... not shown |
965 |
}; |
966 |
|
967 |
The callbacks of the created watchers will not be called as the watchers |
968 |
are destroyed before this can happen (remember EV::check gets called |
969 |
first). |
970 |
|
971 |
The C<check_ns> variant doesn't start (activate) the newly created watcher. |
972 |
|
973 |
=back |
974 |
|
975 |
|
976 |
=head3 FORK WATCHERS - the audacity to resume the event loop after a fork |
977 |
|
978 |
Fork watchers are called when a C<fork ()> was detected. The invocation |
979 |
is done before the event loop blocks next and before C<check> watchers |
980 |
are being called, and only in the child after the fork. |
981 |
|
982 |
=over 4 |
983 |
|
984 |
=item $w = EV::fork $callback |
985 |
|
986 |
=item $w = EV::fork_ns $callback |
987 |
|
988 |
=item $w = $loop->fork ($callback) |
989 |
|
990 |
=item $w = $loop->fork_ns ($callback) |
991 |
|
992 |
Call the callback before the event loop is resumed in the child process |
993 |
after a fork. |
994 |
|
995 |
The C<fork_ns> variant doesn't start (activate) the newly created watcher. |
996 |
|
997 |
=back |
998 |
|
999 |
|
1000 |
=head3 EMBED WATCHERS - when one backend isn't enough... |
1001 |
|
1002 |
This is a rather advanced watcher type that lets you embed one event loop |
1003 |
into another (currently only IO events are supported in the embedded |
1004 |
loop, other types of watchers might be handled in a delayed or incorrect |
1005 |
fashion and must not be used). |
1006 |
|
1007 |
See the libev documentation at |
1008 |
L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#code_ev_embed_code_when_one_backend_> |
1009 |
(locally installed as F<EV::libev>) for more details. |
1010 |
|
1011 |
In short, this watcher is most useful on BSD systems without working |
1012 |
kqueue to still be able to handle a large number of sockets: |
1013 |
|
1014 |
my $socket_loop; |
1015 |
|
1016 |
# check wether we use SELECT or POLL _and_ KQUEUE is supported |
1017 |
if ( |
1018 |
(EV::backend & (EV::BACKEND_POLL | EV::BACKEND_SELECT)) |
1019 |
&& (EV::supported_backends & EV::embeddable_backends & EV::BACKEND_KQUEUE) |
1020 |
) { |
1021 |
# use kqueue for sockets |
1022 |
$socket_loop = new EV::Loop EV::BACKEND_KQUEUE | EV::FLAG_NOENV; |
1023 |
} |
1024 |
|
1025 |
# use the default loop otherwise |
1026 |
$socket_loop ||= EV::default_loop; |
1027 |
|
1028 |
=over 4 |
1029 |
|
1030 |
=item $w = EV::embed $otherloop[, $callback] |
1031 |
|
1032 |
=item $w = EV::embed_ns $otherloop[, $callback] |
1033 |
|
1034 |
=item $w = $loop->embed ($otherloop[, $callback]) |
1035 |
|
1036 |
=item $w = $loop->embed_ns ($otherloop[, $callback]) |
1037 |
|
1038 |
Call the callback when the embedded event loop (C<$otherloop>) has any |
1039 |
I/O activity. The C<$callback> is optional: if it is missing, then the |
1040 |
embedded event loop will be managed automatically (which is recommended), |
1041 |
otherwise you have to invoke C<sweep> yourself. |
1042 |
|
1043 |
The C<embed_ns> variant doesn't start (activate) the newly created watcher. |
1044 |
|
1045 |
=back |
1046 |
|
1047 |
=head3 ASYNC WATCHERS - how to wake up another event loop |
1048 |
|
1049 |
Async watchers are provided by EV, but have little use in perl directly, |
1050 |
as perl neither supports threads running in parallel nor direct access to |
1051 |
signal handlers or other contexts where they could be of value. |
1052 |
|
1053 |
It is, however, possible to use them from the XS level. |
1054 |
|
1055 |
Please see the libev documentation for further details. |
1056 |
|
1057 |
=over 4 |
1058 |
|
1059 |
=item $w = EV::async $callback |
1060 |
|
1061 |
=item $w = EV::async_ns $callback |
1062 |
|
1063 |
=item $w->send |
1064 |
|
1065 |
=item $bool = $w->async_pending |
1066 |
|
1067 |
=back |
1068 |
|
1069 |
|
1070 |
=head1 PERL SIGNALS |
1071 |
|
1072 |
While Perl signal handling (C<%SIG>) is not affected by EV, the behaviour |
1073 |
with EV is as the same as any other C library: Perl-signals will only be |
1074 |
handled when Perl runs, which means your signal handler might be invoked |
1075 |
only the next time an event callback is invoked. |
1076 |
|
1077 |
The solution is to use EV signal watchers (see C<EV::signal>), which will |
1078 |
ensure proper operations with regards to other event watchers. |
1079 |
|
1080 |
If you cannot do this for whatever reason, you can also force a watcher |
1081 |
to be called on every event loop iteration by installing a C<EV::check> |
1082 |
watcher: |
1083 |
|
1084 |
my $async_check = EV::check sub { }; |
1085 |
|
1086 |
This ensures that perl gets into control for a short time to handle any |
1087 |
pending signals, and also ensures (slightly) slower overall operation. |
1088 |
|
1089 |
=head1 ITHREADS |
1090 |
|
1091 |
Ithreads are not supported by this module in any way. Perl pseudo-threads |
1092 |
is evil stuff and must die. Real threads as provided by Coro are fully |
1093 |
supported (and enhanced support is available via L<Coro::EV>). |
1094 |
|
1095 |
=head1 FORK |
1096 |
|
1097 |
Most of the "improved" event delivering mechanisms of modern operating |
1098 |
systems have quite a few problems with fork(2) (to put it bluntly: it is |
1099 |
not supported and usually destructive). Libev makes it possible to work |
1100 |
around this by having a function that recreates the kernel state after |
1101 |
fork in the child. |
1102 |
|
1103 |
On non-win32 platforms, this module requires the pthread_atfork |
1104 |
functionality to do this automatically for you. This function is quite |
1105 |
buggy on most BSDs, though, so YMMV. The overhead for this is quite |
1106 |
negligible, because everything the function currently does is set a flag |
1107 |
that is checked only when the event loop gets used the next time, so when |
1108 |
you do fork but not use EV, the overhead is minimal. |
1109 |
|
1110 |
On win32, there is no notion of fork so all this doesn't apply, of course. |
1111 |
|
1112 |
=cut |
1113 |
|
1114 |
our $DIED = sub { |
1115 |
warn "EV: error in callback (ignoring): $@"; |
1116 |
}; |
1117 |
|
1118 |
default_loop |
1119 |
or die 'EV: cannot initialise libev backend. bad $ENV{LIBEV_FLAGS}?'; |
1120 |
|
1121 |
1; |
1122 |
|
1123 |
=head1 SEE ALSO |
1124 |
|
1125 |
L<EV::ADNS> (asynchronous DNS), L<Glib::EV> (makes Glib/Gtk2 use EV as |
1126 |
event loop), L<EV::Glib> (embed Glib into EV), L<Coro::EV> (efficient |
1127 |
coroutines with EV), L<Net::SNMP::EV> (asynchronous SNMP), L<AnyEvent> for |
1128 |
event-loop agnostic and portable event driven programming. |
1129 |
|
1130 |
=head1 AUTHOR |
1131 |
|
1132 |
Marc Lehmann <schmorp@schmorp.de> |
1133 |
http://home.schmorp.de/ |
1134 |
|
1135 |
=cut |
1136 |
|