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

Comparing EV/EV.pm (file contents):
Revision 1.20 by root, Thu Nov 1 17:17:32 2007 UTC vs.
Revision 1.87 by root, Tue Apr 15 04:41:57 2008 UTC

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::periodic 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, sub { 27 my $w = EV::io *STDIN, EV::READ, sub {
28 my ($w, $revents) = @_; # all callbacks get the watcher object and event mask 28 my ($w, $revents) = @_; # all callbacks receive the watcher and event mask
29 warn "stdin is readable, you entered: ", <STDIN>; 29 warn "stdin is readable, you entered: ", <STDIN>;
30 }; 30 };
31 31
32 # SIGNALS 32 # SIGNALS
33 33
34 my $w = EV::signal 'QUIT', sub { 34 my $w = EV::signal 'QUIT', sub {
35 warn "sigquit received\n"; 35 warn "sigquit received\n";
36 }; 36 };
37 37
38 my $w = EV::signal 3, sub { 38 # CHILD/PID STATUS CHANGES
39 warn "sigquit received (this is GNU/Linux, right?)\n"; 39
40 my $w = EV::child 666, 0, sub {
41 my ($w, $revents) = @_;
42 my $status = $w->rstatus;
40 }; 43 };
41 44
42 # CHILD/PID STATUS CHANGES 45 # STAT CHANGES
43 46 my $w = EV::stat "/etc/passwd", 10, sub {
44 my $w = EV::child 666, sub {
45 my ($w, $revents, $status) = @_; 47 my ($w, $revents) = @_;
48 warn $w->path, " has changed somehow.\n";
46 }; 49 };
47 50
48 # MAINLOOP 51 # MAINLOOP
49 EV::loop; # loop until EV::loop_done is called 52 EV::loop; # loop until EV::unloop is called or all watchers stop
50 EV::loop EV::LOOP_ONESHOT; # block until at least one event could be handled 53 EV::loop EV::LOOP_ONESHOT; # block until at least one event could be handled
51 EV::loop EV::LOOP_NONBLOCK; # try to handle same events, but do not block 54 EV::loop EV::LOOP_NONBLOCK; # try to handle same events, but do not block
52 55
53=head1 DESCRIPTION 56=head1 DESCRIPTION
54 57
55This module provides an interface to libev 58This module provides an interface to libev
56(L<http://software.schmorp.de/pkg/libev.html>). 59(L<http://software.schmorp.de/pkg/libev.html>). While the documentation
60below is comprehensive, one might also consult the documentation of libev
61itself (L<http://cvs.schmorp.de/libev/ev.html>) for more subtle details on
62watcher semantics or some discussion on the available backends, or how to
63force a specific backend with C<LIBEV_FLAGS>, or just about in any case
64because it has much more detailed information.
57 65
58=cut 66=cut
59 67
60package EV; 68package EV;
61 69
62use strict; 70use strict;
63 71
64BEGIN { 72BEGIN {
65 our $VERSION = '0.1'; 73 our $VERSION = '3.2';
66 use XSLoader; 74 use XSLoader;
67 XSLoader::load "EV", $VERSION; 75 XSLoader::load "EV", $VERSION;
68} 76}
69 77
70@EV::Io::ISA = 78@EV::IO::ISA =
71@EV::Timer::ISA = 79@EV::Timer::ISA =
72@EV::Periodic::ISA = 80@EV::Periodic::ISA =
73@EV::Signal::ISA = 81@EV::Signal::ISA =
82@EV::Child::ISA =
83@EV::Stat::ISA =
74@EV::Idle::ISA = 84@EV::Idle::ISA =
75@EV::Prepare::ISA = 85@EV::Prepare::ISA =
76@EV::Check::ISA = 86@EV::Check::ISA =
77@EV::Child::ISA = "EV::Watcher"; 87@EV::Embed::ISA =
88@EV::Fork::ISA =
89@EV::Async::ISA =
90 "EV::Watcher";
91
92@EV::Loop::Default::ISA = "EV::Loop";
93
94=head1 EVENT LOOPS
95
96EV supports multiple event loops: There is a single "default event loop"
97that can handle everything including signals and child watchers, and any
98number of "dynamic event loops" that can use different backends (with
99various limitations), but no child and signal watchers.
100
101You do not have to do anything to create the default event loop: When
102the module is loaded a suitable backend is selected on the premise of
103selecting a working backend (which for example rules out kqueue on most
104BSDs). Modules should, unless they have "special needs" always use the
105default loop as this is fastest (perl-wise), best supported by other
106modules (e.g. AnyEvent or Coro) and most portable event loop.
107
108For specific programs you can create additional event loops dynamically.
109
110=over 4
111
112=item $loop = new EV::loop [$flags]
113
114Create a new event loop as per the specified flags. Please refer to the
115C<ev_loop_new ()> function description in the libev documentation
116(L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#GLOBAL_FUNCTIONS>)
117for more info.
118
119The loop will automatically be destroyed when it is no longer referenced
120by any watcher and the loop object goes out of scope.
121
122Using C<EV::FLAG_FORKCHECK> is recommended, as only the default event loop
123is protected by this module.
124
125=item $loop->loop_fork
126
127Must be called after a fork in the child, before entering or continuing
128the event loop. An alternative is to use C<EV::FLAG_FORKCHECK> which calls
129this fucntion automatically, at some performance loss (refer to the libev
130documentation).
131
132=item $loop = EV::default_loop [$flags]
133
134Return the default loop (which is a singleton object).
135
136=back
137
78 138
79=head1 BASIC INTERFACE 139=head1 BASIC INTERFACE
80 140
81=over 4 141=over 4
82 142
83=item $EV::DIED 143=item $EV::DIED
84 144
85Must contain a reference to a function that is called when a callback 145Must contain a reference to a function that is called when a callback
86throws an exception (with $@ containing thr error). The default prints an 146throws an exception (with $@ containing the error). The default prints an
87informative message and continues. 147informative message and continues.
88 148
89If this callback throws an exception it will be silently ignored. 149If this callback throws an exception it will be silently ignored.
90 150
151=item $flags = EV::supported_backends
152
153=item $flags = EV::recommended_backends
154
155=item $flags = EV::embeddable_backends
156
157Returns the set (see C<EV::BACKEND_*> flags) of backends supported by this
158instance of EV, the set of recommended backends (supposed to be good) for
159this platform and the set of embeddable backends (see EMBED WATCHERS).
160
161=item EV::sleep $seconds
162
163Block the process for the given number of (fractional) seconds.
164
91=item $time = EV::time 165=item $time = EV::time
92 166
93Returns the current time in (fractional) seconds since the epoch. 167Returns the current time in (fractional) seconds since the epoch.
94 168
95=item $time = EV::now 169=item $time = EV::now
170
171=item $time = $loop->now
96 172
97Returns the time the last event loop iteration has been started. This 173Returns the time the last event loop iteration has been started. This
98is the time that (relative) timers are based on, and refering to it is 174is the time that (relative) timers are based on, and refering to it is
99usually faster then calling EV::time. 175usually faster then calling EV::time.
100 176
101=item $method = EV::ev_method 177=item $backend = EV::backend
178
179=item $backend = $loop->backend
102 180
103Returns an integer describing the backend used by libev (EV::METHOD_SELECT 181Returns an integer describing the backend used by libev (EV::METHOD_SELECT
104or EV::METHOD_EPOLL). 182or EV::METHOD_EPOLL).
105 183
106=item EV::loop [$flags] 184=item EV::loop [$flags]
107 185
186=item $loop->loop ([$flags])
187
108Begin checking for events and calling callbacks. It returns when a 188Begin checking for events and calling callbacks. It returns when a
109callback calls EV::loop_done. 189callback calls EV::unloop.
110 190
111The $flags argument can be one of the following: 191The $flags argument can be one of the following:
112 192
113 0 as above 193 0 as above
114 EV::LOOP_ONESHOT block at most once (wait, but do not loop) 194 EV::LOOP_ONESHOT block at most once (wait, but do not loop)
115 EV::LOOP_NONBLOCK do not block at all (fetch/handle events but do not wait) 195 EV::LOOP_NONBLOCK do not block at all (fetch/handle events but do not wait)
116 196
117=item EV::loop_done [$how] 197=item EV::unloop [$how]
118 198
199=item $loop->unloop ([$how])
200
119When called with no arguments or an argument of 1, makes the innermost 201When called with no arguments or an argument of EV::UNLOOP_ONE, makes the
120call to EV::loop return. 202innermost call to EV::loop return.
121 203
122When called with an agrument of 2, all calls to EV::loop will return as 204When called with an argument of EV::UNLOOP_ALL, all calls to EV::loop will return as
123fast as possible. 205fast as possible.
124 206
125=back 207=item $count = EV::loop_count
126 208
127=head2 WATCHER 209=item $count = $loop->loop_count
210
211Return the number of times the event loop has polled for new
212events. Sometiems useful as a generation counter.
213
214=item EV::once $fh_or_undef, $events, $timeout, $cb->($revents)
215
216=item $loop->once ($fh_or_undef, $events, $timeout, $cb->($revents))
217
218This function rolls together an I/O and a timer watcher for a single
219one-shot event without the need for managing a watcher object.
220
221If C<$fh_or_undef> is a filehandle or file descriptor, then C<$events>
222must be a bitset containing either C<EV::READ>, C<EV::WRITE> or C<EV::READ
223| EV::WRITE>, indicating the type of I/O event you want to wait for. If
224you do not want to wait for some I/O event, specify C<undef> for
225C<$fh_or_undef> and C<0> for C<$events>).
226
227If timeout is C<undef> or negative, then there will be no
228timeout. Otherwise a EV::timer with this value will be started.
229
230When an error occurs or either the timeout or I/O watcher triggers, then
231the callback will be called with the received event set (in general
232you can expect it to be a combination of C<EV::ERROR>, C<EV::READ>,
233C<EV::WRITE> and C<EV::TIMEOUT>).
234
235EV::once doesn't return anything: the watchers stay active till either
236of them triggers, then they will be stopped and freed, and the callback
237invoked.
238
239=item EV::feed_fd_event ($fd, $revents)
240
241=item $loop->feed_fd_event ($fd, $revents)
242
243Feed an event on a file descriptor into EV. EV will react to this call as
244if the readyness notifications specified by C<$revents> (a combination of
245C<EV::READ> and C<EV::WRITE>) happened on the file descriptor C<$fd>.
246
247=item EV::feed_signal_event ($signal)
248
249Feed a signal event into EV. EV will react to this call as if the signal
250specified by C<$signal> had occured.
251
252=item EV::set_io_collect_interval $time
253
254=item $loop->set_io_collect_interval ($time)
255
256=item EV::set_timeout_collect_interval $time
257
258=item $loop->set_timeout_collect_interval ($time)
259
260These advanced functions set the minimum block interval when polling for I/O events and the minimum
261wait interval for timer events. See the libev documentation at
262L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP> for
263a more detailed discussion.
264
265=back
266
267
268=head1 WATCHER OBJECTS
128 269
129A watcher is an object that gets created to record your interest in some 270A watcher is an object that gets created to record your interest in some
130event. For instance, if you want to wait for STDIN to become readable, you 271event. For instance, if you want to wait for STDIN to become readable, you
131would create an EV::io watcher for that: 272would create an EV::io watcher for that:
132 273
133 my $watcher = EV::io *STDIN, EV::READ, sub { 274 my $watcher = EV::io *STDIN, EV::READ, sub {
134 my ($watcher, $revents) = @_; 275 my ($watcher, $revents) = @_;
135 warn "yeah, STDIN should not be readable without blocking!\n" 276 warn "yeah, STDIN should now be readable without blocking!\n"
136 }; 277 };
137 278
138All watchers can be active (waiting for events) or inactive (paused). Only 279All watchers can be active (waiting for events) or inactive (paused). Only
139active watchers will have their callbacks invoked. All callbacks will be 280active watchers will have their callbacks invoked. All callbacks will be
140called with at least two arguments: the watcher and a bitmask of received 281called with at least two arguments: the watcher and a bitmask of received
141events. 282events.
142 283
143Each watcher type has its associated bit in revents, so you can use the 284Each watcher type has its associated bit in revents, so you can use the
144same callback for multiple watchers. The event mask is named after the 285same callback for multiple watchers. The event mask is named after the
145type, i..e. EV::child sets EV::CHILD, EV::prepare sets EV::PREPARE, 286type, i..e. EV::child sets EV::CHILD, EV::prepare sets EV::PREPARE,
146EV::periodic sets EV::PERIODIC and so on, with the exception of IO events 287EV::periodic sets EV::PERIODIC and so on, with the exception of I/O events
147(which can set both EV::READ and EV::WRITE bits), and EV::timer (which 288(which can set both EV::READ and EV::WRITE bits), and EV::timer (which
148uses EV::TIMEOUT). 289uses EV::TIMEOUT).
149 290
150In the rare case where one wants to create a watcher but not start it at 291In the rare case where one wants to create a watcher but not start it at
151the same time, each constructor has a variant with a trailing C<_ns> in 292the same time, each constructor has a variant with a trailing C<_ns> in
152its name, e.g. EV::io has a non-starting variant EV::io_ns and so on. 293its name, e.g. EV::io has a non-starting variant EV::io_ns and so on.
153 294
154Please note that a watcher will automatically be stopped when the watcher 295Please note that a watcher will automatically be stopped when the watcher
155object is returned, so you I<need> to keep the watcher objects returned by 296object is destroyed, so you I<need> to keep the watcher objects returned by
156the constructors. 297the constructors.
157 298
158=head2 WATCHER TYPES 299Also, all methods changing some aspect of a watcher (->set, ->priority,
300->fh and so on) automatically stop and start it again if it is active,
301which means pending events get lost.
159 302
160Now lets move to the existing watcher types and asociated methods. 303=head2 COMMON WATCHER METHODS
161 304
162The following methods are available for all watchers. Then followes a 305This section lists methods common to all watchers.
163description of each watcher constructor (EV::io, EV::timer, EV::periodic,
164EV::signal, EV::child, EV::idle, EV::prepare and EV::check), followed by
165any type-specific methods (if any).
166 306
167=over 4 307=over 4
168 308
169=item $w->start 309=item $w->start
170 310
174 314
175=item $w->stop 315=item $w->stop
176 316
177Stop a watcher if it is active. Also clear any pending events (events that 317Stop a watcher if it is active. Also clear any pending events (events that
178have been received but that didn't yet result in a callback invocation), 318have been received but that didn't yet result in a callback invocation),
179regardless of wether the watcher was active or not. 319regardless of whether the watcher was active or not.
180 320
181=item $bool = $w->is_active 321=item $bool = $w->is_active
182 322
183Returns true if the watcher is active, false otherwise. 323Returns true if the watcher is active, false otherwise.
184 324
325=item $current_data = $w->data
326
327=item $old_data = $w->data ($new_data)
328
329Queries a freely usable data scalar on the watcher and optionally changes
330it. This is a way to associate custom data with a watcher:
331
332 my $w = EV::timer 60, 0, sub {
333 warn $_[0]->data;
334 };
335 $w->data ("print me!");
336
185=item $current_cb = $w->cb 337=item $current_cb = $w->cb
186 338
187=item $old_cb = $w->cb ($new_cb) 339=item $old_cb = $w->cb ($new_cb)
188 340
189Queries the callback on the watcher and optionally changes it. You cna do 341Queries the callback on the watcher and optionally changes it. You can do
190this at any time. 342this at any time without the watcher restarting.
191 343
344=item $current_priority = $w->priority
345
346=item $old_priority = $w->priority ($new_priority)
347
348Queries the priority on the watcher and optionally changes it. Pending
349watchers with higher priority will be invoked first. The valid range of
350priorities lies between EV::MAXPRI (default 2) and EV::MINPRI (default
351-2). If the priority is outside this range it will automatically be
352normalised to the nearest valid priority.
353
354The default priority of any newly-created watcher is 0.
355
356Note that the priority semantics have not yet been fleshed out and are
357subject to almost certain change.
358
192=item $w->trigger ($revents) 359=item $w->invoke ($revents)
193 360
194Call the callback *now* with the given event mask. 361Call the callback *now* with the given event mask.
195 362
363=item $w->feed_event ($revents)
364
365Feed some events on this watcher into EV. EV will react to this call as if
366the watcher had received the given C<$revents> mask.
367
368=item $revents = $w->clear_pending
369
370If the watcher is pending, this function clears its pending status and
371returns its C<$revents> bitset (as if its callback was invoked). If the
372watcher isn't pending it does nothing and returns C<0>.
373
374=item $previous_state = $w->keepalive ($bool)
375
376Normally, C<EV::loop> will return when there are no active watchers
377(which is a "deadlock" because no progress can be made anymore). This is
378convinient because it allows you to start your watchers (and your jobs),
379call C<EV::loop> once and when it returns you know that all your jobs are
380finished (or they forgot to register some watchers for their task :).
381
382Sometimes, however, this gets in your way, for example when the module
383that calls C<EV::loop> (usually the main program) is not the same module
384as a long-living watcher (for example a DNS client module written by
385somebody else even). Then you might want any outstanding requests to be
386handled, but you would not want to keep C<EV::loop> from returning just
387because you happen to have this long-running UDP port watcher.
388
389In this case you can clear the keepalive status, which means that even
390though your watcher is active, it won't keep C<EV::loop> from returning.
391
392The initial value for keepalive is true (enabled), and you cna change it
393any time.
394
395Example: Register an I/O watcher for some UDP socket but do not keep the
396event loop from running just because of that watcher.
397
398 my $udp_socket = ...
399 my $udp_watcher = EV::io $udp_socket, EV::READ, sub { ... };
400 $1000udp_watcher->keepalive (0);
401
402=item $loop = $w->loop
403
404Return the loop that this watcher is attached to.
405
406=back
407
408
409=head1 WATCHER TYPES
410
411Each of the following subsections describes a single watcher type.
412
413=head3 I/O WATCHERS - is this file descriptor readable or writable?
414
415=over 4
196 416
197=item $w = EV::io $fileno_or_fh, $eventmask, $callback 417=item $w = EV::io $fileno_or_fh, $eventmask, $callback
198 418
199=item $w = EV::io_ns $fileno_or_fh, $eventmask, $callback 419=item $w = EV::io_ns $fileno_or_fh, $eventmask, $callback
200 420
421=item $w = $loop->io ($fileno_or_fh, $eventmask, $callback)
422
423=item $w = $loop->io_ns ($fileno_or_fh, $eventmask, $callback)
424
201As long as the returned watcher object is alive, call the C<$callback> 425As long as the returned watcher object is alive, call the C<$callback>
202when the events specified in C<$eventmask>. 426when at least one of events specified in C<$eventmask> occurs.
203 427
204The $eventmask can be one or more of these constants ORed together: 428The $eventmask can be one or more of these constants ORed together:
205 429
206 EV::READ wait until read() wouldn't block anymore 430 EV::READ wait until read() wouldn't block anymore
207 EV::WRITE wait until write() wouldn't block anymore 431 EV::WRITE wait until write() wouldn't block anymore
223 447
224=item $old_eventmask = $w->events ($new_eventmask) 448=item $old_eventmask = $w->events ($new_eventmask)
225 449
226Returns the previously set event mask and optionally set a new one. 450Returns the previously set event mask and optionally set a new one.
227 451
452=back
453
454
455=head3 TIMER WATCHERS - relative and optionally repeating timeouts
456
457=over 4
228 458
229=item $w = EV::timer $after, $repeat, $callback 459=item $w = EV::timer $after, $repeat, $callback
230 460
231=item $w = EV::timer_ns $after, $repeat, $callback 461=item $w = EV::timer_ns $after, $repeat, $callback
232 462
233Calls the callback after C<$after> seconds. If C<$repeat> is non-zero, 463=item $w = $loop->timer ($after, $repeat, $callback)
234the timer will be restarted (with the $repeat value as $after) after the 464
235callback returns. 465=item $w = $loop->timer_ns ($after, $repeat, $callback)
466
467Calls the callback after C<$after> seconds (which may be fractional). If
468C<$repeat> is non-zero, the timer will be restarted (with the $repeat
469value as $after) after the callback returns.
236 470
237This means that the callback would be called roughly after C<$after> 471This means that the callback would be called roughly after C<$after>
238seconds, and then every C<$repeat> seconds. "Roughly" because the time of 472seconds, and then every C<$repeat> seconds. The timer does his best not
239callback processing is not taken into account, so the timer will slowly 473to drift, but it will not invoke the timer more often then once per event
240drift. If that isn't acceptable, look at EV::periodic. 474loop iteration, and might drift in other cases. If that isn't acceptable,
475look at EV::periodic, which can provide long-term stable timers.
241 476
242The timer is based on a monotonic clock, that is if somebody is sitting 477The timer is based on a monotonic clock, that is, if somebody is sitting
243in front of the machine while the timer is running and changes the system 478in front of the machine while the timer is running and changes the system
244clock, the timer will nevertheless run (roughly) the same time. 479clock, the timer will nevertheless run (roughly) the same time.
245 480
246The C<timer_ns> variant doesn't start (activate) the newly created watcher. 481The C<timer_ns> variant doesn't start (activate) the newly created watcher.
247 482
248=item $w->set ($after, $repeat) 483=item $w->set ($after, $repeat)
249 484
250Reconfigures the watcher, see the constructor above for details. Can be at 485Reconfigures the watcher, see the constructor above for details. Can be called at
251any time. 486any time.
252 487
253=item $w->again 488=item $w->again
254 489
255Similar to the C<start> method, but has special semantics for repeating timers: 490Similar to the C<start> method, but has special semantics for repeating timers:
491
492If the timer is active and non-repeating, it will be stopped.
256 493
257If the timer is active and repeating, reset the timeout to occur 494If the timer is active and repeating, reset the timeout to occur
258C<$repeat> seconds after now. 495C<$repeat> seconds after now.
259 496
260If the timer is active and non-repeating, it will be stopped.
261
262If the timer is in active and repeating, start it. 497If the timer is inactive and repeating, start it using the repeat value.
263 498
264Otherwise do nothing. 499Otherwise do nothing.
265 500
266This behaviour is useful when you have a timeout for some IO 501This behaviour is useful when you have a timeout for some IO
267operation. You create a timer object with the same value for C<$after> and 502operation. You create a timer object with the same value for C<$after> and
268C<$repeat>, and then, in the read/write watcher, run the C<again> method 503C<$repeat>, and then, in the read/write watcher, run the C<again> method
269on the timeout. 504on the timeout.
270 505
506=back
271 507
508
509=head3 PERIODIC WATCHERS - to cron or not to cron?
510
511=over 4
512
272=item $w = EV::periodic $at, $interval, $callback 513=item $w = EV::periodic $at, $interval, $reschedule_cb, $callback
273 514
274=item $w = EV::periodic_ns $at, $interval, $callback 515=item $w = EV::periodic_ns $at, $interval, $reschedule_cb, $callback
275 516
276Similar to EV::timer, but the time is given as an absolute point in time 517=item $w = $loop->periodic ($at, $interval, $reschedule_cb, $callback)
277(C<$at>), plus an optional C<$interval>.
278 518
279If the C<$interval> is zero, then the callback will be called at the time 519=item $w = $loop->periodic_ns ($at, $interval, $reschedule_cb, $callback)
280C<$at> if that is in the future, or as soon as possible if it is in the
281past. It will not automatically repeat.
282 520
283If the C<$interval> is nonzero, then the watcher will always be scheduled 521Similar to EV::timer, but is not based on relative timeouts but on
284to time out at the next C<$at + N * $interval> time. 522absolute times. Apart from creating "simple" timers that trigger "at" the
523specified time, it can also be used for non-drifting absolute timers and
524more complex, cron-like, setups that are not adversely affected by time
525jumps (i.e. when the system clock is changed by explicit date -s or other
526means such as ntpd). It is also the most complex watcher type in EV.
285 527
286This can be used to schedule a callback to run at very regular intervals, 528It has three distinct "modes":
287as long as the processing time is less then the interval (otherwise 529
288obviously events will be skipped). 530=over 4
531
532=item * absolute timer ($interval = $reschedule_cb = 0)
533
534This time simply fires at the wallclock time C<$at> and doesn't repeat. It
535will not adjust when a time jump occurs, that is, if it is to be run
536at January 1st 2011 then it will run when the system time reaches or
537surpasses this time.
538
539=item * repeating interval timer ($interval > 0, $reschedule_cb = 0)
540
541In this mode the watcher will always be scheduled to time out at the
542next C<$at + N * $interval> time (for some integer N) and then repeat,
543regardless of any time jumps.
544
545This can be used to create timers that do not drift with respect to system
546time:
547
548 my $hourly = EV::periodic 0, 3600, 0, sub { print "once/hour\n" };
549
550That doesn't mean there will always be 3600 seconds in between triggers,
551but only that the the clalback will be called when the system time shows a
552full hour (UTC).
289 553
290Another way to think about it (for the mathematically inclined) is that 554Another way to think about it (for the mathematically inclined) is that
291EV::periodic will try to run the callback at the next possible time where 555EV::periodic will try to run the callback in this mode at the next
292C<$time = $at (mod $interval)>, regardless of any time jumps. 556possible time where C<$time = $at (mod $interval)>, regardless of any time
557jumps.
293 558
294This periodic timer is based on "wallclock time", that is, if the clock 559=item * manual reschedule mode ($reschedule_cb = coderef)
295changes (C<ntp>, C<date -s> etc.), then the timer will nevertheless run at 560
296the specified time. 561In this mode $interval and $at are both being ignored. Instead, each
562time the periodic watcher gets scheduled, the reschedule callback
563($reschedule_cb) will be called with the watcher as first, and the current
564time as second argument.
565
566I<This callback MUST NOT stop or destroy this or any other periodic
567watcher, ever>. If you need to stop it, return 1e30 and stop it
568afterwards.
569
570It must return the next time to trigger, based on the passed time value
571(that is, the lowest time value larger than to the second argument). It
572will usually be called just before the callback will be triggered, but
573might be called at other times, too.
574
575This can be used to create very complex timers, such as a timer that
576triggers on each midnight, local time (actually 24 hours after the last
577midnight, to keep the example simple. If you know a way to do it correctly
578in about the same space (without requiring elaborate modules), drop me a
579note :):
580
581 my $daily = EV::periodic 0, 0, sub {
582 my ($w, $now) = @_;
583
584 use Time::Local ();
585 my (undef, undef, undef, $d, $m, $y) = localtime $now;
586 86400 + Time::Local::timelocal 0, 0, 0, $d, $m, $y
587 }, sub {
588 print "it's midnight or likely shortly after, now\n";
589 };
590
591=back
297 592
298The C<periodic_ns> variant doesn't start (activate) the newly created watcher. 593The C<periodic_ns> variant doesn't start (activate) the newly created watcher.
299 594
300=item $w->set ($at, $interval) 595=item $w->set ($at, $interval, $reschedule_cb)
301 596
302Reconfigures the watcher, see the constructor above for details. Can be at 597Reconfigures the watcher, see the constructor above for details. Can be called at
303any time. 598any time.
304 599
600=item $w->again
601
602Simply stops and starts the watcher again.
603
604=item $time = $w->at
605
606Return the time that the watcher is expected to trigger next.
607
608=back
609
610
611=head3 SIGNAL WATCHERS - signal me when a signal gets signalled!
612
613=over 4
305 614
306=item $w = EV::signal $signal, $callback 615=item $w = EV::signal $signal, $callback
307 616
308=item $w = EV::signal_ns $signal, $callback 617=item $w = EV::signal_ns $signal, $callback
309 618
310Call the callback when $signal is received (the signal can be specified 619Call the callback when $signal is received (the signal can be specified by
311by number or by name, just as with kill or %SIG). 620number or by name, just as with C<kill> or C<%SIG>).
312 621
313EV will grab the signal for the process (the kernel only allows one 622EV will grab the signal for the process (the kernel only allows one
314component to receive a signal at a time) when you start a signal watcher, 623component to receive a signal at a time) when you start a signal watcher,
315and removes it again when you stop it. Perl does the same when you 624and removes it again when you stop it. Perl does the same when you
316add/remove callbacks to %SIG, so watch out. 625add/remove callbacks to C<%SIG>, so watch out.
317 626
318You can have as many signal watchers per signal as you want. 627You can have as many signal watchers per signal as you want.
319 628
320The C<signal_ns> variant doesn't start (activate) the newly created watcher. 629The C<signal_ns> variant doesn't start (activate) the newly created watcher.
321 630
322=item $w->set ($signal) 631=item $w->set ($signal)
323 632
324Reconfigures the watcher, see the constructor above for details. Can be at 633Reconfigures the watcher, see the constructor above for details. Can be
634called at any time.
635
636=item $current_signum = $w->signal
637
638=item $old_signum = $w->signal ($new_signal)
639
640Returns the previously set signal (always as a number not name) and
641optionally set a new one.
642
643=back
644
645
646=head3 CHILD WATCHERS - watch out for process status changes
647
648=over 4
649
650=item $w = EV::child $pid, $trace, $callback
651
652=item $w = EV::child_ns $pid, $trace, $callback
653
654=item $w = $loop->child ($pid, $trace, $callback)
655
656=item $w = $loop->child_ns ($pid, $trace, $callback)
657
658Call the callback when a status change for pid C<$pid> (or any pid
659if C<$pid> is 0) has been received (a status change happens when the
660process terminates or is killed, or, when trace is true, additionally when
661it is stopped or continued). More precisely: when the process receives
662a C<SIGCHLD>, EV will fetch the outstanding exit/wait status for all
663changed/zombie children and call the callback.
664
665It is valid (and fully supported) to install a child watcher after a child
666has exited but before the event loop has started its next iteration (for
667example, first you C<fork>, then the new child process might exit, and
668only then do you install a child watcher in the parent for the new pid).
669
670You can access both exit (or tracing) status and pid by using the
671C<rstatus> and C<rpid> methods on the watcher object.
672
673You can have as many pid watchers per pid as you want, they will all be
674called.
675
676The C<child_ns> variant doesn't start (activate) the newly created watcher.
677
678=item $w->set ($pid, $trace)
679
680Reconfigures the watcher, see the constructor above for details. Can be called at
325any time. 681any time.
326 682
683=item $current_pid = $w->pid
327 684
328=item $w = EV::child $pid, $callback 685Returns the previously set process id and optionally set a new one.
329 686
330=item $w = EV::child_ns $pid, $callback 687=item $exit_status = $w->rstatus
331 688
332Call the callback when a status change for pid C<$pid> (or any pid 689Return the exit/wait status (as returned by waitpid, see the waitpid entry
333if C<$pid> is 0) has been received. More precisely: when the process 690in perlfunc).
334receives a SIGCHLD, EV will fetch the outstanding exit/wait status for all
335changed/zombie children and call the callback.
336 691
337Unlike all other callbacks, this callback will be called with an 692=item $pid = $w->rpid
338additional third argument which is the exit status. See the C<waitpid>
339function for details.
340 693
341You can have as many pid watchers per pid as you want. 694Return the pid of the awaited child (useful when you have installed a
695watcher for all pids).
342 696
697=back
698
699
700=head3 STAT WATCHERS - did the file attributes just change?
701
702=over 4
703
704=item $w = EV::stat $path, $interval, $callback
705
706=item $w = EV::stat_ns $path, $interval, $callback
707
708=item $w = $loop->stat ($path, $interval, $callback)
709
710=item $w = $loop->stat_ns ($path, $interval, $callback)
711
712Call the callback when a file status change has been detected on
713C<$path>. The C<$path> does not need to exist, changing from "path exists"
714to "path does not exist" is a status change like any other.
715
716The C<$interval> is a recommended polling interval for systems where
717OS-supported change notifications don't exist or are not supported. If
718you use C<0> then an unspecified default is used (which is highly
719recommended!), which is to be expected to be around five seconds usually.
720
721This watcher type is not meant for massive numbers of stat watchers,
722as even with OS-supported change notifications, this can be
723resource-intensive.
724
343The C<child_ns> variant doesn't start (activate) the newly created watcher. 725The C<stat_ns> variant doesn't start (activate) the newly created watcher.
344 726
345=item $w->set ($pid) 727=item ... = $w->stat
346 728
729This call is very similar to the perl C<stat> built-in: It stats (using
730C<lstat>) the path specified in the watcher and sets perls stat cache (as
731well as EV's idea of the current stat values) to the values found.
732
733In scalar context, a boolean is return indicating success or failure of
734the stat. In list context, the same 13-value list as with stat is returned
735(except that the blksize and blocks fields are not reliable).
736
737In the case of an error, errno is set to C<ENOENT> (regardless of the
738actual error value) and the C<nlink> value is forced to zero (if the stat
739was successful then nlink is guaranteed to be non-zero).
740
741See also the next two entries for more info.
742
743=item ... = $w->attr
744
745Just like C<< $w->stat >>, but without the initial stat'ing: this returns
746the values most recently detected by EV. See the next entry for more info.
747
748=item ... = $w->prev
749
750Just like C<< $w->stat >>, but without the initial stat'ing: this returns
751the previous set of values, before the change.
752
753That is, when the watcher callback is invoked, C<< $w->prev >> will be set
754to the values found I<before> a change was detected, while C<< $w->attr >>
755returns the values found leading to the change detection. The difference (if any)
756between C<prev> and C<attr> is what triggered the callback.
757
758If you did something to the filesystem object and do not want to trigger
759yet another change, you can call C<stat> to update EV's idea of what the
760current attributes are.
761
762=item $w->set ($path, $interval)
763
347Reconfigures the watcher, see the constructor above for details. Can be at 764Reconfigures the watcher, see the constructor above for details. Can be
348any time. 765called at any time.
349 766
767=item $current_path = $w->path
768
769=item $old_path = $w->path ($new_path)
770
771Returns the previously set path and optionally set a new one.
772
773=item $current_interval = $w->interval
774
775=item $old_interval = $w->interval ($new_interval)
776
777Returns the previously set interval and optionally set a new one. Can be
778used to query the actual interval used.
779
780=back
781
782
783=head3 IDLE WATCHERS - when you've got nothing better to do...
784
785=over 4
350 786
351=item $w = EV::idle $callback 787=item $w = EV::idle $callback
352 788
353=item $w = EV::idle_ns $callback 789=item $w = EV::idle_ns $callback
354 790
355Call the callback when there are no pending io, timer/periodic, signal or 791=item $w = $loop->idle ($callback)
356child events, i.e. when the process is idle. 792
793=item $w = $loop->idle_ns ($callback)
794
795Call the callback when there are no other pending watchers of the same or
796higher priority (excluding check, prepare and other idle watchers of the
797same or lower priority, of course). They are called idle watchers because
798when the watcher is the highest priority pending event in the process, the
799process is considered to be idle at that priority.
800
801If you want a watcher that is only ever called when I<no> other events are
802outstanding you have to set the priority to C<EV::MINPRI>.
357 803
358The process will not block as long as any idle watchers are active, and 804The process will not block as long as any idle watchers are active, and
359they will be called repeatedly until stopped. 805they will be called repeatedly until stopped.
360 806
807For example, if you have idle watchers at priority C<0> and C<1>, and
808an I/O watcher at priority C<0>, then the idle watcher at priority C<1>
809and the I/O watcher will always run when ready. Only when the idle watcher
810at priority C<1> is stopped and the I/O watcher at priority C<0> is not
811pending with the C<0>-priority idle watcher be invoked.
812
361The C<idle_ns> variant doesn't start (activate) the newly created watcher. 813The C<idle_ns> variant doesn't start (activate) the newly created watcher.
362 814
815=back
816
817
818=head3 PREPARE WATCHERS - customise your event loop!
819
820=over 4
363 821
364=item $w = EV::prepare $callback 822=item $w = EV::prepare $callback
365 823
366=item $w = EV::prepare_ns $callback 824=item $w = EV::prepare_ns $callback
825
826=item $w = $loop->prepare ($callback)
827
828=item $w = $loop->prepare_ns ($callback)
367 829
368Call the callback just before the process would block. You can still 830Call the callback just before the process would block. You can still
369create/modify any watchers at this point. 831create/modify any watchers at this point.
370 832
371See the EV::check watcher, below, for explanations and an example. 833See the EV::check watcher, below, for explanations and an example.
372 834
373The C<prepare_ns> variant doesn't start (activate) the newly created watcher. 835The C<prepare_ns> variant doesn't start (activate) the newly created watcher.
374 836
837=back
838
839
840=head3 CHECK WATCHERS - customise your event loop even more!
841
842=over 4
375 843
376=item $w = EV::check $callback 844=item $w = EV::check $callback
377 845
378=item $w = EV::check_ns $callback 846=item $w = EV::check_ns $callback
847
848=item $w = $loop->check ($callback)
849
850=item $w = $loop->check_ns ($callback)
379 851
380Call the callback just after the process wakes up again (after it has 852Call the callback just after the process wakes up again (after it has
381gathered events), but before any other callbacks have been invoked. 853gathered events), but before any other callbacks have been invoked.
382 854
383This is used to integrate other event-based software into the EV 855This is used to integrate other event-based software into the EV
391 # do nothing unless active 863 # do nothing unless active
392 $dispatcher->{_event_queue_h} 864 $dispatcher->{_event_queue_h}
393 or return; 865 or return;
394 866
395 # make the dispatcher handle any outstanding stuff 867 # make the dispatcher handle any outstanding stuff
868 ... not shown
396 869
397 # create an IO watcher for each and every socket 870 # create an I/O watcher for each and every socket
398 @snmp_watcher = ( 871 @snmp_watcher = (
399 (map { EV::io $_, EV::READ, sub { } } 872 (map { EV::io $_, EV::READ, sub { } }
400 keys %{ $dispatcher->{_descriptors} }), 873 keys %{ $dispatcher->{_descriptors} }),
874
875 EV::timer +($event->[Net::SNMP::Dispatcher::_ACTIVE]
876 ? $event->[Net::SNMP::Dispatcher::_TIME] - EV::now : 0),
877 0, sub { },
401 ); 878 );
402
403 # if there are any timeouts, also create a timer
404 push @snmp_watcher, EV::timer $event->[Net::SNMP::Dispatcher::_TIME] - EV::now, 0, sub { }
405 if $event->[Net::SNMP::Dispatcher::_ACTIVE];
406 }; 879 };
407 880
408The callbacks are irrelevant, the only purpose of those watchers is 881The callbacks are irrelevant (and are not even being called), the
409to wake up the process as soon as one of those events occurs (socket 882only purpose of those watchers is to wake up the process as soon as
410readable, or timer timed out). The corresponding EV::check watcher will then 883one of those events occurs (socket readable, or timer timed out). The
411clean up: 884corresponding EV::check watcher will then clean up:
412 885
413 our $snmp_check = EV::check sub { 886 our $snmp_check = EV::check sub {
414 # destroy all watchers 887 # destroy all watchers
415 @snmp_watcher = (); 888 @snmp_watcher = ();
416 889
417 # make the dispatcher handle any new stuff 890 # make the dispatcher handle any new stuff
891 ... not shown
418 }; 892 };
419 893
420The callbacks of the created watchers will not be called as the watchers 894The callbacks of the created watchers will not be called as the watchers
421are destroyed before this cna happen (remember EV::check gets called 895are destroyed before this cna happen (remember EV::check gets called
422first). 896first).
423 897
424The C<check_ns> variant doesn't start (activate) the newly created watcher. 898The C<check_ns> variant doesn't start (activate) the newly created watcher.
425 899
426=back 900=back
427 901
902
903=head3 FORK WATCHERS - the audacity to resume the event loop after a fork
904
905Fork watchers are called when a C<fork ()> was detected. The invocation
906is done before the event loop blocks next and before C<check> watchers
907are being called, and only in the child after the fork.
908
909=over 4
910
911=item $w = EV::fork $callback
912
913=item $w = EV::fork_ns $callback
914
915=item $w = $loop->fork ($callback)
916
917=item $w = $loop->fork_ns ($callback)
918
919Call the callback before the event loop is resumed in the child process
920after a fork.
921
922The C<fork_ns> variant doesn't start (activate) the newly created watcher.
923
924=back
925
926
927=head3 EMBED WATCHERS - when one backend isn't enough...
928
929This is a rather advanced watcher type that lets you embed one event loop
930into another (currently only IO events are supported in the embedded
931loop, other types of watchers might be handled in a delayed or incorrect
932fashion and must not be used).
933
934See the libev documentation at
935L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#code_ev_embed_code_when_one_backend_>
936for more details.
937
938In short, this watcher is most useful on BSD systems without working
939kqueue to still be able to handle a large number of sockets:
940
941 my $socket_loop;
942
943 # check wether we use SELECT or POLL _and_ KQUEUE is supported
944 if (
945 (EV::backend & (EV::BACKEND_POLL | EV::BACKEND_SELECT))
946 && (EV::supported_backends & EV::embeddable_backends & EV::BACKEND_KQUEUE)
947 ) {
948 # use kqueue for sockets
949 $socket_loop = new EV::Loop EV::BACKEND_KQUEUE | EV::FLAG_NOENV;
950 }
951
952 # use the default loop otherwise
953 $socket_loop ||= EV::default_loop;
954
955=over 4
956
957=item $w = EV::embed $otherloop, $callback
958
959=item $w = EV::embed_ns $otherloop, $callback
960
961=item $w = $loop->embed ($otherloop, $callback)
962
963=item $w = $loop->embed_ns ($otherloop, $callback)
964
965Call the callback when the embedded event loop (C<$otherloop>) has any
966I/O activity. The C<$callback> should alwas be specified as C<undef> in
967this version of EV, which means the embedded event loop will be managed
968automatically.
969
970The C<embed_ns> variant doesn't start (activate) the newly created watcher.
971
972=back
973
974=head3 ASYNC WATCHERS - how to wake up another event loop
975
976Async watchers are provided by EV, but have little use in perl directly, as perl
977neither supports threads nor direct access to signal handlers or other
978contexts where they could be of value.
979
980It is, however, possible to use them from the XS level.
981
982Please see the libev documentation for further details.
983
984=over 4
985
986=item $w = EV::async $callback
987
988=item $w = EV::async_ns $callback
989
990=item $w->send
991
992=item $bool = $w->async_pending
993
994=back
995
996
997=head1 PERL SIGNALS
998
999While Perl signal handling (C<%SIG>) is not affected by EV, the behaviour
1000with EV is as the same as any other C library: Perl-signals will only be
1001handled when Perl runs, which means your signal handler might be invoked
1002only the next time an event callback is invoked.
1003
1004The solution is to use EV signal watchers (see C<EV::signal>), which will
1005ensure proper operations with regards to other event watchers.
1006
1007If you cannot do this for whatever reason, you can also force a watcher
1008to be called on every event loop iteration by installing a C<EV::check>
1009watcher:
1010
1011 my $async_check = EV::check sub { };
1012
1013This ensures that perl gets into control for a short time to handle any
1014pending signals, and also ensures (slightly) slower overall operation.
1015
428=head1 THREADS 1016=head1 THREADS
429 1017
430Threads are not supported by this in any way. Perl pseudo-threads is evil 1018Threads are not supported by this module in any way. Perl pseudo-threads
431stuff and must die. 1019is evil stuff and must die. As soon as Perl gains real threads I will work
1020on thread support for it.
1021
1022=head1 FORK
1023
1024Most of the "improved" event delivering mechanisms of modern operating
1025systems have quite a few problems with fork(2) (to put it bluntly: it is
1026not supported and usually destructive). Libev makes it possible to work
1027around this by having a function that recreates the kernel state after
1028fork in the child.
1029
1030On non-win32 platforms, this module requires the pthread_atfork
1031functionality to do this automatically for you. This function is quite
1032buggy on most BSDs, though, so YMMV. The overhead for this is quite
1033negligible, because everything the function currently does is set a flag
1034that is checked only when the event loop gets used the next time, so when
1035you do fork but not use EV, the overhead is minimal.
1036
1037On win32, there is no notion of fork so all this doesn't apply, of course.
432 1038
433=cut 1039=cut
434 1040
435our $DIED = sub { 1041our $DIED = sub {
436 warn "EV: error in callback (ignoring): $@"; 1042 warn "EV: error in callback (ignoring): $@";
437}; 1043};
438 1044
439init; 1045default_loop
440 1046 or die 'EV: cannot initialise libev backend. bad $ENV{LIBEV_FLAGS}?';
441push @AnyEvent::REGISTRY, [EV => "EV::AnyEvent"];
442 1047
4431; 10481;
444 1049
445=head1 SEE ALSO 1050=head1 SEE ALSO
446 1051
447 L<EV::DNS>, L<EV::AnyEvent>. 1052L<EV::ADNS> (asynchronous DNS), L<Glib::EV> (makes Glib/Gtk2 use EV as
1053event loop), L<EV::Glib> (embed Glib into EV), L<Coro::EV> (efficient
1054coroutines with EV), L<Net::SNMP::EV> (asynchronous SNMP).
448 1055
449=head1 AUTHOR 1056=head1 AUTHOR
450 1057
451 Marc Lehmann <schmorp@schmorp.de> 1058 Marc Lehmann <schmorp@schmorp.de>
452 http://home.schmorp.de/ 1059 http://home.schmorp.de/

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines