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

Comparing EV/EV.pm (file contents):
Revision 1.8 by root, Sat Oct 27 19:11:27 2007 UTC vs.
Revision 1.130 by root, Sun Oct 24 17:58:41 2010 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 # IO
22
23 my $w = EV::timer_abs 0, 60, sub { 21 my $w = EV::periodic 0, 60, 0, sub {
24 warn "is called every minute, on the minute, exactly"; 22 warn "is called every minute, on the minute, exactly";
25 }; 23 };
26 24
25 # IO
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 "nothign 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 } 30 };
31
32 # SIGNALS
33
34 my $w = EV::signal 'QUIT', sub {
35 warn "sigquit received\n";
34 }; 36 };
35 $w->timeout (10); 37
36 38 # CHILD/PID STATUS CHANGES
39
40 my $w = EV::child 666, 0, sub {
41 my ($w, $revents) = @_;
42 my $status = $w->rstatus;
43 };
44
45 # STAT CHANGES
46 my $w = EV::stat "/etc/passwd", 10, sub {
47 my ($w, $revents) = @_;
48 warn $w->path, " has changed somehow.\n";
49 };
50
37 # MAINLOOP 51 # MAINLOOP
38 EV::dispatch; # loop as long as watchers are active 52 EV::loop; # loop until EV::unloop is called or all watchers stop
39 EV::loop; # the same thing 53 EV::loop EV::LOOP_ONESHOT; # block until at least one event could be handled
40 EV::loop EV::LOOP_ONCE; 54 EV::loop EV::LOOP_NONBLOCK; # try to handle same events, but do not block
41 EV::loop EV::LOOP_ONSHOT; 55
56=head1 BEFORE YOU START USING THIS MODULE
57
58If you only need timer, I/O, signal, child and idle watchers and not the
59advanced functionality of this module, consider using L<AnyEvent> instead,
60specifically the simplified API described in L<AE>.
61
62When used with EV as backend, the L<AE> API is as fast as the native L<EV>
63API, but your programs/modules will still run with many other event loops.
42 64
43=head1 DESCRIPTION 65=head1 DESCRIPTION
44 66
45This module provides an interface to libevent 67This module provides an interface to libev
46(L<http://monkey.org/~provos/libevent/>). You probably should acquaint 68(L<http://software.schmorp.de/pkg/libev.html>). While the documentation
47yourself with its documentation and source code to be able to use this 69below is comprehensive, one might also consult the documentation of
48module fully. 70libev itself (L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod> or
71F<perldoc EV::libev>) for more subtle details on watcher semantics or some
72discussion on the available backends, or how to force a specific backend
73with C<LIBEV_FLAGS>, or just about in any case because it has much more
74detailed information.
49 75
50Please note thta this module disables the libevent EPOLL method by 76This module is very fast and scalable. It is actually so fast that you
51default, see BUGS, below, if you need to enable it. 77can use it through the L<AnyEvent> module, stay portable to other event
78loops (if you don't rely on any watcher types not available through it)
79and still be faster than with any other event loop currently supported in
80Perl.
81
82=head2 PORTING FROM EV 3.X to 4.X
83
84EV version 4 introduces a number of incompatible changes summarised
85here. According to the depreciation strategy used by libev, there is a
86compatibility layer in place so programs should continue to run unchanged
87(the XS interface lacks this layer, so programs using that one need to be
88updated).
89
90This compatibility layer will be switched off in some future release.
91
92All changes relevant to Perl are renames of symbols, functions and
93methods:
94
95 EV::loop => EV::run
96 EV::LOOP_NONBLOCK => EV::RUN_NOWAIT
97 EV::LOOP_ONESHOT => EV::RUN_ONCE
98
99 EV::unloop => EV::break
100 EV::UNLOOP_CANCEL => EV::BREAK_CANCEL
101 EV::UNLOOP_ONE => EV::BREAK_ONE
102 EV::UNLOOP_ALL => EV::BREAK_ALL
103
104 EV::TIMEOUT => EV::TIMER
105
106 EV::loop_count => EV::iteration
107 EV::loop_depth => EV::depth
108 EV::loop_verify => EV::verify
109
110The loop object methods corresponding to the functions above have been
111similarly renamed.
112
113=head2 MODULE EXPORTS
114
115This module does not export any symbols.
52 116
53=cut 117=cut
54 118
55package EV; 119package EV;
56 120
57use strict; 121use common::sense;
58 122
59BEGIN { 123BEGIN {
60 our $VERSION = '0.01'; 124 our $VERSION = '4.00';
61 use XSLoader; 125 use XSLoader;
62 XSLoader::load "EV", $VERSION; 126 XSLoader::load "EV", $VERSION;
63} 127}
64 128
129@EV::IO::ISA =
130@EV::Timer::ISA =
131@EV::Periodic::ISA =
132@EV::Signal::ISA =
133@EV::Child::ISA =
134@EV::Stat::ISA =
135@EV::Idle::ISA =
136@EV::Prepare::ISA =
137@EV::Check::ISA =
138@EV::Embed::ISA =
139@EV::Fork::ISA =
140@EV::Async::ISA =
141 "EV::Watcher";
142
143@EV::Loop::Default::ISA = "EV::Loop";
144
145=head1 EVENT LOOPS
146
147EV supports multiple event loops: There is a single "default event loop"
148that can handle everything including signals and child watchers, and any
149number of "dynamic event loops" that can use different backends (with
150various limitations), but no child and signal watchers.
151
152You do not have to do anything to create the default event loop: When
153the module is loaded a suitable backend is selected on the premise of
154selecting a working backend (which for example rules out kqueue on most
155BSDs). Modules should, unless they have "special needs" always use the
156default loop as this is fastest (perl-wise), best supported by other
157modules (e.g. AnyEvent or Coro) and most portable event loop.
158
159For specific programs you can create additional event loops dynamically.
160
161If you want to take advantage of kqueue (which often works properly for
162sockets only) even though the default loop doesn't enable it, you can
163I<embed> a kqueue loop into the default loop: running the default loop
164will then also service the kqueue loop to some extent. See the example in
165the section about embed watchers for an example on how to achieve that.
166
167=over 4
168
169=item $loop = new EV::Loop [$flags]
170
171Create a new event loop as per the specified flags. Please refer to
172the C<ev_loop_new ()> function description in the libev documentation
173(L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#GLOBAL_FUNCTIONS>,
174or locally-installed as F<EV::libev> manpage) for more info.
175
176The loop will automatically be destroyed when it is no longer referenced
177by any watcher and the loop object goes out of scope.
178
179If you are not embedding the loop, then Using C<EV::FLAG_FORKCHECK>
180is recommended, as only the default event loop is protected by this
181module. If you I<are> embedding this loop in the default loop, this is not
182necessary, as C<EV::embed> automatically does the right thing on fork.
183
184=item $loop->loop_fork
185
186Must be called after a fork in the child, before entering or continuing
187the event loop. An alternative is to use C<EV::FLAG_FORKCHECK> which calls
188this function automatically, at some performance loss (refer to the libev
189documentation).
190
191=item $loop->loop_verify
192
193Calls C<ev_verify> to make internal consistency checks (for debugging
194libev) and abort the program if any data structures were found to be
195corrupted.
196
197=item $loop = EV::default_loop [$flags]
198
199Return the default loop (which is a singleton object). Since this module
200already creates the default loop with default flags, specifying flags here
201will not have any effect unless you destroy the default loop first, which
202isn't supported. So in short: don't do it, and if you break it, you get to
203keep the pieces.
204
205=back
206
207
65=head1 BASIC INTERFACE 208=head1 BASIC INTERFACE
66 209
67=over 4 210=over 4
68 211
69=item $EV::NPRI
70
71How many priority levels are available.
72
73=item $EV::DIED 212=item $EV::DIED
74 213
75Must contain a reference to a function that is called when a callback 214Must contain a reference to a function that is called when a callback
76throws an exception (with $@ containing thr error). The default prints an 215throws an exception (with $@ containing the error). The default prints an
77informative message and continues. 216informative message and continues.
78 217
79If this callback throws an exception it will be silently ignored. 218If this callback throws an exception it will be silently ignored.
80 219
220=item $flags = EV::supported_backends
221
222=item $flags = EV::recommended_backends
223
224=item $flags = EV::embeddable_backends
225
226Returns the set (see C<EV::BACKEND_*> flags) of backends supported by this
227instance of EV, the set of recommended backends (supposed to be good) for
228this platform and the set of embeddable backends (see EMBED WATCHERS).
229
230=item EV::sleep $seconds
231
232Block the process for the given number of (fractional) seconds.
233
234=item $time = EV::time
235
236Returns the current time in (fractional) seconds since the epoch.
237
81=item $time = EV::now 238=item $time = EV::now
82 239
83Returns the time in (fractional) seconds since the epoch. 240=item $time = $loop->now
84 241
85=item $version = EV::version 242Returns the time the last event loop iteration has been started. This
243is the time that (relative) timers are based on, and referring to it is
244usually faster then calling EV::time.
86 245
87=item $method = EV::method 246=item EV::now_update
88 247
89Return version string and event polling method used. 248=item $loop->now_update
90 249
91=item EV::loop $flags # EV::LOOP_ONCE, EV::LOOP_ONESHOT 250Establishes the current time by querying the kernel, updating the time
251returned by C<EV::now> in the progress. This is a costly operation and
252is usually done automatically within C<EV::loop>.
92 253
93=item EV::loopexit $after 254This function is rarely useful, but when some event callback runs for a
255very long time without entering the event loop, updating libev's idea of
256the current time is a good idea.
94 257
95Exit any active loop or dispatch after C<$after> seconds or immediately if 258=item EV::suspend
96C<$after> is missing or zero.
97 259
98=item EV::dispatch 260=item $loop->suspend
99 261
100Same as C<EV::loop 0>. 262=item EV::resume
101 263
102=item EV::event $callback 264=item $loop->resume
103 265
104Creates a new event watcher waiting for nothing, calling the given callback. 266These two functions suspend and resume a loop, for use when the loop is
267not used for a while and timeouts should not be processed.
105 268
269A typical use case would be an interactive program such as a game: When
270the user presses C<^Z> to suspend the game and resumes it an hour later it
271would be best to handle timeouts as if no time had actually passed while
272the program was suspended. This can be achieved by calling C<suspend>
273in your C<SIGTSTP> handler, sending yourself a C<SIGSTOP> and calling
274C<resume> directly afterwards to resume timer processing.
275
276Effectively, all C<timer> watchers will be delayed by the time spend
277between C<suspend> and C<resume>, and all C<periodic> watchers
278will be rescheduled (that is, they will lose any events that would have
279occured while suspended).
280
281After calling C<suspend> you B<must not> call I<any> function on the given
282loop other than C<resume>, and you B<must not> call C<resume>
283without a previous call to C<suspend>.
284
285Calling C<suspend>/C<resume> has the side effect of updating the event
286loop time (see C<now_update>).
287
288=item $backend = EV::backend
289
290=item $backend = $loop->backend
291
292Returns an integer describing the backend used by libev (EV::BACKEND_SELECT
293or EV::BACKEND_EPOLL).
294
295=item EV::loop [$flags]
296
297=item $loop->loop ([$flags])
298
299Begin checking for events and calling callbacks. It returns when a
300callback calls EV::unloop.
301
302The $flags argument can be one of the following:
303
304 0 as above
305 EV::LOOP_ONESHOT block at most once (wait, but do not loop)
306 EV::LOOP_NONBLOCK do not block at all (fetch/handle events but do not wait)
307
308=item EV::unloop [$how]
309
310=item $loop->unloop ([$how])
311
312When called with no arguments or an argument of EV::UNLOOP_ONE, makes the
313innermost call to EV::loop return.
314
315When called with an argument of EV::UNLOOP_ALL, all calls to EV::loop will return as
316fast as possible.
317
318=item $count = EV::loop_count
319
320=item $count = $loop->loop_count
321
322Return the number of times the event loop has polled for new
323events. Sometimes useful as a generation counter.
324
325=item EV::once $fh_or_undef, $events, $timeout, $cb->($revents)
326
327=item $loop->once ($fh_or_undef, $events, $timeout, $cb->($revents))
328
329This function rolls together an I/O and a timer watcher for a single
330one-shot event without the need for managing a watcher object.
331
332If C<$fh_or_undef> is a filehandle or file descriptor, then C<$events>
333must be a bitset containing either C<EV::READ>, C<EV::WRITE> or C<EV::READ
334| EV::WRITE>, indicating the type of I/O event you want to wait for. If
335you do not want to wait for some I/O event, specify C<undef> for
336C<$fh_or_undef> and C<0> for C<$events>).
337
338If timeout is C<undef> or negative, then there will be no
339timeout. Otherwise a EV::timer with this value will be started.
340
341When an error occurs or either the timeout or I/O watcher triggers, then
342the callback will be called with the received event set (in general
343you can expect it to be a combination of C<EV::ERROR>, C<EV::READ>,
344C<EV::WRITE> and C<EV::TIMER>).
345
346EV::once doesn't return anything: the watchers stay active till either
347of them triggers, then they will be stopped and freed, and the callback
348invoked.
349
350=item EV::feed_fd_event ($fd, $revents)
351
352=item $loop->feed_fd_event ($fd, $revents)
353
354Feed an event on a file descriptor into EV. EV will react to this call as
355if the readyness notifications specified by C<$revents> (a combination of
356C<EV::READ> and C<EV::WRITE>) happened on the file descriptor C<$fd>.
357
358=item EV::feed_signal_event ($signal)
359
360Feed a signal event into EV. EV will react to this call as if the signal
361specified by C<$signal> had occured.
362
363=item EV::set_io_collect_interval $time
364
365=item $loop->set_io_collect_interval ($time)
366
367=item EV::set_timeout_collect_interval $time
368
369=item $loop->set_timeout_collect_interval ($time)
370
371These advanced functions set the minimum block interval when polling for I/O events and the minimum
372wait interval for timer events. See the libev documentation at
373L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP>
374(locally installed as F<EV::libev>) for a more detailed discussion.
375
376=item $count = EV::pending_count
377
378=item $count = $loop->pending_count
379
380Returns the number of currently pending watchers.
381
382=item EV::invoke_pending
383
384=item $loop->invoke_pending
385
386Invoke all currently pending watchers.
387
388=back
389
390
391=head1 WATCHER OBJECTS
392
393A watcher is an object that gets created to record your interest in some
394event. For instance, if you want to wait for STDIN to become readable, you
395would create an EV::io watcher for that:
396
397 my $watcher = EV::io *STDIN, EV::READ, sub {
398 my ($watcher, $revents) = @_;
399 warn "yeah, STDIN should now be readable without blocking!\n"
400 };
401
402All watchers can be active (waiting for events) or inactive (paused). Only
403active watchers will have their callbacks invoked. All callbacks will be
404called with at least two arguments: the watcher and a bitmask of received
405events.
406
407Each watcher type has its associated bit in revents, so you can use the
408same callback for multiple watchers. The event mask is named after the
409type, i.e. EV::child sets EV::CHILD, EV::prepare sets EV::PREPARE,
410EV::periodic sets EV::PERIODIC and so on, with the exception of I/O events
411(which can set both EV::READ and EV::WRITE bits).
412
413In the rare case where one wants to create a watcher but not start it at
414the same time, each constructor has a variant with a trailing C<_ns> in
415its name, e.g. EV::io has a non-starting variant EV::io_ns and so on.
416
417Please note that a watcher will automatically be stopped when the watcher
418object is destroyed, so you I<need> to keep the watcher objects returned by
419the constructors.
420
421Also, all methods changing some aspect of a watcher (->set, ->priority,
422->fh and so on) automatically stop and start it again if it is active,
423which means pending events get lost.
424
425=head2 COMMON WATCHER METHODS
426
427This section lists methods common to all watchers.
428
429=over 4
430
431=item $w->start
432
433Starts a watcher if it isn't active already. Does nothing to an already
434active watcher. By default, all watchers start out in the active state
435(see the description of the C<_ns> variants if you need stopped watchers).
436
437=item $w->stop
438
439Stop a watcher if it is active. Also clear any pending events (events that
440have been received but that didn't yet result in a callback invocation),
441regardless of whether the watcher was active or not.
442
443=item $bool = $w->is_active
444
445Returns true if the watcher is active, false otherwise.
446
447=item $current_data = $w->data
448
449=item $old_data = $w->data ($new_data)
450
451Queries a freely usable data scalar on the watcher and optionally changes
452it. This is a way to associate custom data with a watcher:
453
454 my $w = EV::timer 60, 0, sub {
455 warn $_[0]->data;
456 };
457 $w->data ("print me!");
458
459=item $current_cb = $w->cb
460
461=item $old_cb = $w->cb ($new_cb)
462
463Queries the callback on the watcher and optionally changes it. You can do
464this at any time without the watcher restarting.
465
466=item $current_priority = $w->priority
467
468=item $old_priority = $w->priority ($new_priority)
469
470Queries the priority on the watcher and optionally changes it. Pending
471watchers with higher priority will be invoked first. The valid range of
472priorities lies between EV::MAXPRI (default 2) and EV::MINPRI (default
473-2). If the priority is outside this range it will automatically be
474normalised to the nearest valid priority.
475
476The default priority of any newly-created watcher is 0.
477
478Note that the priority semantics have not yet been fleshed out and are
479subject to almost certain change.
480
481=item $w->invoke ($revents)
482
483Call the callback *now* with the given event mask.
484
485=item $w->feed_event ($revents)
486
487Feed some events on this watcher into EV. EV will react to this call as if
488the watcher had received the given C<$revents> mask.
489
490=item $revents = $w->clear_pending
491
492If the watcher is pending, this function clears its pending status and
493returns its C<$revents> bitset (as if its callback was invoked). If the
494watcher isn't pending it does nothing and returns C<0>.
495
496=item $previous_state = $w->keepalive ($bool)
497
498Normally, C<EV::loop> will return when there are no active watchers
499(which is a "deadlock" because no progress can be made anymore). This is
500convenient because it allows you to start your watchers (and your jobs),
501call C<EV::loop> once and when it returns you know that all your jobs are
502finished (or they forgot to register some watchers for their task :).
503
504Sometimes, however, this gets in your way, for example when the module
505that calls C<EV::loop> (usually the main program) is not the same module
506as a long-living watcher (for example a DNS client module written by
507somebody else even). Then you might want any outstanding requests to be
508handled, but you would not want to keep C<EV::loop> from returning just
509because you happen to have this long-running UDP port watcher.
510
511In this case you can clear the keepalive status, which means that even
512though your watcher is active, it won't keep C<EV::loop> from returning.
513
514The initial value for keepalive is true (enabled), and you can change it
515any time.
516
517Example: Register an I/O watcher for some UDP socket but do not keep the
518event loop from running just because of that watcher.
519
520 my $udp_socket = ...
521 my $udp_watcher = EV::io $udp_socket, EV::READ, sub { ... };
522 $udp_watcher->keepalive (0);
523
524=item $loop = $w->loop
525
526Return the loop that this watcher is attached to.
527
528=back
529
530
531=head1 WATCHER TYPES
532
533Each of the following subsections describes a single watcher type.
534
535=head3 I/O WATCHERS - is this file descriptor readable or writable?
536
537=over 4
538
106=item my $w = EV::io $fileno_or_fh, $eventmask, $callback 539=item $w = EV::io $fileno_or_fh, $eventmask, $callback
107 540
108=item my $w = EV::io_ns $fileno_or_fh, $eventmask, $callback 541=item $w = EV::io_ns $fileno_or_fh, $eventmask, $callback
542
543=item $w = $loop->io ($fileno_or_fh, $eventmask, $callback)
544
545=item $w = $loop->io_ns ($fileno_or_fh, $eventmask, $callback)
109 546
110As long as the returned watcher object is alive, call the C<$callback> 547As long as the returned watcher object is alive, call the C<$callback>
111when the events specified in C<$eventmask> happen. Initially, the timeout 548when at least one of events specified in C<$eventmask> occurs.
112is disabled.
113 549
114Youc an additionall set a timeout to occur on the watcher, but note that
115this timeout will not be reset when you get an I/O event in the EV::PERSIST
116case, and reaching a timeout will always stop the watcher even in the
117EV::PERSIST case.
118
119If you want a timeout to occur only after a specific time of inactivity, set
120a repeating timeout and do NOT use EV::PERSIST.
121
122Eventmask can be one or more of these constants ORed together: 550The $eventmask can be one or more of these constants ORed together:
123 551
124 EV::READ wait until read() wouldn't block anymore 552 EV::READ wait until read() wouldn't block anymore
125 EV::WRITE wait until write() wouldn't block anymore 553 EV::WRITE wait until write() wouldn't block anymore
126 EV::PERSIST stay active after a (non-timeout) event occured
127 554
128The C<io_ns> variant doesn't add/start the newly created watcher. 555The C<io_ns> variant doesn't start (activate) the newly created watcher.
129 556
557=item $w->set ($fileno_or_fh, $eventmask)
558
559Reconfigures the watcher, see the constructor above for details. Can be
560called at any time.
561
562=item $current_fh = $w->fh
563
564=item $old_fh = $w->fh ($new_fh)
565
566Returns the previously set filehandle and optionally set a new one.
567
568=item $current_eventmask = $w->events
569
570=item $old_eventmask = $w->events ($new_eventmask)
571
572Returns the previously set event mask and optionally set a new one.
573
574=back
575
576
577=head3 TIMER WATCHERS - relative and optionally repeating timeouts
578
579=over 4
580
130=item my $w = EV::timer $after, $repeat, $callback 581=item $w = EV::timer $after, $repeat, $callback
131 582
132=item my $w = EV::timer_ns $after, $repeat, $callback 583=item $w = EV::timer_ns $after, $repeat, $callback
133 584
134Calls the callback after C<$after> seconds. If C<$repeat> is true, the 585=item $w = $loop->timer ($after, $repeat, $callback)
135timer will be restarted after the callback returns. This means that the
136callback would be called roughly every C<$after> seconds, prolonged by the
137time the callback takes.
138 586
587=item $w = $loop->timer_ns ($after, $repeat, $callback)
588
589Calls the callback after C<$after> seconds (which may be fractional). If
590C<$repeat> is non-zero, the timer will be restarted (with the $repeat
591value as $after) after the callback returns.
592
593This means that the callback would be called roughly after C<$after>
594seconds, and then every C<$repeat> seconds. The timer does his best not
595to drift, but it will not invoke the timer more often then once per event
596loop iteration, and might drift in other cases. If that isn't acceptable,
597look at EV::periodic, which can provide long-term stable timers.
598
599The timer is based on a monotonic clock, that is, if somebody is sitting
600in front of the machine while the timer is running and changes the system
601clock, the timer will nevertheless run (roughly) the same time.
602
139The C<timer_ns> variant doesn't add/start the newly created watcher. 603The C<timer_ns> variant doesn't start (activate) the newly created watcher.
140 604
141=item my $w = EV::timer_abs $at, $interval, $callback 605=item $w->set ($after, $repeat)
142 606
143=item my $w = EV::timer_abs_ns $at, $interval, $callback 607Reconfigures the watcher, see the constructor above for details. Can be called at
608any time.
144 609
145Similar to EV::timer, but the time is given as an absolute point in time 610=item $w->again
146(C<$at>), plus an optional C<$interval>.
147 611
148If the C<$interval> is zero, then the callback will be called at the time 612Similar to the C<start> method, but has special semantics for repeating timers:
149C<$at> if that is in the future, or as soon as possible if its in the
150past. It will not automatically repeat.
151 613
152If the C<$interval> is nonzero, then the watcher will always be scheduled 614If the timer is active and non-repeating, it will be stopped.
153to time out at the next C<$at + integer * $interval> time.
154 615
155This can be used to schedule a callback to run at very regular intervals, 616If the timer is active and repeating, reset the timeout to occur
156as long as the processing time is less then the interval (otherwise 617C<$repeat> seconds after now.
157obviously events will be skipped). 618
619If the timer is inactive and repeating, start it using the repeat value.
620
621Otherwise do nothing.
622
623This behaviour is useful when you have a timeout for some IO
624operation. You create a timer object with the same value for C<$after> and
625C<$repeat>, and then, in the read/write watcher, run the C<again> method
626on the timeout.
627
628=back
629
630
631=head3 PERIODIC WATCHERS - to cron or not to cron?
632
633=over 4
634
635=item $w = EV::periodic $at, $interval, $reschedule_cb, $callback
636
637=item $w = EV::periodic_ns $at, $interval, $reschedule_cb, $callback
638
639=item $w = $loop->periodic ($at, $interval, $reschedule_cb, $callback)
640
641=item $w = $loop->periodic_ns ($at, $interval, $reschedule_cb, $callback)
642
643Similar to EV::timer, but is not based on relative timeouts but on
644absolute times. Apart from creating "simple" timers that trigger "at" the
645specified time, it can also be used for non-drifting absolute timers and
646more complex, cron-like, setups that are not adversely affected by time
647jumps (i.e. when the system clock is changed by explicit date -s or other
648means such as ntpd). It is also the most complex watcher type in EV.
649
650It has three distinct "modes":
651
652=over 4
653
654=item * absolute timer ($interval = $reschedule_cb = 0)
655
656This time simply fires at the wallclock time C<$at> and doesn't repeat. It
657will not adjust when a time jump occurs, that is, if it is to be run
658at January 1st 2011 then it will run when the system time reaches or
659surpasses this time.
660
661=item * repeating interval timer ($interval > 0, $reschedule_cb = 0)
662
663In this mode the watcher will always be scheduled to time out at the
664next C<$at + N * $interval> time (for some integer N) and then repeat,
665regardless of any time jumps.
666
667This can be used to create timers that do not drift with respect to system
668time:
669
670 my $hourly = EV::periodic 0, 3600, 0, sub { print "once/hour\n" };
671
672That doesn't mean there will always be 3600 seconds in between triggers,
673but only that the the clalback will be called when the system time shows a
674full hour (UTC).
158 675
159Another way to think about it (for the mathematically inclined) is that 676Another way to think about it (for the mathematically inclined) is that
160C<timer_abs> will try to tun the callback at the next possible time where 677EV::periodic will try to run the callback in this mode at the next
161C<$time = $at (mod $interval)>, regardless of any time jumps. 678possible time where C<$time = $at (mod $interval)>, regardless of any time
679jumps.
162 680
681=item * manual reschedule mode ($reschedule_cb = coderef)
682
683In this mode $interval and $at are both being ignored. Instead, each
684time the periodic watcher gets scheduled, the reschedule callback
685($reschedule_cb) will be called with the watcher as first, and the current
686time as second argument.
687
688I<This callback MUST NOT stop or destroy this or any other periodic
689watcher, ever, and MUST NOT call any event loop functions or methods>. If
690you need to stop it, return 1e30 and stop it afterwards. You may create
691and start a C<EV::prepare> watcher for this task.
692
693It must return the next time to trigger, based on the passed time value
694(that is, the lowest time value larger than or equal to to the second
695argument). It will usually be called just before the callback will be
696triggered, but might be called at other times, too.
697
698This can be used to create very complex timers, such as a timer that
699triggers on each midnight, local time (actually 24 hours after the last
700midnight, to keep the example simple. If you know a way to do it correctly
701in about the same space (without requiring elaborate modules), drop me a
702note :):
703
704 my $daily = EV::periodic 0, 0, sub {
705 my ($w, $now) = @_;
706
707 use Time::Local ();
708 my (undef, undef, undef, $d, $m, $y) = localtime $now;
709 86400 + Time::Local::timelocal 0, 0, 0, $d, $m, $y
710 }, sub {
711 print "it's midnight or likely shortly after, now\n";
712 };
713
714=back
715
163The C<timer_abs_ns> variant doesn't add/start the newly created watcher. 716The C<periodic_ns> variant doesn't start (activate) the newly created watcher.
164 717
718=item $w->set ($at, $interval, $reschedule_cb)
719
720Reconfigures the watcher, see the constructor above for details. Can be called at
721any time.
722
723=item $w->again
724
725Simply stops and starts the watcher again.
726
727=item $time = $w->at
728
729Return the time that the watcher is expected to trigger next.
730
731=back
732
733
734=head3 SIGNAL WATCHERS - signal me when a signal gets signalled!
735
736=over 4
737
165=item my $w = EV::signal $signum, $callback 738=item $w = EV::signal $signal, $callback
166 739
167=item my $w = EV::signal_ns $signum, $callback 740=item $w = EV::signal_ns $signal, $callback
168 741
169Call the callback when signal $signum is received. 742=item $w = $loop->signal ($signal, $callback)
170 743
744=item $w = $loop->signal_ns ($signal, $callback)
745
746Call the callback when $signal is received (the signal can be specified by
747number or by name, just as with C<kill> or C<%SIG>).
748
749Only one event loop can grab a given signal - attempting to grab the same
750signal from two EV loops will crash the program immediately or cause data
751corruption.
752
753EV will grab the signal for the process (the kernel only allows one
754component to receive a signal at a time) when you start a signal watcher,
755and removes it again when you stop it. Perl does the same when you
756add/remove callbacks to C<%SIG>, so watch out.
757
758You can have as many signal watchers per signal as you want.
759
171The C<signal_ns> variant doesn't add/start the newly created watcher. 760The C<signal_ns> variant doesn't start (activate) the newly created watcher.
172 761
173=back 762=item $w->set ($signal)
174 763
175=head1 THE EV::Event CLASS 764Reconfigures the watcher, see the constructor above for details. Can be
765called at any time.
176 766
177All EV functions creating an event watcher (designated by C<my $w => 767=item $current_signum = $w->signal
178above) support the following methods on the returned watcher object:
179 768
180=over 4 769=item $old_signum = $w->signal ($new_signal)
181 770
182=item $w->add ($timeout) 771Returns the previously set signal (always as a number not name) and
772optionally set a new one.
183 773
184Stops and (re-)starts the event watcher, setting the optional timeout to 774=back
185the given value, or clearing the timeout if none is given.
186 775
776
777=head3 CHILD WATCHERS - watch out for process status changes
778
779=over 4
780
781=item $w = EV::child $pid, $trace, $callback
782
783=item $w = EV::child_ns $pid, $trace, $callback
784
785=item $w = $loop->child ($pid, $trace, $callback)
786
787=item $w = $loop->child_ns ($pid, $trace, $callback)
788
789Call the callback when a status change for pid C<$pid> (or any pid
790if C<$pid> is 0) has been received (a status change happens when the
791process terminates or is killed, or, when trace is true, additionally when
792it is stopped or continued). More precisely: when the process receives
793a C<SIGCHLD>, EV will fetch the outstanding exit/wait status for all
794changed/zombie children and call the callback.
795
796It is valid (and fully supported) to install a child watcher after a child
797has exited but before the event loop has started its next iteration (for
798example, first you C<fork>, then the new child process might exit, and
799only then do you install a child watcher in the parent for the new pid).
800
801You can access both exit (or tracing) status and pid by using the
802C<rstatus> and C<rpid> methods on the watcher object.
803
804You can have as many pid watchers per pid as you want, they will all be
805called.
806
807The C<child_ns> variant doesn't start (activate) the newly created watcher.
808
809=item $w->set ($pid, $trace)
810
811Reconfigures the watcher, see the constructor above for details. Can be called at
812any time.
813
814=item $current_pid = $w->pid
815
816Returns the previously set process id and optionally set a new one.
817
818=item $exit_status = $w->rstatus
819
820Return the exit/wait status (as returned by waitpid, see the waitpid entry
821in perlfunc).
822
823=item $pid = $w->rpid
824
825Return the pid of the awaited child (useful when you have installed a
826watcher for all pids).
827
828=back
829
830
831=head3 STAT WATCHERS - did the file attributes just change?
832
833=over 4
834
835=item $w = EV::stat $path, $interval, $callback
836
837=item $w = EV::stat_ns $path, $interval, $callback
838
839=item $w = $loop->stat ($path, $interval, $callback)
840
841=item $w = $loop->stat_ns ($path, $interval, $callback)
842
843Call the callback when a file status change has been detected on
844C<$path>. The C<$path> does not need to exist, changing from "path exists"
845to "path does not exist" is a status change like any other.
846
847The C<$interval> is a recommended polling interval for systems where
848OS-supported change notifications don't exist or are not supported. If
849you use C<0> then an unspecified default is used (which is highly
850recommended!), which is to be expected to be around five seconds usually.
851
852This watcher type is not meant for massive numbers of stat watchers,
853as even with OS-supported change notifications, this can be
854resource-intensive.
855
856The C<stat_ns> variant doesn't start (activate) the newly created watcher.
857
187=item $w->start 858=item ... = $w->stat
188 859
189Stops and (re-)starts the event watcher without touching the timeout. 860This call is very similar to the perl C<stat> built-in: It stats (using
861C<lstat>) the path specified in the watcher and sets perls stat cache (as
862well as EV's idea of the current stat values) to the values found.
190 863
191=item $w->del 864In scalar context, a boolean is return indicating success or failure of
865the stat. In list context, the same 13-value list as with stat is returned
866(except that the blksize and blocks fields are not reliable).
192 867
193=item $w->stop 868In the case of an error, errno is set to C<ENOENT> (regardless of the
869actual error value) and the C<nlink> value is forced to zero (if the stat
870was successful then nlink is guaranteed to be non-zero).
194 871
195Stop the event watcher if it was started. 872See also the next two entries for more info.
196 873
197=item $current_callback = $w->cb 874=item ... = $w->attr
198 875
199=item $old_callback = $w->cb ($new_callback) 876Just like C<< $w->stat >>, but without the initial stat'ing: this returns
877the values most recently detected by EV. See the next entry for more info.
200 878
879=item ... = $w->prev
880
881Just like C<< $w->stat >>, but without the initial stat'ing: this returns
882the previous set of values, before the change.
883
884That is, when the watcher callback is invoked, C<< $w->prev >> will be set
885to the values found I<before> a change was detected, while C<< $w->attr >>
886returns the values found leading to the change detection. The difference (if any)
887between C<prev> and C<attr> is what triggered the callback.
888
889If you did something to the filesystem object and do not want to trigger
890yet another change, you can call C<stat> to update EV's idea of what the
891current attributes are.
892
893=item $w->set ($path, $interval)
894
895Reconfigures the watcher, see the constructor above for details. Can be
896called at any time.
897
898=item $current_path = $w->path
899
900=item $old_path = $w->path ($new_path)
901
201Return the previously set callback and optionally set a new one. 902Returns the previously set path and optionally set a new one.
202 903
203=item $current_fh = $w->fh 904=item $current_interval = $w->interval
204 905
205=item $old_fh = $w->fh ($new_fh) 906=item $old_interval = $w->interval ($new_interval)
206 907
207Returns the previously set filehandle and optionally set a new one.
208
209=item $current_eventmask = $w->events
210
211=item $old_eventmask = $w->events ($new_eventmask)
212
213Returns the previously set event mask and optionally set a new one. 908Returns the previously set interval and optionally set a new one. Can be
909used to query the actual interval used.
214 910
215=item $w->timeout ($after, $repeat)
216
217Resets the timeout (see C<EV::timer> for details).
218
219=item $w->timeout_abs ($at, $interval)
220
221Resets the timeout (see C<EV::timer_abs> for details).
222
223=item $w->priority_set ($priority)
224
225Set the priority of the watcher to C<$priority> (0 <= $priority < $EV::NPRI).
226
227=back 911=back
228 912
229=head1 BUGS
230 913
231Lots. Libevent itself isn't well tested and rather buggy, and this module 914=head3 IDLE WATCHERS - when you've got nothing better to do...
232is quite new at the moment.
233 915
234Please note that the epoll method is not, in general, reliable in programs 916=over 4
235that use fork (even if no libveent calls are being made in the forked
236process). If your program behaves erratically, try setting the environment
237variable C<EVENT_NOEPOLL> first when running the program.
238 917
239In general, if you fork, then you can only use the EV module in one of the 918=item $w = EV::idle $callback
240children. 919
920=item $w = EV::idle_ns $callback
921
922=item $w = $loop->idle ($callback)
923
924=item $w = $loop->idle_ns ($callback)
925
926Call the callback when there are no other pending watchers of the same or
927higher priority (excluding check, prepare and other idle watchers of the
928same or lower priority, of course). They are called idle watchers because
929when the watcher is the highest priority pending event in the process, the
930process is considered to be idle at that priority.
931
932If you want a watcher that is only ever called when I<no> other events are
933outstanding you have to set the priority to C<EV::MINPRI>.
934
935The process will not block as long as any idle watchers are active, and
936they will be called repeatedly until stopped.
937
938For example, if you have idle watchers at priority C<0> and C<1>, and
939an I/O watcher at priority C<0>, then the idle watcher at priority C<1>
940and the I/O watcher will always run when ready. Only when the idle watcher
941at priority C<1> is stopped and the I/O watcher at priority C<0> is not
942pending with the C<0>-priority idle watcher be invoked.
943
944The C<idle_ns> variant doesn't start (activate) the newly created watcher.
945
946=back
947
948
949=head3 PREPARE WATCHERS - customise your event loop!
950
951=over 4
952
953=item $w = EV::prepare $callback
954
955=item $w = EV::prepare_ns $callback
956
957=item $w = $loop->prepare ($callback)
958
959=item $w = $loop->prepare_ns ($callback)
960
961Call the callback just before the process would block. You can still
962create/modify any watchers at this point.
963
964See the EV::check watcher, below, for explanations and an example.
965
966The C<prepare_ns> variant doesn't start (activate) the newly created watcher.
967
968=back
969
970
971=head3 CHECK WATCHERS - customise your event loop even more!
972
973=over 4
974
975=item $w = EV::check $callback
976
977=item $w = EV::check_ns $callback
978
979=item $w = $loop->check ($callback)
980
981=item $w = $loop->check_ns ($callback)
982
983Call the callback just after the process wakes up again (after it has
984gathered events), but before any other callbacks have been invoked.
985
986This can be used to integrate other event-based software into the EV
987mainloop: You register a prepare callback and in there, you create io and
988timer watchers as required by the other software. Here is a real-world
989example of integrating Net::SNMP (with some details left out):
990
991 our @snmp_watcher;
992
993 our $snmp_prepare = EV::prepare sub {
994 # do nothing unless active
995 $dispatcher->{_event_queue_h}
996 or return;
997
998 # make the dispatcher handle any outstanding stuff
999 ... not shown
1000
1001 # create an I/O watcher for each and every socket
1002 @snmp_watcher = (
1003 (map { EV::io $_, EV::READ, sub { } }
1004 keys %{ $dispatcher->{_descriptors} }),
1005
1006 EV::timer +($event->[Net::SNMP::Dispatcher::_ACTIVE]
1007 ? $event->[Net::SNMP::Dispatcher::_TIME] - EV::now : 0),
1008 0, sub { },
1009 );
1010 };
1011
1012The callbacks are irrelevant (and are not even being called), the
1013only purpose of those watchers is to wake up the process as soon as
1014one of those events occurs (socket readable, or timer timed out). The
1015corresponding EV::check watcher will then clean up:
1016
1017 our $snmp_check = EV::check sub {
1018 # destroy all watchers
1019 @snmp_watcher = ();
1020
1021 # make the dispatcher handle any new stuff
1022 ... not shown
1023 };
1024
1025The callbacks of the created watchers will not be called as the watchers
1026are destroyed before this can happen (remember EV::check gets called
1027first).
1028
1029The C<check_ns> variant doesn't start (activate) the newly created watcher.
1030
1031=item EV::CHECK constant issues
1032
1033Like all other watcher types, there is a bitmask constant for use in
1034C<$revents> and other places. The C<EV::CHECK> is special as it has
1035the same name as the C<CHECK> sub called by Perl. This doesn't cause
1036big issues on newer perls (beginning with 5.8.9), but it means thatthe
1037constant must be I<inlined>, i.e. runtime calls will not work. That means
1038that as long as you always C<use EV> and then C<EV::CHECK> you are on the
1039safe side.
1040
1041=back
1042
1043
1044=head3 FORK WATCHERS - the audacity to resume the event loop after a fork
1045
1046Fork watchers are called when a C<fork ()> was detected. The invocation
1047is done before the event loop blocks next and before C<check> watchers
1048are being called, and only in the child after the fork.
1049
1050=over 4
1051
1052=item $w = EV::fork $callback
1053
1054=item $w = EV::fork_ns $callback
1055
1056=item $w = $loop->fork ($callback)
1057
1058=item $w = $loop->fork_ns ($callback)
1059
1060Call the callback before the event loop is resumed in the child process
1061after a fork.
1062
1063The C<fork_ns> variant doesn't start (activate) the newly created watcher.
1064
1065=back
1066
1067
1068=head3 EMBED WATCHERS - when one backend isn't enough...
1069
1070This is a rather advanced watcher type that lets you embed one event loop
1071into another (currently only IO events are supported in the embedded
1072loop, other types of watchers might be handled in a delayed or incorrect
1073fashion and must not be used).
1074
1075See the libev documentation at
1076L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#code_ev_embed_code_when_one_backend_>
1077(locally installed as F<EV::libev>) for more details.
1078
1079In short, this watcher is most useful on BSD systems without working
1080kqueue to still be able to handle a large number of sockets:
1081
1082 my $socket_loop;
1083
1084 # check wether we use SELECT or POLL _and_ KQUEUE is supported
1085 if (
1086 (EV::backend & (EV::BACKEND_POLL | EV::BACKEND_SELECT))
1087 && (EV::supported_backends & EV::embeddable_backends & EV::BACKEND_KQUEUE)
1088 ) {
1089 # use kqueue for sockets
1090 $socket_loop = new EV::Loop EV::BACKEND_KQUEUE | EV::FLAG_NOENV;
1091 }
1092
1093 # use the default loop otherwise
1094 $socket_loop ||= EV::default_loop;
1095
1096=over 4
1097
1098=item $w = EV::embed $otherloop[, $callback]
1099
1100=item $w = EV::embed_ns $otherloop[, $callback]
1101
1102=item $w = $loop->embed ($otherloop[, $callback])
1103
1104=item $w = $loop->embed_ns ($otherloop[, $callback])
1105
1106Call the callback when the embedded event loop (C<$otherloop>) has any
1107I/O activity. The C<$callback> is optional: if it is missing, then the
1108embedded event loop will be managed automatically (which is recommended),
1109otherwise you have to invoke C<sweep> yourself.
1110
1111The C<embed_ns> variant doesn't start (activate) the newly created watcher.
1112
1113=back
1114
1115=head3 ASYNC WATCHERS - how to wake up another event loop
1116
1117Async watchers are provided by EV, but have little use in perl directly,
1118as perl neither supports threads running in parallel nor direct access to
1119signal handlers or other contexts where they could be of value.
1120
1121It is, however, possible to use them from the XS level.
1122
1123Please see the libev documentation for further details.
1124
1125=over 4
1126
1127=item $w = EV::async $callback
1128
1129=item $w = EV::async_ns $callback
1130
1131=item $w->send
1132
1133=item $bool = $w->async_pending
1134
1135=back
1136
1137
1138=head1 PERL SIGNALS
1139
1140While Perl signal handling (C<%SIG>) is not affected by EV, the behaviour
1141with EV is as the same as any other C library: Perl-signals will only be
1142handled when Perl runs, which means your signal handler might be invoked
1143only the next time an event callback is invoked.
1144
1145The solution is to use EV signal watchers (see C<EV::signal>), which will
1146ensure proper operations with regards to other event watchers.
1147
1148If you cannot do this for whatever reason, you can also force a watcher
1149to be called on every event loop iteration by installing a C<EV::check>
1150watcher:
1151
1152 my $async_check = EV::check sub { };
1153
1154This ensures that perl gets into control for a short time to handle any
1155pending signals, and also ensures (slightly) slower overall operation.
1156
1157=head1 ITHREADS
1158
1159Ithreads are not supported by this module in any way. Perl pseudo-threads
1160is evil stuff and must die. Real threads as provided by Coro are fully
1161supported (and enhanced support is available via L<Coro::EV>).
1162
1163=head1 FORK
1164
1165Most of the "improved" event delivering mechanisms of modern operating
1166systems have quite a few problems with fork(2) (to put it bluntly: it is
1167not supported and usually destructive). Libev makes it possible to work
1168around this by having a function that recreates the kernel state after
1169fork in the child.
1170
1171On non-win32 platforms, this module requires the pthread_atfork
1172functionality to do this automatically for you. This function is quite
1173buggy on most BSDs, though, so YMMV. The overhead for this is quite
1174negligible, because everything the function currently does is set a flag
1175that is checked only when the event loop gets used the next time, so when
1176you do fork but not use EV, the overhead is minimal.
1177
1178On win32, there is no notion of fork so all this doesn't apply, of course.
241 1179
242=cut 1180=cut
243 1181
244our $DIED = sub { 1182our $DIED = sub {
245 warn "EV: error in callback (ignoring): $@"; 1183 warn "EV: error in callback (ignoring): $@";
246}; 1184};
247 1185
248our $NPRI = 4; 1186default_loop
249our $BASE = init; 1187 or die 'EV: cannot initialise libev backend. bad $ENV{LIBEV_FLAGS}?';
250priority_init $NPRI;
251
252push @AnyEvent::REGISTRY, [EV => "EV::AnyEvent"];
253 1188
2541; 11891;
255 1190
256=head1 SEE ALSO 1191=head1 SEE ALSO
257 1192
258 L<EV::DNS>, L<event(3)>, L<event.h>, L<evdns.h>. 1193L<EV::MakeMaker> - MakeMaker interface to XS API, L<EV::ADNS>
259 L<EV::AnyEvent>. 1194(asynchronous DNS), L<Glib::EV> (makes Glib/Gtk2 use EV as event
1195loop), L<EV::Glib> (embed Glib into EV), L<Coro::EV> (efficient thread
1196integration), L<Net::SNMP::EV> (asynchronous SNMP), L<AnyEvent> for
1197event-loop agnostic and portable event driven programming.
260 1198
261=head1 AUTHOR 1199=head1 AUTHOR
262 1200
263 Marc Lehmann <schmorp@schmorp.de> 1201 Marc Lehmann <schmorp@schmorp.de>
264 http://home.schmorp.de/ 1202 http://home.schmorp.de/
265 1203
266=cut 1204=cut
267 1205

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines