ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/EV/EV.pm
(Generate patch)

Comparing EV/EV.pm (file contents):
Revision 1.14 by root, Wed Oct 31 10:53:53 2007 UTC vs.
Revision 1.62 by root, Fri Dec 7 18:09:38 2007 UTC

1=head1 NAME 1=head1 NAME
2 2
3EV - perl interface to libevent, monkey.org/~provos/libevent/ 3EV - perl interface to libev, a high performance full-featured event loop
4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 use EV; 7 use EV;
8 8
9 # TIMER 9 # TIMERS
10 10
11 my $w = EV::timer 2, 0, sub { 11 my $w = EV::timer 2, 0, sub {
12 warn "is called after 2s"; 12 warn "is called after 2s";
13 }; 13 };
14 14
15 my $w = EV::timer 2, 1, sub { 15 my $w = EV::timer 2, 2, sub {
16 warn "is called roughly every 2s (repeat = 1)"; 16 warn "is called roughly every 2s (repeat = 2)";
17 }; 17 };
18 18
19 undef $w; # destroy event watcher again 19 undef $w; # destroy event watcher again
20 20
21 my $w = EV::timer_abs 0, 60, sub { 21 my $w = EV::periodic 0, 60, 0, sub {
22 warn "is called every minute, on the minute, exactly"; 22 warn "is called every minute, on the minute, exactly";
23 }; 23 };
24 24
25 # IO 25 # IO
26 26
27 my $w = EV::io \*STDIN, EV::READ | EV::PERSIST, sub { 27 my $w = EV::io *STDIN, EV::READ, sub {
28 my ($w, $events) = @_; # all callbacks get the watcher object and event mask 28 my ($w, $revents) = @_; # all callbacks receive the watcher and event mask
29 if ($events & EV::TIMEOUT) {
30 warn "nothing received on stdin for 10 seconds, retrying";
31 } else {
32 warn "stdin is readable, you entered: ", <STDIN>; 29 warn "stdin is readable, you entered: ", <STDIN>;
33 }
34 };
35 $w->timeout (10);
36
37 my $w = EV::timed_io \*STDIN, EV::READ, 30, sub {
38 my ($w, $events) = @_;
39 if ($_[1] & EV::TIMEOUT) {
40 warn "nothing entered within 30 seconds, bye bye.\n";
41 $w->stop;
42 } else {
43 my $line = <STDIN>;
44 warn "you entered something, you again have 30 seconds.\n";
45 }
46 }; 30 };
47 31
48 # SIGNALS 32 # SIGNALS
49 33
50 my $w = EV::signal 'QUIT', sub { 34 my $w = EV::signal 'QUIT', sub {
51 warn "sigquit received\n"; 35 warn "sigquit received\n";
52 }; 36 };
53 37
38 # CHILD/PID STATUS CHANGES
39
54 my $w = EV::signal 3, sub { 40 my $w = EV::child 666, sub {
55 warn "sigquit received (this is GNU/Linux, right?)\n"; 41 my ($w, $revents) = @_;
42 my $status = $w->rstatus;
56 }; 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 };
57 50
58 # MAINLOOP 51 # MAINLOOP
59 EV::dispatch; # loop as long as watchers are active 52 EV::loop; # loop until EV::unloop is called or all watchers stop
60 EV::loop; # the same thing
61 EV::loop EV::LOOP_ONCE; # block until some events could be handles 53 EV::loop EV::LOOP_ONESHOT; # block until at least one event could be handled
62 EV::loop EV::LOOP_NONBLOCK; # check and handle some events, but do not wait 54 EV::loop EV::LOOP_NONBLOCK; # try to handle same events, but do not block
63 55
64=head1 DESCRIPTION 56=head1 DESCRIPTION
65 57
66This module provides an interface to libevent 58This module provides an interface to libev
67(L<http://monkey.org/~provos/libevent/>). You probably should acquaint 59(L<http://software.schmorp.de/pkg/libev.html>). While the documentation
68yourself with its documentation and source code to be able to use this 60below is comprehensive, one might also consult the documentation of libev
69module fully. 61itself (L<http://cvs.schmorp.de/libev/ev.html>) for more subtle details on
70 62watcher semantics or some discussion on the available backends, or how to
71Please note thta this module disables the libevent EPOLL method by 63force a specific backend with C<LIBEV_FLAGS>, or just about in any case
72default, see BUGS, below, if you need to enable it. 64because it has much more detailed information.
73 65
74=cut 66=cut
75 67
76package EV; 68package EV;
77 69
78use strict; 70use strict;
79 71
80BEGIN { 72BEGIN {
81 our $VERSION = '0.03'; 73 our $VERSION = '1.7';
82 use XSLoader; 74 use XSLoader;
83 XSLoader::load "EV", $VERSION; 75 XSLoader::load "EV", $VERSION;
84} 76}
85 77
78@EV::IO::ISA =
79@EV::Timer::ISA =
80@EV::Periodic::ISA =
81@EV::Signal::ISA =
82@EV::Child::ISA =
83@EV::Stat::ISA =
84@EV::Idle::ISA =
85@EV::Prepare::ISA =
86@EV::Check::ISA =
87@EV::Embed::ISA =
88@EV::Fork::ISA =
89 "EV::Watcher";
90
86=head1 BASIC INTERFACE 91=head1 BASIC INTERFACE
87 92
88=over 4 93=over 4
89
90=item $EV::NPRI
91
92How many priority levels are available.
93 94
94=item $EV::DIED 95=item $EV::DIED
95 96
96Must contain a reference to a function that is called when a callback 97Must contain a reference to a function that is called when a callback
97throws an exception (with $@ containing thr error). The default prints an 98throws an exception (with $@ containing thr error). The default prints an
98informative message and continues. 99informative message and continues.
99 100
100If this callback throws an exception it will be silently ignored. 101If this callback throws an exception it will be silently ignored.
101 102
103=item $time = EV::time
104
105Returns the current time in (fractional) seconds since the epoch.
106
102=item $time = EV::now 107=item $time = EV::now
103 108
104Returns the time in (fractional) seconds since the epoch. 109Returns the time the last event loop iteration has been started. This
105 110is the time that (relative) timers are based on, and refering to it is
106=item $version = EV::version 111usually faster then calling EV::time.
107 112
108=item $method = EV::method 113=item $method = EV::method
109 114
110Return version string and event polling method used. 115Returns an integer describing the backend used by libev (EV::METHOD_SELECT
116or EV::METHOD_EPOLL).
111 117
112=item EV::loop $flags # EV::LOOP_ONCE, EV::LOOP_ONESHOT 118=item EV::loop [$flags]
113 119
114=item EV::loopexit $after 120Begin checking for events and calling callbacks. It returns when a
121callback calls EV::unloop.
115 122
116Exit any active loop or dispatch after C<$after> seconds or immediately if 123The $flags argument can be one of the following:
117C<$after> is missing or zero.
118 124
119=item EV::dispatch 125 0 as above
126 EV::LOOP_ONESHOT block at most once (wait, but do not loop)
127 EV::LOOP_NONBLOCK do not block at all (fetch/handle events but do not wait)
120 128
121Same as C<EV::loop 0>. 129=item EV::unloop [$how]
122 130
123=item EV::event $callback 131When called with no arguments or an argument of EV::UNLOOP_ONE, makes the
132innermost call to EV::loop return.
124 133
125Creates a new event watcher waiting for nothing, calling the given callback. 134When called with an argument of EV::UNLOOP_ALL, all calls to EV::loop will return as
135fast as possible.
126 136
137=item $count = EV::loop_count
138
139Return the number of times the event loop has polled for new
140events. Sometiems useful as a generation counter.
141
142=item EV::once $fh_or_undef, $events, $timeout, $cb->($revents)
143
144This function rolls together an I/O and a timer watcher for a single
145one-shot event without the need for managing a watcher object.
146
147If C<$fh_or_undef> is a filehandle or file descriptor, then C<$events>
148must be a bitset containing either C<EV::READ>, C<EV::WRITE> or C<EV::READ
149| EV::WRITE>, indicating the type of I/O event you want to wait for. If
150you do not want to wait for some I/O event, specify C<undef> for
151C<$fh_or_undef> and C<0> for C<$events>).
152
153If timeout is C<undef> or negative, then there will be no
154timeout. Otherwise a EV::timer with this value will be started.
155
156When an error occurs or either the timeout or I/O watcher triggers, then
157the callback will be called with the received event set (in general
158you can expect it to be a combination of C<EV:ERROR>, C<EV::READ>,
159C<EV::WRITE> and C<EV::TIMEOUT>).
160
161EV::once doesn't return anything: the watchers stay active till either
162of them triggers, then they will be stopped and freed, and the callback
163invoked.
164
165=back
166
167=head2 WATCHER OBJECTS
168
169A watcher is an object that gets created to record your interest in some
170event. For instance, if you want to wait for STDIN to become readable, you
171would create an EV::io watcher for that:
172
173 my $watcher = EV::io *STDIN, EV::READ, sub {
174 my ($watcher, $revents) = @_;
175 warn "yeah, STDIN should not be readable without blocking!\n"
176 };
177
178All watchers can be active (waiting for events) or inactive (paused). Only
179active watchers will have their callbacks invoked. All callbacks will be
180called with at least two arguments: the watcher and a bitmask of received
181events.
182
183Each watcher type has its associated bit in revents, so you can use the
184same callback for multiple watchers. The event mask is named after the
185type, i..e. EV::child sets EV::CHILD, EV::prepare sets EV::PREPARE,
186EV::periodic sets EV::PERIODIC and so on, with the exception of I/O events
187(which can set both EV::READ and EV::WRITE bits), and EV::timer (which
188uses EV::TIMEOUT).
189
190In the rare case where one wants to create a watcher but not start it at
191the same time, each constructor has a variant with a trailing C<_ns> in
192its name, e.g. EV::io has a non-starting variant EV::io_ns and so on.
193
194Please note that a watcher will automatically be stopped when the watcher
195object is destroyed, so you I<need> to keep the watcher objects returned by
196the constructors.
197
198Also, all methods changing some aspect of a watcher (->set, ->priority,
199->fh and so on) automatically stop and start it again if it is active,
200which means pending events get lost.
201
202=head2 COMMON WATCHER METHODS
203
204This section lists methods common to all watchers.
205
206=over 4
207
208=item $w->start
209
210Starts a watcher if it isn't active already. Does nothing to an already
211active watcher. By default, all watchers start out in the active state
212(see the description of the C<_ns> variants if you need stopped watchers).
213
214=item $w->stop
215
216Stop a watcher if it is active. Also clear any pending events (events that
217have been received but that didn't yet result in a callback invocation),
218regardless of whether the watcher was active or not.
219
220=item $bool = $w->is_active
221
222Returns true if the watcher is active, false otherwise.
223
224=item $current_data = $w->data
225
226=item $old_data = $w->data ($new_data)
227
228Queries a freely usable data scalar on the watcher and optionally changes
229it. This is a way to associate custom data with a watcher:
230
231 my $w = EV::timer 60, 0, sub {
232 warn $_[0]->data;
233 };
234 $w->data ("print me!");
235
236=item $current_cb = $w->cb
237
238=item $old_cb = $w->cb ($new_cb)
239
240Queries the callback on the watcher and optionally changes it. You can do
241this at any time without the watcher restarting.
242
243=item $current_priority = $w->priority
244
245=item $old_priority = $w->priority ($new_priority)
246
247Queries the priority on the watcher and optionally changes it. Pending
248watchers with higher priority will be invoked first. The valid range of
249priorities lies between EV::MAXPRI (default 2) and EV::MINPRI (default
250-2). If the priority is outside this range it will automatically be
251normalised to the nearest valid priority.
252
253The default priority of any newly-created watcher is 0.
254
255Note that the priority semantics have not yet been fleshed out and are
256subject to almost certain change.
257
258=item $w->trigger ($revents)
259
260Call the callback *now* with the given event mask.
261
262=item $previous_state = $w->keepalive ($bool)
263
264Normally, C<EV::loop> will return when there are no active watchers
265(which is a "deadlock" because no progress can be made anymore). This is
266convinient because it allows you to start your watchers (and your jobs),
267call C<EV::loop> once and when it returns you know that all your jobs are
268finished (or they forgot to register some watchers for their task :).
269
270Sometimes, however, this gets in your way, for example when you the module
271that calls C<EV::loop> (usually the main program) is not the same module
272as a long-living watcher (for example a DNS client module written by
273somebody else even). Then you might want any outstanding requests to be
274handled, but you would not want to keep C<EV::loop> from returning just
275because you happen to have this long-running UDP port watcher.
276
277In this case you can clear the keepalive status, which means that even
278though your watcher is active, it won't keep C<EV::loop> from returning.
279
280The initial value for keepalive is true (enabled), and you cna change it
281any time.
282
283Example: Register an I/O watcher for some UDP socket but do not keep the
284event loop from running just because of that watcher.
285
286 my $udp_socket = ...
287 my $udp_watcher = EV::io $udp_socket, EV::READ, sub { ... };
288 $udp_watcher->keepalive (0);
289
290=back
291
292
293=head2 WATCHER TYPES
294
295Each of the following subsections describes a single watcher type.
296
297=head3 I/O WATCHERS - is this file descriptor readable or writable?
298
299=over 4
300
127=item my $w = EV::io $fileno_or_fh, $eventmask, $callback 301=item $w = EV::io $fileno_or_fh, $eventmask, $callback
128 302
129=item my $w = EV::io_ns $fileno_or_fh, $eventmask, $callback 303=item $w = EV::io_ns $fileno_or_fh, $eventmask, $callback
130 304
131As long as the returned watcher object is alive, call the C<$callback> 305As long as the returned watcher object is alive, call the C<$callback>
132when the events specified in C<$eventmask> happen. Initially, the timeout 306when at least one of events specified in C<$eventmask> occurs.
133is disabled.
134 307
135You can additionall set a timeout to occur on the watcher, but note that
136this timeout will not be reset when you get an I/O event in the EV::PERSIST
137case, and reaching a timeout will always stop the watcher even in the
138EV::PERSIST case.
139
140If you want a timeout to occur only after a specific time of inactivity, set
141a repeating timeout and do NOT use EV::PERSIST.
142
143Eventmask can be one or more of these constants ORed together: 308The $eventmask can be one or more of these constants ORed together:
144 309
145 EV::READ wait until read() wouldn't block anymore 310 EV::READ wait until read() wouldn't block anymore
146 EV::WRITE wait until write() wouldn't block anymore 311 EV::WRITE wait until write() wouldn't block anymore
147 EV::PERSIST stay active after a (non-timeout) event occured
148 312
149The C<io_ns> variant doesn't add/start the newly created watcher. 313The C<io_ns> variant doesn't start (activate) the newly created watcher.
150 314
151=item my $w = EV::timed_io $fileno_or_fh, $eventmask, $timeout, $callback 315=item $w->set ($fileno_or_fh, $eventmask)
152 316
153=item my $w = EV::timed_io_ns $fileno_or_fh, $eventmask, $timeout, $callback 317Reconfigures the watcher, see the constructor above for details. Can be
318called at any time.
154 319
155Same as C<io> and C<io_ns>, but also specifies a timeout (as if there was 320=item $current_fh = $w->fh
156a call to C<< $w->timeout ($timout, 1) >>. The persist flag is not allowed
157and will automatically be cleared. The watcher will be restarted after each event.
158 321
159If the timeout is zero or undef, no timeout will be set, and a normal 322=item $old_fh = $w->fh ($new_fh)
160watcher (with the persist flag set!) will be created.
161 323
162This has the effect of timing out after the specified period of inactivity 324Returns the previously set filehandle and optionally set a new one.
163has happened.
164 325
165Due to the design of libevent, this is also relatively inefficient, having 326=item $current_eventmask = $w->events
166one or two io watchers and a separate timeout watcher that you reset on
167activity (by calling its C<start> method) is usually more efficient.
168 327
328=item $old_eventmask = $w->events ($new_eventmask)
329
330Returns the previously set event mask and optionally set a new one.
331
332=back
333
334
335=head3 TIMER WATCHERS - relative and optionally repeating timeouts
336
337=over 4
338
169=item my $w = EV::timer $after, $repeat, $callback 339=item $w = EV::timer $after, $repeat, $callback
170 340
171=item my $w = EV::timer_ns $after, $repeat, $callback 341=item $w = EV::timer_ns $after, $repeat, $callback
172 342
173Calls the callback after C<$after> seconds. If C<$repeat> is true, the 343Calls the callback after C<$after> seconds (which may be fractional). If
174timer will be restarted after the callback returns. This means that the 344C<$repeat> is non-zero, the timer will be restarted (with the $repeat
175callback would be called roughly every C<$after> seconds, prolonged by the 345value as $after) after the callback returns.
176time the callback takes.
177 346
347This means that the callback would be called roughly after C<$after>
348seconds, and then every C<$repeat> seconds. The timer does his best not
349to drift, but it will not invoke the timer more often then once per event
350loop iteration, and might drift in other cases. If that isn't acceptable,
351look at EV::periodic, which can provide long-term stable timers.
352
353The timer is based on a monotonic clock, that is, if somebody is sitting
354in front of the machine while the timer is running and changes the system
355clock, the timer will nevertheless run (roughly) the same time.
356
178The C<timer_ns> variant doesn't add/start the newly created watcher. 357The C<timer_ns> variant doesn't start (activate) the newly created watcher.
179 358
180=item my $w = EV::timer_abs $at, $interval, $callback 359=item $w->set ($after, $repeat)
181 360
182=item my $w = EV::timer_abs_ns $at, $interval, $callback 361Reconfigures the watcher, see the constructor above for details. Can be called at
362any time.
183 363
184Similar to EV::timer, but the time is given as an absolute point in time 364=item $w->again
185(C<$at>), plus an optional C<$interval>.
186 365
187If the C<$interval> is zero, then the callback will be called at the time 366Similar to the C<start> method, but has special semantics for repeating timers:
188C<$at> if that is in the future, or as soon as possible if its in the
189past. It will not automatically repeat.
190 367
191If the C<$interval> is nonzero, then the watcher will always be scheduled 368If the timer is active and non-repeating, it will be stopped.
192to time out at the next C<$at + integer * $interval> time.
193 369
194This can be used to schedule a callback to run at very regular intervals, 370If the timer is active and repeating, reset the timeout to occur
195as long as the processing time is less then the interval (otherwise 371C<$repeat> seconds after now.
196obviously events will be skipped). 372
373If the timer is inactive and repeating, start it using the repeat value.
374
375Otherwise do nothing.
376
377This behaviour is useful when you have a timeout for some IO
378operation. You create a timer object with the same value for C<$after> and
379C<$repeat>, and then, in the read/write watcher, run the C<again> method
380on the timeout.
381
382=back
383
384
385=head3 PERIODIC WATCHERS - to cron or not to cron?
386
387=over 4
388
389=item $w = EV::periodic $at, $interval, $reschedule_cb, $callback
390
391=item $w = EV::periodic_ns $at, $interval, $reschedule_cb, $callback
392
393Similar to EV::timer, but is not based on relative timeouts but on
394absolute times. Apart from creating "simple" timers that trigger "at" the
395specified time, it can also be used for non-drifting absolute timers and
396more complex, cron-like, setups that are not adversely affected by time
397jumps (i.e. when the system clock is changed by explicit date -s or other
398means such as ntpd). It is also the most complex watcher type in EV.
399
400It has three distinct "modes":
401
402=over 4
403
404=item * absolute timer ($interval = $reschedule_cb = 0)
405
406This time simply fires at the wallclock time C<$at> and doesn't repeat. It
407will not adjust when a time jump occurs, that is, if it is to be run
408at January 1st 2011 then it will run when the system time reaches or
409surpasses this time.
410
411=item * non-repeating interval timer ($interval > 0, $reschedule_cb = 0)
412
413In this mode the watcher will always be scheduled to time out at the
414next C<$at + N * $interval> time (for some integer N) and then repeat,
415regardless of any time jumps.
416
417This can be used to create timers that do not drift with respect to system
418time:
419
420 my $hourly = EV::periodic 0, 3600, 0, sub { print "once/hour\n" };
421
422That doesn't mean there will always be 3600 seconds in between triggers,
423but only that the the clalback will be called when the system time shows a
424full hour (UTC).
197 425
198Another way to think about it (for the mathematically inclined) is that 426Another way to think about it (for the mathematically inclined) is that
199C<timer_abs> will try to tun the callback at the next possible time where 427EV::periodic will try to run the callback in this mode at the next
200C<$time = $at (mod $interval)>, regardless of any time jumps. 428possible time where C<$time = $at (mod $interval)>, regardless of any time
429jumps.
201 430
431=item * manual reschedule mode ($reschedule_cb = coderef)
432
433In this mode $interval and $at are both being ignored. Instead, each
434time the periodic watcher gets scheduled, the reschedule callback
435($reschedule_cb) will be called with the watcher as first, and the current
436time as second argument.
437
438I<This callback MUST NOT stop or destroy this or any other periodic
439watcher, ever>. If you need to stop it, return 1e30 and stop it
440afterwards.
441
442It must return the next time to trigger, based on the passed time value
443(that is, the lowest time value larger than to the second argument). It
444will usually be called just before the callback will be triggered, but
445might be called at other times, too.
446
447This can be used to create very complex timers, such as a timer that
448triggers on each midnight, local time (actually 24 hours after the last
449midnight, to keep the example simple. If you know a way to do it correctly
450in about the same space (without requiring elaborate modules), drop me a
451note :):
452
453 my $daily = EV::periodic 0, 0, sub {
454 my ($w, $now) = @_;
455
456 use Time::Local ();
457 my (undef, undef, undef, $d, $m, $y) = localtime $now;
458 86400 + Time::Local::timelocal 0, 0, 0, $d, $m, $y
459 }, sub {
460 print "it's midnight or likely shortly after, now\n";
461 };
462
463=back
464
202The C<timer_abs_ns> variant doesn't add/start the newly created watcher. 465The C<periodic_ns> variant doesn't start (activate) the newly created watcher.
203 466
467=item $w->set ($at, $interval, $reschedule_cb)
468
469Reconfigures the watcher, see the constructor above for details. Can be called at
470any time.
471
472=item $w->again
473
474Simply stops and starts the watcher again.
475
476=back
477
478
479=head3 SIGNAL WATCHERS - signal me when a signal gets signalled!
480
481=over 4
482
204=item my $w = EV::signal $signal, $callback 483=item $w = EV::signal $signal, $callback
205 484
206=item my $w = EV::signal_ns $signal, $callback 485=item $w = EV::signal_ns $signal, $callback
207 486
208Call the callback when $signal is received (the signal can be specified 487Call the callback when $signal is received (the signal can be specified by
209by number or by name, just as with kill or %SIG). Signal watchers are 488number or by name, just as with C<kill> or C<%SIG>).
210persistent no natter what.
211 489
212EV will grab the signal for the process (the kernel only allows one 490EV will grab the signal for the process (the kernel only allows one
213component to receive signals) when you start a signal watcher, and 491component to receive a signal at a time) when you start a signal watcher,
214removes it again when you stop it. Pelr does the same when you add/remove 492and removes it again when you stop it. Perl does the same when you
215callbacks to %SIG, so watch out. 493add/remove callbacks to C<%SIG>, so watch out.
216 494
217Unfortunately, only one handler can be registered per signal. Screw 495You can have as many signal watchers per signal as you want.
218libevent.
219 496
220The C<signal_ns> variant doesn't add/start the newly created watcher. 497The C<signal_ns> variant doesn't start (activate) the newly created watcher.
221 498
222=back 499=item $w->set ($signal)
223 500
224=head1 THE EV::Event CLASS 501Reconfigures the watcher, see the constructor above for details. Can be
502called at any time.
225 503
226All EV functions creating an event watcher (designated by C<my $w =>
227above) support the following methods on the returned watcher object:
228
229=over 4
230
231=item $w->add ($timeout)
232
233Stops and (re-)starts the event watcher, setting the optional timeout to
234the given value, or clearing the timeout if none is given.
235
236=item $w->start
237
238Stops and (re-)starts the event watcher without touching the timeout.
239
240=item $w->del
241
242=item $w->stop
243
244Stop the event watcher if it was started.
245
246=item $current_callback = $w->cb
247
248=item $old_callback = $w->cb ($new_callback)
249
250Return the previously set callback and optionally set a new one.
251
252=item $current_fh = $w->fh
253
254=item $old_fh = $w->fh ($new_fh)
255
256Returns the previously set filehandle and optionally set a new one (also
257clears the EV::SIGNAL flag when setting a filehandle).
258
259=item $current_signal = $w->signal 504=item $current_signum = $w->signal
260 505
261=item $old_signal = $w->signal ($new_signal) 506=item $old_signum = $w->signal ($new_signal)
262 507
508Returns the previously set signal (always as a number not name) and
509optionally set a new one.
510
511=back
512
513
514=head3 CHILD WATCHERS - watch out for process status changes
515
516=over 4
517
518=item $w = EV::child $pid, $callback
519
520=item $w = EV::child_ns $pid, $callback
521
522Call the callback when a status change for pid C<$pid> (or any pid if
523C<$pid> is 0) has been received. More precisely: when the process receives
524a C<SIGCHLD>, EV will fetch the outstanding exit/wait status for all
525changed/zombie children and call the callback.
526
527It is valid (and fully supported) to install a child watcher after a child
528has exited but before the event loop has started its next iteration (for
529example, first you C<fork>, then the new child process might exit, and
530only then do you install a child watcher in the parent for the new pid).
531
532You can access both exit (or tracing) status and pid by using the
533C<rstatus> and C<rpid> methods on the watcher object.
534
535You can have as many pid watchers per pid as you want, they will all be
536called.
537
538The C<child_ns> variant doesn't start (activate) the newly created watcher.
539
540=item $w->set ($pid)
541
542Reconfigures the watcher, see the constructor above for details. Can be called at
543any time.
544
545=item $current_pid = $w->pid
546
547=item $old_pid = $w->pid ($new_pid)
548
263Returns the previously set signal number and optionally set a new one (also sets 549Returns the previously set process id and optionally set a new one.
264the EV::SIGNAL flag when setting a signal).
265 550
266=item $current_eventmask = $w->events 551=item $exit_status = $w->rstatus
267 552
268=item $old_eventmask = $w->events ($new_eventmask) 553Return the exit/wait status (as returned by waitpid, see the waitpid entry
554in perlfunc).
269 555
556=item $pid = $w->rpid
557
558Return the pid of the awaited child (useful when you have installed a
559watcher for all pids).
560
561=back
562
563
564=head3 STAT WATCHERS - did the file attributes just change?
565
566=over 4
567
568=item $w = EV::stat $path, $interval, $callback
569
570=item $w = EV::stat_ns $path, $interval, $callback
571
572Call the callback when a file status change has been detected on
573C<$path>. The C<$path> does not need to exist, changing from "path exists"
574to "path does not exist" is a status change like any other.
575
576The C<$interval> is a recommended polling interval for systems where
577OS-supported change notifications don't exist or are not supported. If
578you use C<0> then an unspecified default is used (which is highly
579recommended!), which is to be expected to be around five seconds usually.
580
581This watcher type is not meant for massive numbers of stat watchers,
582as even with OS-supported change notifications, this can be
583resource-intensive.
584
585The C<stat_ns> variant doesn't start (activate) the newly created watcher.
586
587=item ... = $w->stat
588
589This call is very similar to the perl C<stat> built-in: It stats (using
590C<lstat>) the path specified in the watcher and sets perls stat cache (as
591well as EV's idea of the current stat values) to the values found.
592
593In scalar context, a boolean is return indicating success or failure of
594the stat. In list context, the same 13-value list as with stat is returned
595(except that the blksize and blocks fields are not reliable).
596
597In the case of an error, errno is set to C<ENOENT> (regardless of the
598actual error value) and the C<nlink> value is forced to zero (if the stat
599was successful then nlink is guaranteed to be non-zero).
600
601See also the next two entries for more info.
602
603=item ... = $w->attr
604
605Just like C<< $w->stat >>, but without the initial stat'ing: this returns
606the values most recently detected by EV. See the next entry for more info.
607
608=item ... = $w->prev
609
610Just like C<< $w->stat >>, but without the initial stat'ing: this returns
611the previous set of values, before the change.
612
613That is, when the watcher callback is invoked, C<< $w->prev >> will be set
614to the values found I<before> a change was detected, while C<< $w->attr >>
615returns the values found leading to the change detection. The difference (if any)
616between C<prev> and C<attr> is what triggered the callback.
617
618If you did something to the filesystem object and do not want to trigger
619yet another change, you can call C<stat> to update EV's idea of what the
620current attributes are.
621
622=item $w->set ($path, $interval)
623
624Reconfigures the watcher, see the constructor above for details. Can be
625called at any time.
626
627=item $current_path = $w->path
628
629=item $old_path = $w->path ($new_path)
630
270Returns the previously set event mask and optionally set a new one. 631Returns the previously set path and optionally set a new one.
271 632
272=item $w->timeout ($after, $repeat) 633=item $current_interval = $w->interval
273 634
274Resets the timeout (see C<EV::timer> for details). 635=item $old_interval = $w->interval ($new_interval)
275 636
276=item $w->timeout_abs ($at, $interval) 637Returns the previously set interval and optionally set a new one. Can be
638used to query the actual interval used.
277 639
278Resets the timeout (see C<EV::timer_abs> for details).
279
280=item $w->priority_set ($priority)
281
282Set the priority of the watcher to C<$priority> (0 <= $priority < $EV::NPRI).
283
284=back 640=back
641
642
643=head3 IDLE WATCHERS - when you've got nothing better to do...
644
645=over 4
646
647=item $w = EV::idle $callback
648
649=item $w = EV::idle_ns $callback
650
651Call the callback when there are no other pending watchers of the same or
652higher priority (excluding check, prepare and other idle watchers of the
653same or lower priority, of course). They are called idle watchers because
654when the watcher is the highest priority pending event in the process, the
655process is considered to be idle at that priority.
656
657If you want a watcher that is only ever called when I<no> other events are
658outstanding you have to set the priority to C<EV::MINPRI>.
659
660The process will not block as long as any idle watchers are active, and
661they will be called repeatedly until stopped.
662
663For example, if you have idle watchers at priority C<0> and C<1>, and
664an I/O watcher at priority C<0>, then the idle watcher at priority C<1>
665and the I/O watcher will always run when ready. Only when the idle watcher
666at priority C<1> is stopped and the I/O watcher at priority C<0> is not
667pending with the C<0>-priority idle watcher be invoked.
668
669The C<idle_ns> variant doesn't start (activate) the newly created watcher.
670
671=back
672
673
674=head3 PREPARE WATCHERS - customise your event loop!
675
676=over 4
677
678=item $w = EV::prepare $callback
679
680=item $w = EV::prepare_ns $callback
681
682Call the callback just before the process would block. You can still
683create/modify any watchers at this point.
684
685See the EV::check watcher, below, for explanations and an example.
686
687The C<prepare_ns> variant doesn't start (activate) the newly created watcher.
688
689=back
690
691
692=head3 CHECK WATCHERS - customise your event loop even more!
693
694=over 4
695
696=item $w = EV::check $callback
697
698=item $w = EV::check_ns $callback
699
700Call the callback just after the process wakes up again (after it has
701gathered events), but before any other callbacks have been invoked.
702
703This is used to integrate other event-based software into the EV
704mainloop: You register a prepare callback and in there, you create io and
705timer watchers as required by the other software. Here is a real-world
706example of integrating Net::SNMP (with some details left out):
707
708 our @snmp_watcher;
709
710 our $snmp_prepare = EV::prepare sub {
711 # do nothing unless active
712 $dispatcher->{_event_queue_h}
713 or return;
714
715 # make the dispatcher handle any outstanding stuff
716 ... not shown
717
718 # create an I/O watcher for each and every socket
719 @snmp_watcher = (
720 (map { EV::io $_, EV::READ, sub { } }
721 keys %{ $dispatcher->{_descriptors} }),
722
723 EV::timer +($event->[Net::SNMP::Dispatcher::_ACTIVE]
724 ? $event->[Net::SNMP::Dispatcher::_TIME] - EV::now : 0),
725 0, sub { },
726 );
727 };
728
729The callbacks are irrelevant (and are not even being called), the
730only purpose of those watchers is to wake up the process as soon as
731one of those events occurs (socket readable, or timer timed out). The
732corresponding EV::check watcher will then clean up:
733
734 our $snmp_check = EV::check sub {
735 # destroy all watchers
736 @snmp_watcher = ();
737
738 # make the dispatcher handle any new stuff
739 ... not shown
740 };
741
742The callbacks of the created watchers will not be called as the watchers
743are destroyed before this cna happen (remember EV::check gets called
744first).
745
746The C<check_ns> variant doesn't start (activate) the newly created watcher.
747
748=back
749
750
751=head3 FORK WATCHERS - the audacity to resume the event loop after a fork
752
753Fork watchers are called when a C<fork ()> was detected. The invocation
754is done before the event loop blocks next and before C<check> watchers
755are being called, and only in the child after the fork.
756
757=over 4
758
759=item $w = EV::fork $callback
760
761=item $w = EV::fork_ns $callback
762
763Call the callback before the event loop is resumed in the child process
764after a fork.
765
766The C<fork_ns> variant doesn't start (activate) the newly created watcher.
767
768=back
769
770
771=head1 PERL SIGNALS
772
773While Perl signal handling (C<%SIG>) is not affected by EV, the behaviour
774with EV is as the same as any other C library: Perl-signals will only be
775handled when Perl runs, which means your signal handler might be invoked
776only the next time an event callback is invoked.
777
778The solution is to use EV signal watchers (see C<EV::signal>), which will
779ensure proper operations with regards to other event watchers.
780
781If you cannot do this for whatever reason, you can also force a watcher
782to be called on every event loop iteration by installing a C<EV::check>
783watcher:
784
785 my $async_check = EV::check sub { };
786
787This ensures that perl shortly gets into control for a short time, and
788also ensures slower overall operation.
285 789
286=head1 THREADS 790=head1 THREADS
287 791
288Threads are not supported by this in any way. Perl pseudo-threads is evil 792Threads are not supported by this module in any way. Perl pseudo-threads
289and must die. 793is evil stuff and must die. As soon as Perl gains real threads I will work
794on thread support for it.
290 795
291=head1 BUGS 796=head1 FORK
292 797
293Lots. Libevent itself isn't well tested and rather buggy, and this module 798Most of the "improved" event delivering mechanisms of modern operating
294is quite new at the moment. 799systems have quite a few problems with fork(2) (to put it bluntly: it is
800not supported and usually destructive). Libev makes it possible to work
801around this by having a function that recreates the kernel state after
802fork in the child.
295 803
296Please note that the epoll method is not, in general, reliable in programs 804On non-win32 platforms, this module requires the pthread_atfork
297that use fork (even if no libveent calls are being made in the forked 805functionality to do this automatically for you. This function is quite
298process). If your program behaves erratically, try setting the environment 806buggy on most BSDs, though, so YMMV. The overhead for this is quite
299variable C<EVENT_NOEPOLL> first when running the program. 807negligible, because everything the function currently does is set a flag
808that is checked only when the event loop gets used the next time, so when
809you do fork but not use EV, the overhead is minimal.
300 810
301In general, if you fork, then you can only use the EV module in one of the 811On win32, there is no notion of fork so all this doesn't apply, of course.
302children.
303 812
304=cut 813=cut
305 814
306our $DIED = sub { 815our $DIED = sub {
307 warn "EV: error in callback (ignoring): $@"; 816 warn "EV: error in callback (ignoring): $@";
308}; 817};
309 818
310init; 819default_loop
311 820 or die 'EV: cannot initialise libev backend. bad $ENV{LIBEV_METHODS}?';
312push @AnyEvent::REGISTRY, [EV => "EV::AnyEvent"];
313 821
3141; 8221;
315 823
316=head1 SEE ALSO 824=head1 SEE ALSO
317 825
318 L<EV::DNS>, L<event(3)>, L<event.h>, L<evdns.h>. 826L<EV::ADNS> (asynchronous dns), L<Glib::EV> (makes Glib/Gtk2 use EV as
319 L<EV::AnyEvent>. 827event loop), L<Coro::EV> (efficient coroutines with EV).
320 828
321=head1 AUTHOR 829=head1 AUTHOR
322 830
323 Marc Lehmann <schmorp@schmorp.de> 831 Marc Lehmann <schmorp@schmorp.de>
324 http://home.schmorp.de/ 832 http://home.schmorp.de/

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines