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

Comparing EV/README (file contents):
Revision 1.7 by root, Sat Nov 3 16:25:49 2007 UTC vs.
Revision 1.32 by root, Tue Apr 28 00:50:57 2009 UTC

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

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines