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::loop; # loop until EV::unloop is called or all watchers stop |
52 |
EV::loop EV::LOOP_ONESHOT; # block until at least one event could be handled |
53 |
EV::loop EV::LOOP_NONBLOCK; # try to handle same events, but do not block |
54 |
|
55 |
DESCRIPTION |
56 |
This module provides an interface to libev |
57 |
(<http://software.schmorp.de/pkg/libev.html>). While the documentation |
58 |
below is comprehensive, one might also consult the documentation of |
59 |
libev itself (<http://cvs.schmorp.de/libev/ev.html>) for more subtle |
60 |
details on watcher semantics or some discussion on the available |
61 |
backends, or how to force a specific backend with "LIBEV_FLAGS", or just |
62 |
about in any case because it has much more detailed information. |
63 |
|
64 |
EVENT LOOPS |
65 |
EV supports multiple event loops: There is a single "default event loop" |
66 |
that can handle everything including signals and child watchers, and any |
67 |
number of "dynamic event loops" that can use different backends (with |
68 |
various limitations), but no child and signal watchers. |
69 |
|
70 |
You do not have to do anything to create the default event loop: When |
71 |
the module is loaded a suitable backend is selected on the premise of |
72 |
selecting a working backend (which for example rules out kqueue on most |
73 |
BSDs). Modules should, unless they have "special needs" always use the |
74 |
default loop as this is fastest (perl-wise), best supported by other |
75 |
modules (e.g. AnyEvent or Coro) and most portable event loop. |
76 |
|
77 |
For specific programs you can create additional event loops dynamically. |
78 |
|
79 |
$loop = new EV::loop [$flags] |
80 |
Create a new event loop as per the specified flags. Please refer to |
81 |
the "ev_loop_new ()" function description in the libev documentation |
82 |
(<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#GLOBAL_FUNCTI |
83 |
ONS>) for more info. |
84 |
|
85 |
The loop will automatically be destroyed when it is no longer |
86 |
referenced by any watcher and the loop object goes out of scope. |
87 |
|
88 |
Using "EV::FLAG_FORKCHECK" is recommended, as only the default event |
89 |
loop is protected by this module. |
90 |
|
91 |
$loop->loop_fork |
92 |
Must be called after a fork in the child, before entering or |
93 |
continuing the event loop. An alternative is to use |
94 |
"EV::FLAG_FORKCHECK" which calls this fucntion automatically, at |
95 |
some performance loss (refer to the libev documentation). |
96 |
|
97 |
$loop = EV::default_loop [$flags] |
98 |
Return the default loop (which is a singleton object). |
99 |
|
100 |
BASIC INTERFACE |
101 |
$EV::DIED |
102 |
Must contain a reference to a function that is called when a |
103 |
callback throws an exception (with $@ containing the error). The |
104 |
default prints an informative message and continues. |
105 |
|
106 |
If this callback throws an exception it will be silently ignored. |
107 |
|
108 |
$flags = EV::supported_backends |
109 |
$flags = EV::recommended_backends |
110 |
$flags = EV::embeddable_backends |
111 |
Returns the set (see "EV::BACKEND_*" flags) of backends supported by |
112 |
this instance of EV, the set of recommended backends (supposed to be |
113 |
good) for this platform and the set of embeddable backends (see |
114 |
EMBED WATCHERS). |
115 |
|
116 |
EV::sleep $seconds |
117 |
Block the process for the given number of (fractional) seconds. |
118 |
|
119 |
$time = EV::time |
120 |
Returns the current time in (fractional) seconds since the epoch. |
121 |
|
122 |
$time = EV::now |
123 |
$time = $loop->now |
124 |
Returns the time the last event loop iteration has been started. |
125 |
This is the time that (relative) timers are based on, and refering |
126 |
to it is usually faster then calling EV::time. |
127 |
|
128 |
$backend = EV::backend |
129 |
$backend = $loop->backend |
130 |
Returns an integer describing the backend used by libev |
131 |
(EV::METHOD_SELECT or EV::METHOD_EPOLL). |
132 |
|
133 |
EV::loop [$flags] |
134 |
$loop->loop ([$flags]) |
135 |
Begin checking for events and calling callbacks. It returns when a |
136 |
callback calls EV::unloop. |
137 |
|
138 |
The $flags argument can be one of the following: |
139 |
|
140 |
0 as above |
141 |
EV::LOOP_ONESHOT block at most once (wait, but do not loop) |
142 |
EV::LOOP_NONBLOCK do not block at all (fetch/handle events but do not wait) |
143 |
|
144 |
EV::unloop [$how] |
145 |
$loop->unloop ([$how]) |
146 |
When called with no arguments or an argument of EV::UNLOOP_ONE, |
147 |
makes the innermost call to EV::loop return. |
148 |
|
149 |
When called with an argument of EV::UNLOOP_ALL, all calls to |
150 |
EV::loop will return as fast as possible. |
151 |
|
152 |
$count = EV::loop_count |
153 |
$count = $loop->loop_count |
154 |
Return the number of times the event loop has polled for new events. |
155 |
Sometiems useful as a generation counter. |
156 |
|
157 |
EV::once $fh_or_undef, $events, $timeout, $cb->($revents) |
158 |
$loop->once ($fh_or_undef, $events, $timeout, $cb->($revents)) |
159 |
This function rolls together an I/O and a timer watcher for a single |
160 |
one-shot event without the need for managing a watcher object. |
161 |
|
162 |
If $fh_or_undef is a filehandle or file descriptor, then $events |
163 |
must be a bitset containing either "EV::READ", "EV::WRITE" or |
164 |
"EV::READ | EV::WRITE", indicating the type of I/O event you want to |
165 |
wait for. If you do not want to wait for some I/O event, specify |
166 |
"undef" for $fh_or_undef and 0 for $events). |
167 |
|
168 |
If timeout is "undef" or negative, then there will be no timeout. |
169 |
Otherwise a EV::timer with this value will be started. |
170 |
|
171 |
When an error occurs or either the timeout or I/O watcher triggers, |
172 |
then the callback will be called with the received event set (in |
173 |
general you can expect it to be a combination of "EV::ERROR", |
174 |
"EV::READ", "EV::WRITE" and "EV::TIMEOUT"). |
175 |
|
176 |
EV::once doesn't return anything: the watchers stay active till |
177 |
either of them triggers, then they will be stopped and freed, and |
178 |
the callback invoked. |
179 |
|
180 |
EV::feed_fd_event ($fd, $revents) |
181 |
$loop->feed_fd_event ($fd, $revents) |
182 |
Feed an event on a file descriptor into EV. EV will react to this |
183 |
call as if the readyness notifications specified by $revents (a |
184 |
combination of "EV::READ" and "EV::WRITE") happened on the file |
185 |
descriptor $fd. |
186 |
|
187 |
EV::feed_signal_event ($signal) |
188 |
Feed a signal event into EV. EV will react to this call as if the |
189 |
signal specified by $signal had occured. |
190 |
|
191 |
EV::set_io_collect_interval $time |
192 |
$loop->set_io_collect_interval ($time) |
193 |
EV::set_timeout_collect_interval $time |
194 |
$loop->set_timeout_collect_interval ($time) |
195 |
These advanced functions set the minimum block interval when polling |
196 |
for I/O events and the minimum wait interval for timer events. See |
197 |
the libev documentation at |
198 |
<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#FUNCTIONS_CONT |
199 |
ROLLING_THE_EVENT_LOOP> for a more detailed discussion. |
200 |
|
201 |
WATCHER OBJECTS |
202 |
A watcher is an object that gets created to record your interest in some |
203 |
event. For instance, if you want to wait for STDIN to become readable, |
204 |
you would create an EV::io watcher for that: |
205 |
|
206 |
my $watcher = EV::io *STDIN, EV::READ, sub { |
207 |
my ($watcher, $revents) = @_; |
208 |
warn "yeah, STDIN should now be readable without blocking!\n" |
209 |
}; |
210 |
|
211 |
All watchers can be active (waiting for events) or inactive (paused). |
212 |
Only active watchers will have their callbacks invoked. All callbacks |
213 |
will be called with at least two arguments: the watcher and a bitmask of |
214 |
received events. |
215 |
|
216 |
Each watcher type has its associated bit in revents, so you can use the |
217 |
same callback for multiple watchers. The event mask is named after the |
218 |
type, i..e. EV::child sets EV::CHILD, EV::prepare sets EV::PREPARE, |
219 |
EV::periodic sets EV::PERIODIC and so on, with the exception of I/O |
220 |
events (which can set both EV::READ and EV::WRITE bits), and EV::timer |
221 |
(which uses EV::TIMEOUT). |
222 |
|
223 |
In the rare case where one wants to create a watcher but not start it at |
224 |
the same time, each constructor has a variant with a trailing "_ns" in |
225 |
its name, e.g. EV::io has a non-starting variant EV::io_ns and so on. |
226 |
|
227 |
Please note that a watcher will automatically be stopped when the |
228 |
watcher object is destroyed, so you *need* to keep the watcher objects |
229 |
returned by the constructors. |
230 |
|
231 |
Also, all methods changing some aspect of a watcher (->set, ->priority, |
232 |
->fh and so on) automatically stop and start it again if it is active, |
233 |
which means pending events get lost. |
234 |
|
235 |
COMMON WATCHER METHODS |
236 |
This section lists methods common to all watchers. |
237 |
|
238 |
$w->start |
239 |
Starts a watcher if it isn't active already. Does nothing to an |
240 |
already active watcher. By default, all watchers start out in the |
241 |
active state (see the description of the "_ns" variants if you need |
242 |
stopped watchers). |
243 |
|
244 |
$w->stop |
245 |
Stop a watcher if it is active. Also clear any pending events |
246 |
(events that have been received but that didn't yet result in a |
247 |
callback invocation), regardless of whether the watcher was active |
248 |
or not. |
249 |
|
250 |
$bool = $w->is_active |
251 |
Returns true if the watcher is active, false otherwise. |
252 |
|
253 |
$current_data = $w->data |
254 |
$old_data = $w->data ($new_data) |
255 |
Queries a freely usable data scalar on the watcher and optionally |
256 |
changes it. This is a way to associate custom data with a watcher: |
257 |
|
258 |
my $w = EV::timer 60, 0, sub { |
259 |
warn $_[0]->data; |
260 |
}; |
261 |
$w->data ("print me!"); |
262 |
|
263 |
$current_cb = $w->cb |
264 |
$old_cb = $w->cb ($new_cb) |
265 |
Queries the callback on the watcher and optionally changes it. You |
266 |
can do this at any time without the watcher restarting. |
267 |
|
268 |
$current_priority = $w->priority |
269 |
$old_priority = $w->priority ($new_priority) |
270 |
Queries the priority on the watcher and optionally changes it. |
271 |
Pending watchers with higher priority will be invoked first. The |
272 |
valid range of priorities lies between EV::MAXPRI (default 2) and |
273 |
EV::MINPRI (default -2). If the priority is outside this range it |
274 |
will automatically be normalised to the nearest valid priority. |
275 |
|
276 |
The default priority of any newly-created watcher is 0. |
277 |
|
278 |
Note that the priority semantics have not yet been fleshed out and |
279 |
are subject to almost certain change. |
280 |
|
281 |
$w->invoke ($revents) |
282 |
Call the callback *now* with the given event mask. |
283 |
|
284 |
$w->feed_event ($revents) |
285 |
Feed some events on this watcher into EV. EV will react to this call |
286 |
as if the watcher had received the given $revents mask. |
287 |
|
288 |
$revents = $w->clear_pending |
289 |
If the watcher is pending, this function clears its pending status |
290 |
and returns its $revents bitset (as if its callback was invoked). If |
291 |
the watcher isn't pending it does nothing and returns 0. |
292 |
|
293 |
$previous_state = $w->keepalive ($bool) |
294 |
Normally, "EV::loop" will return when there are no active watchers |
295 |
(which is a "deadlock" because no progress can be made anymore). |
296 |
This is convinient because it allows you to start your watchers (and |
297 |
your jobs), call "EV::loop" once and when it returns you know that |
298 |
all your jobs are finished (or they forgot to register some watchers |
299 |
for their task :). |
300 |
|
301 |
Sometimes, however, this gets in your way, for example when the |
302 |
module that calls "EV::loop" (usually the main program) is not the |
303 |
same module as a long-living watcher (for example a DNS client |
304 |
module written by somebody else even). Then you might want any |
305 |
outstanding requests to be handled, but you would not want to keep |
306 |
"EV::loop" from returning just because you happen to have this |
307 |
long-running UDP port watcher. |
308 |
|
309 |
In this case you can clear the keepalive status, which means that |
310 |
even though your watcher is active, it won't keep "EV::loop" from |
311 |
returning. |
312 |
|
313 |
The initial value for keepalive is true (enabled), and you cna |
314 |
change it any time. |
315 |
|
316 |
Example: Register an I/O watcher for some UDP socket but do not keep |
317 |
the event loop from running just because of that watcher. |
318 |
|
319 |
my $udp_socket = ... |
320 |
my $udp_watcher = EV::io $udp_socket, EV::READ, sub { ... }; |
321 |
$1000udp_watcher->keepalive (0); |
322 |
|
323 |
$loop = $w->loop |
324 |
Return the loop that this watcher is attached to. |
325 |
|
326 |
WATCHER TYPES |
327 |
Each of the following subsections describes a single watcher type. |
328 |
|
329 |
I/O WATCHERS - is this file descriptor readable or writable? |
330 |
$w = EV::io $fileno_or_fh, $eventmask, $callback |
331 |
$w = EV::io_ns $fileno_or_fh, $eventmask, $callback |
332 |
$w = $loop->io ($fileno_or_fh, $eventmask, $callback) |
333 |
$w = $loop->io_ns ($fileno_or_fh, $eventmask, $callback) |
334 |
As long as the returned watcher object is alive, call the $callback |
335 |
when at least one of events specified in $eventmask occurs. |
336 |
|
337 |
The $eventmask can be one or more of these constants ORed together: |
338 |
|
339 |
EV::READ wait until read() wouldn't block anymore |
340 |
EV::WRITE wait until write() wouldn't block anymore |
341 |
|
342 |
The "io_ns" variant doesn't start (activate) the newly created |
343 |
watcher. |
344 |
|
345 |
$w->set ($fileno_or_fh, $eventmask) |
346 |
Reconfigures the watcher, see the constructor above for details. Can |
347 |
be called at any time. |
348 |
|
349 |
$current_fh = $w->fh |
350 |
$old_fh = $w->fh ($new_fh) |
351 |
Returns the previously set filehandle and optionally set a new one. |
352 |
|
353 |
$current_eventmask = $w->events |
354 |
$old_eventmask = $w->events ($new_eventmask) |
355 |
Returns the previously set event mask and optionally set a new one. |
356 |
|
357 |
TIMER WATCHERS - relative and optionally repeating timeouts |
358 |
$w = EV::timer $after, $repeat, $callback |
359 |
$w = EV::timer_ns $after, $repeat, $callback |
360 |
$w = $loop->timer ($after, $repeat, $callback) |
361 |
$w = $loop->timer_ns ($after, $repeat, $callback) |
362 |
Calls the callback after $after seconds (which may be fractional). |
363 |
If $repeat is non-zero, the timer will be restarted (with the |
364 |
$repeat value as $after) after the callback returns. |
365 |
|
366 |
This means that the callback would be called roughly after $after |
367 |
seconds, and then every $repeat seconds. The timer does his best not |
368 |
to drift, but it will not invoke the timer more often then once per |
369 |
event loop iteration, and might drift in other cases. If that isn't |
370 |
acceptable, look at EV::periodic, which can provide long-term stable |
371 |
timers. |
372 |
|
373 |
The timer is based on a monotonic clock, that is, if somebody is |
374 |
sitting in front of the machine while the timer is running and |
375 |
changes the system clock, the timer will nevertheless run (roughly) |
376 |
the same time. |
377 |
|
378 |
The "timer_ns" variant doesn't start (activate) the newly created |
379 |
watcher. |
380 |
|
381 |
$w->set ($after, $repeat) |
382 |
Reconfigures the watcher, see the constructor above for details. Can |
383 |
be called at any time. |
384 |
|
385 |
$w->again |
386 |
Similar to the "start" method, but has special semantics for |
387 |
repeating timers: |
388 |
|
389 |
If the timer is active and non-repeating, it will be stopped. |
390 |
|
391 |
If the timer is active and repeating, reset the timeout to occur |
392 |
$repeat seconds after now. |
393 |
|
394 |
If the timer is inactive and repeating, start it using the repeat |
395 |
value. |
396 |
|
397 |
Otherwise do nothing. |
398 |
|
399 |
This behaviour is useful when you have a timeout for some IO |
400 |
operation. You create a timer object with the same value for $after |
401 |
and $repeat, and then, in the read/write watcher, run the "again" |
402 |
method on the timeout. |
403 |
|
404 |
PERIODIC WATCHERS - to cron or not to cron? |
405 |
$w = EV::periodic $at, $interval, $reschedule_cb, $callback |
406 |
$w = EV::periodic_ns $at, $interval, $reschedule_cb, $callback |
407 |
$w = $loop->periodic ($at, $interval, $reschedule_cb, $callback) |
408 |
$w = $loop->periodic_ns ($at, $interval, $reschedule_cb, $callback) |
409 |
Similar to EV::timer, but is not based on relative timeouts but on |
410 |
absolute times. Apart from creating "simple" timers that trigger |
411 |
"at" the specified time, it can also be used for non-drifting |
412 |
absolute timers and more complex, cron-like, setups that are not |
413 |
adversely affected by time jumps (i.e. when the system clock is |
414 |
changed by explicit date -s or other means such as ntpd). It is also |
415 |
the most complex watcher type in EV. |
416 |
|
417 |
It has three distinct "modes": |
418 |
|
419 |
* absolute timer ($interval = $reschedule_cb = 0) |
420 |
|
421 |
This time simply fires at the wallclock time $at and doesn't |
422 |
repeat. It will not adjust when a time jump occurs, that is, if |
423 |
it is to be run at January 1st 2011 then it will run when the |
424 |
system time reaches or surpasses this time. |
425 |
|
426 |
* repeating interval timer ($interval > 0, $reschedule_cb = 0) |
427 |
|
428 |
In this mode the watcher will always be scheduled to time out at |
429 |
the next "$at + N * $interval" time (for some integer N) and |
430 |
then repeat, regardless of any time jumps. |
431 |
|
432 |
This can be used to create timers that do not drift with respect |
433 |
to system time: |
434 |
|
435 |
my $hourly = EV::periodic 0, 3600, 0, sub { print "once/hour\n" }; |
436 |
|
437 |
That doesn't mean there will always be 3600 seconds in between |
438 |
triggers, but only that the the clalback will be called when the |
439 |
system time shows a full hour (UTC). |
440 |
|
441 |
Another way to think about it (for the mathematically inclined) |
442 |
is that EV::periodic will try to run the callback in this mode |
443 |
at the next possible time where "$time = $at (mod $interval)", |
444 |
regardless of any time jumps. |
445 |
|
446 |
* manual reschedule mode ($reschedule_cb = coderef) |
447 |
|
448 |
In this mode $interval and $at are both being ignored. Instead, |
449 |
each time the periodic watcher gets scheduled, the reschedule |
450 |
callback ($reschedule_cb) will be called with the watcher as |
451 |
first, and the current time as second argument. |
452 |
|
453 |
*This callback MUST NOT stop or destroy this or any other |
454 |
periodic watcher, ever*. If you need to stop it, return 1e30 and |
455 |
stop it afterwards. |
456 |
|
457 |
It must return the next time to trigger, based on the passed |
458 |
time value (that is, the lowest time value larger than to the |
459 |
second argument). It will usually be called just before the |
460 |
callback will be triggered, but might be called at other times, |
461 |
too. |
462 |
|
463 |
This can be used to create very complex timers, such as a timer |
464 |
that triggers on each midnight, local time (actually 24 hours |
465 |
after the last midnight, to keep the example simple. If you know |
466 |
a way to do it correctly in about the same space (without |
467 |
requiring elaborate modules), drop me a note :): |
468 |
|
469 |
my $daily = EV::periodic 0, 0, sub { |
470 |
my ($w, $now) = @_; |
471 |
|
472 |
use Time::Local (); |
473 |
my (undef, undef, undef, $d, $m, $y) = localtime $now; |
474 |
86400 + Time::Local::timelocal 0, 0, 0, $d, $m, $y |
475 |
}, sub { |
476 |
print "it's midnight or likely shortly after, now\n"; |
477 |
}; |
478 |
|
479 |
The "periodic_ns" variant doesn't start (activate) the newly created |
480 |
watcher. |
481 |
|
482 |
$w->set ($at, $interval, $reschedule_cb) |
483 |
Reconfigures the watcher, see the constructor above for details. Can |
484 |
be called at any time. |
485 |
|
486 |
$w->again |
487 |
Simply stops and starts the watcher again. |
488 |
|
489 |
$time = $w->at |
490 |
Return the time that the watcher is expected to trigger next. |
491 |
|
492 |
SIGNAL WATCHERS - signal me when a signal gets signalled! |
493 |
$w = EV::signal $signal, $callback |
494 |
$w = EV::signal_ns $signal, $callback |
495 |
Call the callback when $signal is received (the signal can be |
496 |
specified by number or by name, just as with "kill" or %SIG). |
497 |
|
498 |
EV will grab the signal for the process (the kernel only allows one |
499 |
component to receive a signal at a time) when you start a signal |
500 |
watcher, and removes it again when you stop it. Perl does the same |
501 |
when you add/remove callbacks to %SIG, so watch out. |
502 |
|
503 |
You can have as many signal watchers per signal as you want. |
504 |
|
505 |
The "signal_ns" variant doesn't start (activate) the newly created |
506 |
watcher. |
507 |
|
508 |
$w->set ($signal) |
509 |
Reconfigures the watcher, see the constructor above for details. Can |
510 |
be called at any time. |
511 |
|
512 |
$current_signum = $w->signal |
513 |
$old_signum = $w->signal ($new_signal) |
514 |
Returns the previously set signal (always as a number not name) and |
515 |
optionally set a new one. |
516 |
|
517 |
CHILD WATCHERS - watch out for process status changes |
518 |
$w = EV::child $pid, $trace, $callback |
519 |
$w = EV::child_ns $pid, $trace, $callback |
520 |
$w = $loop->child ($pid, $trace, $callback) |
521 |
$w = $loop->child_ns ($pid, $trace, $callback) |
522 |
Call the callback when a status change for pid $pid (or any pid if |
523 |
$pid is 0) has been received (a status change happens when the |
524 |
process terminates or is killed, or, when trace is true, |
525 |
additionally when it is stopped or continued). More precisely: when |
526 |
the process receives a "SIGCHLD", EV will fetch the outstanding |
527 |
exit/wait status for all changed/zombie children and call the |
528 |
callback. |
529 |
|
530 |
It is valid (and fully supported) to install a child watcher after a |
531 |
child has exited but before the event loop has started its next |
532 |
iteration (for example, first you "fork", then the new child process |
533 |
might exit, and only then do you install a child watcher in the |
534 |
parent for the new pid). |
535 |
|
536 |
You can access both exit (or tracing) status and pid by using the |
537 |
"rstatus" and "rpid" methods on the watcher object. |
538 |
|
539 |
You can have as many pid watchers per pid as you want, they will all |
540 |
be called. |
541 |
|
542 |
The "child_ns" variant doesn't start (activate) the newly created |
543 |
watcher. |
544 |
|
545 |
$w->set ($pid, $trace) |
546 |
Reconfigures the watcher, see the constructor above for details. Can |
547 |
be called at any time. |
548 |
|
549 |
$current_pid = $w->pid |
550 |
Returns the previously set process id and optionally set a new one. |
551 |
|
552 |
$exit_status = $w->rstatus |
553 |
Return the exit/wait status (as returned by waitpid, see the waitpid |
554 |
entry in perlfunc). |
555 |
|
556 |
$pid = $w->rpid |
557 |
Return the pid of the awaited child (useful when you have installed |
558 |
a watcher for all pids). |
559 |
|
560 |
STAT WATCHERS - did the file attributes just change? |
561 |
$w = EV::stat $path, $interval, $callback |
562 |
$w = EV::stat_ns $path, $interval, $callback |
563 |
$w = $loop->stat ($path, $interval, $callback) |
564 |
$w = $loop->stat_ns ($path, $interval, $callback) |
565 |
Call the callback when a file status change has been detected on |
566 |
$path. The $path does not need to exist, changing from "path exists" |
567 |
to "path does not exist" is a status change like any other. |
568 |
|
569 |
The $interval is a recommended polling interval for systems where |
570 |
OS-supported change notifications don't exist or are not supported. |
571 |
If you use 0 then an unspecified default is used (which is highly |
572 |
recommended!), which is to be expected to be around five seconds |
573 |
usually. |
574 |
|
575 |
This watcher type is not meant for massive numbers of stat watchers, |
576 |
as even with OS-supported change notifications, this can be |
577 |
resource-intensive. |
578 |
|
579 |
The "stat_ns" variant doesn't start (activate) the newly created |
580 |
watcher. |
581 |
|
582 |
... = $w->stat |
583 |
This call is very similar to the perl "stat" built-in: It stats |
584 |
(using "lstat") the path specified in the watcher and sets perls |
585 |
stat cache (as well as EV's idea of the current stat values) to the |
586 |
values found. |
587 |
|
588 |
In scalar context, a boolean is return indicating success or failure |
589 |
of the stat. In list context, the same 13-value list as with stat is |
590 |
returned (except that the blksize and blocks fields are not |
591 |
reliable). |
592 |
|
593 |
In the case of an error, errno is set to "ENOENT" (regardless of the |
594 |
actual error value) and the "nlink" value is forced to zero (if the |
595 |
stat was successful then nlink is guaranteed to be non-zero). |
596 |
|
597 |
See also the next two entries for more info. |
598 |
|
599 |
... = $w->attr |
600 |
Just like "$w->stat", but without the initial stat'ing: this returns |
601 |
the values most recently detected by EV. See the next entry for more |
602 |
info. |
603 |
|
604 |
... = $w->prev |
605 |
Just like "$w->stat", but without the initial stat'ing: this returns |
606 |
the previous set of values, before the change. |
607 |
|
608 |
That is, when the watcher callback is invoked, "$w->prev" will be |
609 |
set to the values found *before* a change was detected, while |
610 |
"$w->attr" returns the values found leading to the change detection. |
611 |
The difference (if any) between "prev" and "attr" is what triggered |
612 |
the callback. |
613 |
|
614 |
If you did something to the filesystem object and do not want to |
615 |
trigger yet another change, you can call "stat" to update EV's idea |
616 |
of what the current attributes are. |
617 |
|
618 |
$w->set ($path, $interval) |
619 |
Reconfigures the watcher, see the constructor above for details. Can |
620 |
be called at any time. |
621 |
|
622 |
$current_path = $w->path |
623 |
$old_path = $w->path ($new_path) |
624 |
Returns the previously set path and optionally set a new one. |
625 |
|
626 |
$current_interval = $w->interval |
627 |
$old_interval = $w->interval ($new_interval) |
628 |
Returns the previously set interval and optionally set a new one. |
629 |
Can be used to query the actual interval used. |
630 |
|
631 |
IDLE WATCHERS - when you've got nothing better to do... |
632 |
$w = EV::idle $callback |
633 |
$w = EV::idle_ns $callback |
634 |
$w = $loop->idle ($callback) |
635 |
$w = $loop->idle_ns ($callback) |
636 |
Call the callback when there are no other pending watchers of the |
637 |
same or higher priority (excluding check, prepare and other idle |
638 |
watchers of the same or lower priority, of course). They are called |
639 |
idle watchers because when the watcher is the highest priority |
640 |
pending event in the process, the process is considered to be idle |
641 |
at that priority. |
642 |
|
643 |
If you want a watcher that is only ever called when *no* other |
644 |
events are outstanding you have to set the priority to "EV::MINPRI". |
645 |
|
646 |
The process will not block as long as any idle watchers are active, |
647 |
and they will be called repeatedly until stopped. |
648 |
|
649 |
For example, if you have idle watchers at priority 0 and 1, and an |
650 |
I/O watcher at priority 0, then the idle watcher at priority 1 and |
651 |
the I/O watcher will always run when ready. Only when the idle |
652 |
watcher at priority 1 is stopped and the I/O watcher at priority 0 |
653 |
is not pending with the 0-priority idle watcher be invoked. |
654 |
|
655 |
The "idle_ns" variant doesn't start (activate) the newly created |
656 |
watcher. |
657 |
|
658 |
PREPARE WATCHERS - customise your event loop! |
659 |
$w = EV::prepare $callback |
660 |
$w = EV::prepare_ns $callback |
661 |
$w = $loop->prepare ($callback) |
662 |
$w = $loop->prepare_ns ($callback) |
663 |
Call the callback just before the process would block. You can still |
664 |
create/modify any watchers at this point. |
665 |
|
666 |
See the EV::check watcher, below, for explanations and an example. |
667 |
|
668 |
The "prepare_ns" variant doesn't start (activate) the newly created |
669 |
watcher. |
670 |
|
671 |
CHECK WATCHERS - customise your event loop even more! |
672 |
$w = EV::check $callback |
673 |
$w = EV::check_ns $callback |
674 |
$w = $loop->check ($callback) |
675 |
$w = $loop->check_ns ($callback) |
676 |
Call the callback just after the process wakes up again (after it |
677 |
has gathered events), but before any other callbacks have been |
678 |
invoked. |
679 |
|
680 |
This is used to integrate other event-based software into the EV |
681 |
mainloop: You register a prepare callback and in there, you create |
682 |
io and timer watchers as required by the other software. Here is a |
683 |
real-world example of integrating Net::SNMP (with some details left |
684 |
out): |
685 |
|
686 |
our @snmp_watcher; |
687 |
|
688 |
our $snmp_prepare = EV::prepare sub { |
689 |
# do nothing unless active |
690 |
$dispatcher->{_event_queue_h} |
691 |
or return; |
692 |
|
693 |
# make the dispatcher handle any outstanding stuff |
694 |
... not shown |
695 |
|
696 |
# create an I/O watcher for each and every socket |
697 |
@snmp_watcher = ( |
698 |
(map { EV::io $_, EV::READ, sub { } } |
699 |
keys %{ $dispatcher->{_descriptors} }), |
700 |
|
701 |
EV::timer +($event->[Net::SNMP::Dispatcher::_ACTIVE] |
702 |
? $event->[Net::SNMP::Dispatcher::_TIME] - EV::now : 0), |
703 |
0, sub { }, |
704 |
); |
705 |
}; |
706 |
|
707 |
The callbacks are irrelevant (and are not even being called), the |
708 |
only purpose of those watchers is to wake up the process as soon as |
709 |
one of those events occurs (socket readable, or timer timed out). |
710 |
The corresponding EV::check watcher will then clean up: |
711 |
|
712 |
our $snmp_check = EV::check sub { |
713 |
# destroy all watchers |
714 |
@snmp_watcher = (); |
715 |
|
716 |
# make the dispatcher handle any new stuff |
717 |
... not shown |
718 |
}; |
719 |
|
720 |
The callbacks of the created watchers will not be called as the |
721 |
watchers are destroyed before this cna happen (remember EV::check |
722 |
gets called first). |
723 |
|
724 |
The "check_ns" variant doesn't start (activate) the newly created |
725 |
watcher. |
726 |
|
727 |
FORK WATCHERS - the audacity to resume the event loop after a fork |
728 |
Fork watchers are called when a "fork ()" was detected. The invocation |
729 |
is done before the event loop blocks next and before "check" watchers |
730 |
are being called, and only in the child after the fork. |
731 |
|
732 |
$w = EV::fork $callback |
733 |
$w = EV::fork_ns $callback |
734 |
$w = $loop->fork ($callback) |
735 |
$w = $loop->fork_ns ($callback) |
736 |
Call the callback before the event loop is resumed in the child |
737 |
process after a fork. |
738 |
|
739 |
The "fork_ns" variant doesn't start (activate) the newly created |
740 |
watcher. |
741 |
|
742 |
EMBED WATCHERS - when one backend isn't enough... |
743 |
This is a rather advanced watcher type that lets you embed one event |
744 |
loop into another (currently only IO events are supported in the |
745 |
embedded loop, other types of watchers might be handled in a delayed or |
746 |
incorrect fashion and must not be used). |
747 |
|
748 |
See the libev documentation at |
749 |
<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#code_ev_embed_code |
750 |
_when_one_backend_> for more details. |
751 |
|
752 |
In short, this watcher is most useful on BSD systems without working |
753 |
kqueue to still be able to handle a large number of sockets: |
754 |
|
755 |
my $socket_loop; |
756 |
|
757 |
# check wether we use SELECT or POLL _and_ KQUEUE is supported |
758 |
if ( |
759 |
(EV::backend & (EV::BACKEND_POLL | EV::BACKEND_SELECT)) |
760 |
&& (EV::supported_backends & EV::embeddable_backends & EV::BACKEND_KQUEUE) |
761 |
) { |
762 |
# use kqueue for sockets |
763 |
$socket_loop = new EV::Loop EV::BACKEND_KQUEUE | EV::FLAG_NOENV; |
764 |
} |
765 |
|
766 |
# use the default loop otherwise |
767 |
$socket_loop ||= EV::default_loop; |
768 |
|
769 |
$w = EV::embed $otherloop, $callback |
770 |
$w = EV::embed_ns $otherloop, $callback |
771 |
$w = $loop->embed ($otherloop, $callback) |
772 |
$w = $loop->embed_ns ($otherloop, $callback) |
773 |
Call the callback when the embedded event loop ($otherloop) has any |
774 |
I/O activity. The $callback should alwas be specified as "undef" in |
775 |
this version of EV, which means the embedded event loop will be |
776 |
managed automatically. |
777 |
|
778 |
The "embed_ns" variant doesn't start (activate) the newly created |
779 |
watcher. |
780 |
|
781 |
ASYNC WATCHERS - how to wake up another event loop |
782 |
Async watchers are provided by EV, but have little use in perl directly, |
783 |
as perl neither supports threads nor direct access to signal handlers or |
784 |
other contexts where they could be of value. |
785 |
|
786 |
It is, however, possible to use them from the XS level. |
787 |
|
788 |
Please see the libev documentation for further details. |
789 |
|
790 |
PERL SIGNALS |
791 |
While Perl signal handling (%SIG) is not affected by EV, the behaviour |
792 |
with EV is as the same as any other C library: Perl-signals will only be |
793 |
handled when Perl runs, which means your signal handler might be invoked |
794 |
only the next time an event callback is invoked. |
795 |
|
796 |
The solution is to use EV signal watchers (see "EV::signal"), which will |
797 |
ensure proper operations with regards to other event watchers. |
798 |
|
799 |
If you cannot do this for whatever reason, you can also force a watcher |
800 |
to be called on every event loop iteration by installing a "EV::check" |
801 |
watcher: |
802 |
|
803 |
my $async_check = EV::check sub { }; |
804 |
|
805 |
This ensures that perl gets into control for a short time to handle any |
806 |
pending signals, and also ensures (slightly) slower overall operation. |
807 |
|
808 |
THREADS |
809 |
Threads are not supported by this module in any way. Perl pseudo-threads |
810 |
is evil stuff and must die. As soon as Perl gains real threads I will |
811 |
work on thread support for it. |
812 |
|
813 |
FORK |
814 |
Most of the "improved" event delivering mechanisms of modern operating |
815 |
systems have quite a few problems with fork(2) (to put it bluntly: it is |
816 |
not supported and usually destructive). Libev makes it possible to work |
817 |
around this by having a function that recreates the kernel state after |
818 |
fork in the child. |
819 |
|
820 |
On non-win32 platforms, this module requires the pthread_atfork |
821 |
functionality to do this automatically for you. This function is quite |
822 |
buggy on most BSDs, though, so YMMV. The overhead for this is quite |
823 |
negligible, because everything the function currently does is set a flag |
824 |
that is checked only when the event loop gets used the next time, so |
825 |
when you do fork but not use EV, the overhead is minimal. |
826 |
|
827 |
On win32, there is no notion of fork so all this doesn't apply, of |
828 |
course. |
829 |
|
830 |
SEE ALSO |
831 |
EV::ADNS (asynchronous DNS), Glib::EV (makes Glib/Gtk2 use EV as event |
832 |
loop), EV::Glib (embed Glib into EV), Coro::EV (efficient coroutines |
833 |
with EV), Net::SNMP::EV (asynchronous SNMP). |
834 |
|
835 |
AUTHOR |
836 |
Marc Lehmann <schmorp@schmorp.de> |
837 |
http://home.schmorp.de/ |
838 |
|