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

Comparing AnyEvent/README (file contents):
Revision 1.19 by root, Mon Apr 28 08:02:14 2008 UTC vs.
Revision 1.75 by root, Thu Jul 16 12:48:38 2015 UTC

1NAME 1NAME
2 AnyEvent - provide framework for multiple event loops 2 AnyEvent - the DBI of event loop programming
3 3
4 EV, Event, Coro::EV, Coro::Event, Glib, Tk, Perl, Event::Lib, Qt, POE - 4 EV, Event, Glib, Tk, UV, Perl, Event::Lib, Irssi, rxvt-unicode,
5 various supported event loops 5 IO::Async, Qt, FLTK and POE are various supported event
6 loops/environments.
6 7
7SYNOPSIS 8SYNOPSIS
8 use AnyEvent; 9 use AnyEvent;
9 10
11 # if you prefer function calls, look at the AE manpage for
12 # an alternative API.
13
14 # file handle or descriptor readable
10 my $w = AnyEvent->io (fh => $fh, poll => "r|w", cb => sub { 15 my $w = AnyEvent->io (fh => $fh, poll => "r", cb => sub { ... });
16
17 # one-shot or repeating timers
18 my $w = AnyEvent->timer (after => $seconds, cb => sub { ... });
19 my $w = AnyEvent->timer (after => $seconds, interval => $seconds, cb => ...);
20
21 print AnyEvent->now; # prints current event loop time
22 print AnyEvent->time; # think Time::HiRes::time or simply CORE::time.
23
24 # POSIX signal
25 my $w = AnyEvent->signal (signal => "TERM", cb => sub { ... });
26
27 # child process exit
28 my $w = AnyEvent->child (pid => $pid, cb => sub {
29 my ($pid, $status) = @_;
11 ... 30 ...
12 }); 31 });
13 32
14 my $w = AnyEvent->timer (after => $seconds, cb => sub { 33 # called when event loop idle (if applicable)
15 ... 34 my $w = AnyEvent->idle (cb => sub { ... });
16 });
17 35
18 my $w = AnyEvent->condvar; # stores whether a condition was flagged 36 my $w = AnyEvent->condvar; # stores whether a condition was flagged
37 $w->send; # wake up current and all future recv's
19 $w->wait; # enters "main loop" till $condvar gets ->broadcast 38 $w->recv; # enters "main loop" till $condvar gets ->send
20 $w->broadcast; # wake up current and all future wait's 39 # use a condvar in callback mode:
40 $w->cb (sub { $_[0]->recv });
41
42INTRODUCTION/TUTORIAL
43 This manpage is mainly a reference manual. If you are interested in a
44 tutorial or some gentle introduction, have a look at the AnyEvent::Intro
45 manpage.
46
47SUPPORT
48 An FAQ document is available as AnyEvent::FAQ.
49
50 There also is a mailinglist for discussing all things AnyEvent, and an
51 IRC channel, too.
52
53 See the AnyEvent project page at the Schmorpforge Ta-Sa Software
54 Repository, at <http://anyevent.schmorp.de>, for more info.
21 55
22WHY YOU SHOULD USE THIS MODULE (OR NOT) 56WHY YOU SHOULD USE THIS MODULE (OR NOT)
23 Glib, POE, IO::Async, Event... CPAN offers event models by the dozen 57 Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
24 nowadays. So what is different about AnyEvent? 58 nowadays. So what is different about AnyEvent?
25 59
26 Executive Summary: AnyEvent is *compatible*, AnyEvent is *free of 60 Executive Summary: AnyEvent is *compatible*, AnyEvent is *free of
27 policy* and AnyEvent is *small and efficient*. 61 policy* and AnyEvent is *small and efficient*.
28 62
29 First and foremost, *AnyEvent is not an event model* itself, it only 63 First and foremost, *AnyEvent is not an event model* itself, it only
30 interfaces to whatever event model the main program happens to use in a 64 interfaces to whatever event model the main program happens to use, in a
31 pragmatic way. For event models and certain classes of immortals alike, 65 pragmatic way. For event models and certain classes of immortals alike,
32 the statement "there can only be one" is a bitter reality: In general, 66 the statement "there can only be one" is a bitter reality: In general,
33 only one event loop can be active at the same time in a process. 67 only one event loop can be active at the same time in a process.
34 AnyEvent helps hiding the differences between those event loops. 68 AnyEvent cannot change this, but it can hide the differences between
69 those event loops.
35 70
36 The goal of AnyEvent is to offer module authors the ability to do event 71 The goal of AnyEvent is to offer module authors the ability to do event
37 programming (waiting for I/O or timer events) without subscribing to a 72 programming (waiting for I/O or timer events) without subscribing to a
38 religion, a way of living, and most importantly: without forcing your 73 religion, a way of living, and most importantly: without forcing your
39 module users into the same thing by forcing them to use the same event 74 module users into the same thing by forcing them to use the same event
40 model you use. 75 model you use.
41 76
42 For modules like POE or IO::Async (which is a total misnomer as it is 77 For modules like POE or IO::Async (which is a total misnomer as it is
43 actually doing all I/O *synchronously*...), using them in your module is 78 actually doing all I/O *synchronously*...), using them in your module is
44 like joining a cult: After you joined, you are dependent on them and you 79 like joining a cult: After you join, you are dependent on them and you
45 cannot use anything else, as it is simply incompatible to everything 80 cannot use anything else, as they are simply incompatible to everything
46 that isn't itself. What's worse, all the potential users of your module 81 that isn't them. What's worse, all the potential users of your module
47 are *also* forced to use the same event loop you use. 82 are *also* forced to use the same event loop you use.
48 83
49 AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works 84 AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
50 fine. AnyEvent + Tk works fine etc. etc. but none of these work together 85 fine. AnyEvent + Tk works fine etc. etc. but none of these work together
51 with the rest: POE + IO::Async? no go. Tk + Event? no go. Again: if your 86 with the rest: POE + EV? No go. Tk + Event? No go. Again: if your module
52 module uses one of those, every user of your module has to use it, too. 87 uses one of those, every user of your module has to use it, too. But if
53 But if your module uses AnyEvent, it works transparently with all event 88 your module uses AnyEvent, it works transparently with all event models
54 models it supports (including stuff like POE and IO::Async, as long as 89 it supports (including stuff like IO::Async, as long as those use one of
55 those use one of the supported event loops. It is trivial to add new 90 the supported event loops. It is easy to add new event loops to
56 event loops to AnyEvent, too, so it is future-proof). 91 AnyEvent, too, so it is future-proof).
57 92
58 In addition to being free of having to use *the one and only true event 93 In addition to being free of having to use *the one and only true event
59 model*, AnyEvent also is free of bloat and policy: with POE or similar 94 model*, AnyEvent also is free of bloat and policy: with POE or similar
60 modules, you get an enourmous amount of code and strict rules you have 95 modules, you get an enormous amount of code and strict rules you have to
61 to follow. AnyEvent, on the other hand, is lean and up to the point, by 96 follow. AnyEvent, on the other hand, is lean and to the point, by only
62 only offering the functionality that is necessary, in as thin as a 97 offering the functionality that is necessary, in as thin as a wrapper as
63 wrapper as technically possible. 98 technically possible.
64 99
100 Of course, AnyEvent comes with a big (and fully optional!) toolbox of
101 useful functionality, such as an asynchronous DNS resolver, 100%
102 non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
103 such as Windows) and lots of real-world knowledge and workarounds for
104 platform bugs and differences.
105
65 Of course, if you want lots of policy (this can arguably be somewhat 106 Now, if you *do want* lots of policy (this can arguably be somewhat
66 useful) and you want to force your users to use the one and only event 107 useful) and you want to force your users to use the one and only event
67 model, you should *not* use this module. 108 model, you should *not* use this module.
68 109
69DESCRIPTION 110DESCRIPTION
70 AnyEvent provides an identical interface to multiple event loops. This 111 AnyEvent provides a uniform interface to various event loops. This
71 allows module authors to utilise an event loop without forcing module 112 allows module authors to use event loop functionality without forcing
72 users to use the same event loop (as only a single event loop can 113 module users to use a specific event loop implementation (since more
73 coexist peacefully at any one time). 114 than one event loop cannot coexist peacefully).
74 115
75 The interface itself is vaguely similar, but not identical to the Event 116 The interface itself is vaguely similar, but not identical to the Event
76 module. 117 module.
77 118
78 During the first call of any watcher-creation method, the module tries 119 During the first call of any watcher-creation method, the module tries
79 to detect the currently loaded event loop by probing whether one of the 120 to detect the currently loaded event loop by probing whether one of the
80 following modules is already loaded: Coro::EV, Coro::Event, EV, Event, 121 following modules is already loaded: EV, AnyEvent::Loop, Event, Glib,
81 Glib, AnyEvent::Impl::Perl, Tk, Event::Lib, Qt, POE. The first one found 122 Tk, Event::Lib, Qt, POE. The first one found is used. If none are
82 is used. If none are found, the module tries to load these modules 123 detected, the module tries to load the first four modules in the order
83 (excluding Tk, Event::Lib, Qt and POE as the pure perl adaptor should 124 given; but note that if EV is not available, the pure-perl
84 always succeed) in the order given. The first one that can be 125 AnyEvent::Loop should always work, so the other two are not normally
85 successfully loaded will be used. If, after this, still none could be 126 tried.
86 found, AnyEvent will fall back to a pure-perl event loop, which is not
87 very efficient, but should work everywhere.
88 127
89 Because AnyEvent first checks for modules that are already loaded, 128 Because AnyEvent first checks for modules that are already loaded,
90 loading an event model explicitly before first using AnyEvent will 129 loading an event model explicitly before first using AnyEvent will
91 likely make that model the default. For example: 130 likely make that model the default. For example:
92 131
… …
94 use AnyEvent; 133 use AnyEvent;
95 134
96 # .. AnyEvent will likely default to Tk 135 # .. AnyEvent will likely default to Tk
97 136
98 The *likely* means that, if any module loads another event model and 137 The *likely* means that, if any module loads another event model and
99 starts using it, all bets are off. Maybe you should tell their authors 138 starts using it, all bets are off - this case should be very rare
100 to use AnyEvent so their modules work together with others seamlessly... 139 though, as very few modules hardcode event loops without announcing this
140 very loudly.
101 141
102 The pure-perl implementation of AnyEvent is called 142 The pure-perl implementation of AnyEvent is called "AnyEvent::Loop".
103 "AnyEvent::Impl::Perl". Like other event modules you can load it 143 Like other event modules you can load it explicitly and enjoy the high
104 explicitly. 144 availability of that event loop :)
105 145
106WATCHERS 146WATCHERS
107 AnyEvent has the central concept of a *watcher*, which is an object that 147 AnyEvent has the central concept of a *watcher*, which is an object that
108 stores relevant data for each kind of event you are waiting for, such as 148 stores relevant data for each kind of event you are waiting for, such as
109 the callback to call, the filehandle to watch, etc. 149 the callback to call, the file handle to watch, etc.
110 150
111 These watchers are normal Perl objects with normal Perl lifetime. After 151 These watchers are normal Perl objects with normal Perl lifetime. After
112 creating a watcher it will immediately "watch" for events and invoke the 152 creating a watcher it will immediately "watch" for events and invoke the
113 callback when the event occurs (of course, only when the event model is 153 callback when the event occurs (of course, only when the event model is
114 in control). 154 in control).
115 155
156 Note that callbacks must not permanently change global variables
157 potentially in use by the event loop (such as $_ or $[) and that
158 callbacks must not "die". The former is good programming practice in
159 Perl and the latter stems from the fact that exception handling differs
160 widely between event loops.
161
116 To disable the watcher you have to destroy it (e.g. by setting the 162 To disable a watcher you have to destroy it (e.g. by setting the
117 variable you store it in to "undef" or otherwise deleting all references 163 variable you store it in to "undef" or otherwise deleting all references
118 to it). 164 to it).
119 165
120 All watchers are created by calling a method on the "AnyEvent" class. 166 All watchers are created by calling a method on the "AnyEvent" class.
121 167
122 Many watchers either are used with "recursion" (repeating timers for 168 Many watchers either are used with "recursion" (repeating timers for
123 example), or need to refer to their watcher object in other ways. 169 example), or need to refer to their watcher object in other ways.
124 170
125 An any way to achieve that is this pattern: 171 One way to achieve that is this pattern:
126 172
127 my $w; $w = AnyEvent->type (arg => value ..., cb => sub { 173 my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
128 # you can use $w here, for example to undef it 174 # you can use $w here, for example to undef it
129 undef $w; 175 undef $w;
130 }); 176 });
131 177
132 Note that "my $w; $w =" combination. This is necessary because in Perl, 178 Note that "my $w; $w =" combination. This is necessary because in Perl,
133 my variables are only visible after the statement in which they are 179 my variables are only visible after the statement in which they are
134 declared. 180 declared.
135 181
136 I/O WATCHERS 182 I/O WATCHERS
183 $w = AnyEvent->io (
184 fh => <filehandle_or_fileno>,
185 poll => <"r" or "w">,
186 cb => <callback>,
187 );
188
137 You can create an I/O watcher by calling the "AnyEvent->io" method with 189 You can create an I/O watcher by calling the "AnyEvent->io" method with
138 the following mandatory key-value pairs as arguments: 190 the following mandatory key-value pairs as arguments:
139 191
140 "fh" the Perl *file handle* (*not* file descriptor) to watch for events. 192 "fh" is the Perl *file handle* (or a naked file descriptor) to watch for
193 events (AnyEvent might or might not keep a reference to this file
194 handle). Note that only file handles pointing to things for which
195 non-blocking operation makes sense are allowed. This includes sockets,
196 most character devices, pipes, fifos and so on, but not for example
197 files or block devices.
198
141 "poll" must be a string that is either "r" or "w", which creates a 199 "poll" must be a string that is either "r" or "w", which creates a
142 watcher waiting for "r"eadable or "w"ritable events, respectively. "cb" 200 watcher waiting for "r"eadable or "w"ritable events, respectively.
201
143 is the callback to invoke each time the file handle becomes ready. 202 "cb" is the callback to invoke each time the file handle becomes ready.
144 203
145 Although the callback might get passed parameters, their value and 204 Although the callback might get passed parameters, their value and
146 presence is undefined and you cannot rely on them. Portable AnyEvent 205 presence is undefined and you cannot rely on them. Portable AnyEvent
147 callbacks cannot use arguments passed to I/O watcher callbacks. 206 callbacks cannot use arguments passed to I/O watcher callbacks.
148 207
149 The I/O watcher might use the underlying file descriptor or a copy of 208 The I/O watcher might use the underlying file descriptor or a copy of
150 it. You must not close a file handle as long as any watcher is active on 209 it. You must not close a file handle as long as any watcher is active on
151 the underlying file descriptor. 210 the underlying file descriptor.
152 211
153 Some event loops issue spurious readyness notifications, so you should 212 Some event loops issue spurious readiness notifications, so you should
154 always use non-blocking calls when reading/writing from/to your file 213 always use non-blocking calls when reading/writing from/to your file
155 handles. 214 handles.
156 215
157 Example:
158
159 # wait for readability of STDIN, then read a line and disable the watcher 216 Example: wait for readability of STDIN, then read a line and disable the
217 watcher.
218
160 my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub { 219 my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
161 chomp (my $input = <STDIN>); 220 chomp (my $input = <STDIN>);
162 warn "read: $input\n"; 221 warn "read: $input\n";
163 undef $w; 222 undef $w;
164 }); 223 });
165 224
166 TIME WATCHERS 225 TIME WATCHERS
226 $w = AnyEvent->timer (after => <seconds>, cb => <callback>);
227
228 $w = AnyEvent->timer (
229 after => <fractional_seconds>,
230 interval => <fractional_seconds>,
231 cb => <callback>,
232 );
233
167 You can create a time watcher by calling the "AnyEvent->timer" method 234 You can create a time watcher by calling the "AnyEvent->timer" method
168 with the following mandatory arguments: 235 with the following mandatory arguments:
169 236
170 "after" specifies after how many seconds (fractional values are 237 "after" specifies after how many seconds (fractional values are
171 supported) the callback should be invoked. "cb" is the callback to 238 supported) the callback should be invoked. "cb" is the callback to
… …
173 240
174 Although the callback might get passed parameters, their value and 241 Although the callback might get passed parameters, their value and
175 presence is undefined and you cannot rely on them. Portable AnyEvent 242 presence is undefined and you cannot rely on them. Portable AnyEvent
176 callbacks cannot use arguments passed to time watcher callbacks. 243 callbacks cannot use arguments passed to time watcher callbacks.
177 244
178 The timer callback will be invoked at most once: if you want a repeating 245 The callback will normally be invoked only once. If you specify another
179 timer you have to create a new watcher (this is a limitation by both Tk 246 parameter, "interval", as a strictly positive number (> 0), then the
180 and Glib). 247 callback will be invoked regularly at that interval (in fractional
248 seconds) after the first invocation. If "interval" is specified with a
249 false value, then it is treated as if it were not specified at all.
181 250
182 Example: 251 The callback will be rescheduled before invoking the callback, but no
252 attempt is made to avoid timer drift in most backends, so the interval
253 is only approximate.
183 254
184 # fire an event after 7.7 seconds 255 Example: fire an event after 7.7 seconds.
256
185 my $w = AnyEvent->timer (after => 7.7, cb => sub { 257 my $w = AnyEvent->timer (after => 7.7, cb => sub {
186 warn "timeout\n"; 258 warn "timeout\n";
187 }); 259 });
188 260
189 # to cancel the timer: 261 # to cancel the timer:
190 undef $w; 262 undef $w;
191 263
192 Example 2:
193
194 # fire an event after 0.5 seconds, then roughly every second 264 Example 2: fire an event after 0.5 seconds, then roughly every second.
195 my $w;
196 265
197 my $cb = sub {
198 # cancel the old timer while creating a new one
199 $w = AnyEvent->timer (after => 1, cb => $cb); 266 my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
267 warn "timeout\n";
200 }; 268 });
201
202 # start the "loop" by creating the first watcher
203 $w = AnyEvent->timer (after => 0.5, cb => $cb);
204 269
205 TIMING ISSUES 270 TIMING ISSUES
206 There are two ways to handle timers: based on real time (relative, "fire 271 There are two ways to handle timers: based on real time (relative, "fire
207 in 10 seconds") and based on wallclock time (absolute, "fire at 12 272 in 10 seconds") and based on wallclock time (absolute, "fire at 12
208 o'clock"). 273 o'clock").
209 274
210 While most event loops expect timers to specified in a relative way, 275 While most event loops expect timers to specified in a relative way,
211 they use absolute time internally. This makes a difference when your 276 they use absolute time internally. This makes a difference when your
212 clock "jumps", for example, when ntp decides to set your clock backwards 277 clock "jumps", for example, when ntp decides to set your clock backwards
213 from the wrong date of 2014-01-01 to 2008-01-01, a watcher that is 278 from the wrong date of 2014-01-01 to 2008-01-01, a watcher that is
214 supposed to fire "after" a second might actually take six years to 279 supposed to fire "after a second" might actually take six years to
215 finally fire. 280 finally fire.
216 281
217 AnyEvent cannot compensate for this. The only event loop that is 282 AnyEvent cannot compensate for this. The only event loop that is
218 conscious about these issues is EV, which offers both relative 283 conscious of these issues is EV, which offers both relative (ev_timer,
219 (ev_timer, based on true relative time) and absolute (ev_periodic, based 284 based on true relative time) and absolute (ev_periodic, based on
220 on wallclock time) timers. 285 wallclock time) timers.
221 286
222 AnyEvent always prefers relative timers, if available, matching the 287 AnyEvent always prefers relative timers, if available, matching the
223 AnyEvent API. 288 AnyEvent API.
224 289
290 AnyEvent has two additional methods that return the "current time":
291
292 AnyEvent->time
293 This returns the "current wallclock time" as a fractional number of
294 seconds since the Epoch (the same thing as "time" or
295 "Time::HiRes::time" return, and the result is guaranteed to be
296 compatible with those).
297
298 It progresses independently of any event loop processing, i.e. each
299 call will check the system clock, which usually gets updated
300 frequently.
301
302 AnyEvent->now
303 This also returns the "current wallclock time", but unlike "time",
304 above, this value might change only once per event loop iteration,
305 depending on the event loop (most return the same time as "time",
306 above). This is the time that AnyEvent's timers get scheduled
307 against.
308
309 *In almost all cases (in all cases if you don't care), this is the
310 function to call when you want to know the current time.*
311
312 This function is also often faster then "AnyEvent->time", and thus
313 the preferred method if you want some timestamp (for example,
314 AnyEvent::Handle uses this to update its activity timeouts).
315
316 The rest of this section is only of relevance if you try to be very
317 exact with your timing; you can skip it without a bad conscience.
318
319 For a practical example of when these times differ, consider
320 Event::Lib and EV and the following set-up:
321
322 The event loop is running and has just invoked one of your callbacks
323 at time=500 (assume no other callbacks delay processing). In your
324 callback, you wait a second by executing "sleep 1" (blocking the
325 process for a second) and then (at time=501) you create a relative
326 timer that fires after three seconds.
327
328 With Event::Lib, "AnyEvent->time" and "AnyEvent->now" will both
329 return 501, because that is the current time, and the timer will be
330 scheduled to fire at time=504 (501 + 3).
331
332 With EV, "AnyEvent->time" returns 501 (as that is the current time),
333 but "AnyEvent->now" returns 500, as that is the time the last event
334 processing phase started. With EV, your timer gets scheduled to run
335 at time=503 (500 + 3).
336
337 In one sense, Event::Lib is more exact, as it uses the current time
338 regardless of any delays introduced by event processing. However,
339 most callbacks do not expect large delays in processing, so this
340 causes a higher drift (and a lot more system calls to get the
341 current time).
342
343 In another sense, EV is more exact, as your timer will be scheduled
344 at the same time, regardless of how long event processing actually
345 took.
346
347 In either case, if you care (and in most cases, you don't), then you
348 can get whatever behaviour you want with any event loop, by taking
349 the difference between "AnyEvent->time" and "AnyEvent->now" into
350 account.
351
352 AnyEvent->now_update
353 Some event loops (such as EV or AnyEvent::Loop) cache the current
354 time for each loop iteration (see the discussion of AnyEvent->now,
355 above).
356
357 When a callback runs for a long time (or when the process sleeps),
358 then this "current" time will differ substantially from the real
359 time, which might affect timers and time-outs.
360
361 When this is the case, you can call this method, which will update
362 the event loop's idea of "current time".
363
364 A typical example would be a script in a web server (e.g.
365 "mod_perl") - when mod_perl executes the script, then the event loop
366 will have the wrong idea about the "current time" (being potentially
367 far in the past, when the script ran the last time). In that case
368 you should arrange a call to "AnyEvent->now_update" each time the
369 web server process wakes up again (e.g. at the start of your script,
370 or in a handler).
371
372 Note that updating the time *might* cause some events to be handled.
373
225 SIGNAL WATCHERS 374 SIGNAL WATCHERS
375 $w = AnyEvent->signal (signal => <uppercase_signal_name>, cb => <callback>);
376
226 You can watch for signals using a signal watcher, "signal" is the signal 377 You can watch for signals using a signal watcher, "signal" is the signal
227 *name* without any "SIG" prefix, "cb" is the Perl callback to be invoked 378 *name* in uppercase and without any "SIG" prefix, "cb" is the Perl
228 whenever a signal occurs. 379 callback to be invoked whenever a signal occurs.
229 380
230 Although the callback might get passed parameters, their value and 381 Although the callback might get passed parameters, their value and
231 presence is undefined and you cannot rely on them. Portable AnyEvent 382 presence is undefined and you cannot rely on them. Portable AnyEvent
232 callbacks cannot use arguments passed to signal watcher callbacks. 383 callbacks cannot use arguments passed to signal watcher callbacks.
233 384
234 Multiple signal occurances can be clumped together into one callback 385 Multiple signal occurrences can be clumped together into one callback
235 invocation, and callback invocation will be synchronous. synchronous 386 invocation, and callback invocation will be synchronous. Synchronous
236 means that it might take a while until the signal gets handled by the 387 means that it might take a while until the signal gets handled by the
237 process, but it is guarenteed not to interrupt any other callbacks. 388 process, but it is guaranteed not to interrupt any other callbacks.
238 389
239 The main advantage of using these watchers is that you can share a 390 The main advantage of using these watchers is that you can share a
240 signal between multiple watchers. 391 signal between multiple watchers, and AnyEvent will ensure that signals
392 will not interrupt your program at bad times.
241 393
242 This watcher might use %SIG, so programs overwriting those signals 394 This watcher might use %SIG (depending on the event loop used), so
243 directly will likely not work correctly. 395 programs overwriting those signals directly will likely not work
396 correctly.
244 397
245 Example: exit on SIGINT 398 Example: exit on SIGINT
246 399
247 my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 }); 400 my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
248 401
402 Restart Behaviour
403 While restart behaviour is up to the event loop implementation, most
404 will not restart syscalls (that includes Async::Interrupt and AnyEvent's
405 pure perl implementation).
406
407 Safe/Unsafe Signals
408 Perl signals can be either "safe" (synchronous to opcode handling) or
409 "unsafe" (asynchronous) - the former might delay signal delivery
410 indefinitely, the latter might corrupt your memory.
411
412 AnyEvent signal handlers are, in addition, synchronous to the event
413 loop, i.e. they will not interrupt your running perl program but will
414 only be called as part of the normal event handling (just like timer,
415 I/O etc. callbacks, too).
416
417 Signal Races, Delays and Workarounds
418 Many event loops (e.g. Glib, Tk, Qt, IO::Async) do not support attaching
419 callbacks to signals in a generic way, which is a pity, as you cannot do
420 race-free signal handling in perl, requiring C libraries for this.
421 AnyEvent will try to do its best, which means in some cases, signals
422 will be delayed. The maximum time a signal might be delayed is 10
423 seconds by default, but can be overriden via
424 $ENV{PERL_ANYEVENT_MAX_SIGNAL_LATENCY} or $AnyEvent::MAX_SIGNAL_LATENCY
425 - see the "ENVIRONMENT VARIABLES" section for details.
426
427 All these problems can be avoided by installing the optional
428 Async::Interrupt module, which works with most event loops. It will not
429 work with inherently broken event loops such as Event or Event::Lib (and
430 not with POE currently). For those, you just have to suffer the delays.
431
249 CHILD PROCESS WATCHERS 432 CHILD PROCESS WATCHERS
433 $w = AnyEvent->child (pid => <process id>, cb => <callback>);
434
250 You can also watch on a child process exit and catch its exit status. 435 You can also watch for a child process exit and catch its exit status.
251 436
252 The child process is specified by the "pid" argument (if set to 0, it 437 The child process is specified by the "pid" argument (on some backends,
253 watches for any child process exit). The watcher will trigger as often 438 using 0 watches for any child process exit, on others this will croak).
254 as status change for the child are received. This works by installing a 439 The watcher will be triggered only when the child process has finished
255 signal handler for "SIGCHLD". The callback will be called with the pid 440 and an exit status is available, not on any trace events
256 and exit status (as returned by waitpid), so unlike other watcher types, 441 (stopped/continued).
257 you *can* rely on child watcher callback arguments. 442
443 The callback will be called with the pid and exit status (as returned by
444 waitpid), so unlike other watcher types, you *can* rely on child watcher
445 callback arguments.
446
447 This watcher type works by installing a signal handler for "SIGCHLD",
448 and since it cannot be shared, nothing else should use SIGCHLD or reap
449 random child processes (waiting for specific child processes, e.g.
450 inside "system", is just fine).
258 451
259 There is a slight catch to child watchers, however: you usually start 452 There is a slight catch to child watchers, however: you usually start
260 them *after* the child process was created, and this means the process 453 them *after* the child process was created, and this means the process
261 could have exited already (and no SIGCHLD will be sent anymore). 454 could have exited already (and no SIGCHLD will be sent anymore).
262 455
263 Not all event models handle this correctly (POE doesn't), but even for 456 Not all event models handle this correctly (neither POE nor IO::Async
457 do, see their AnyEvent::Impl manpages for details), but even for event
264 event models that *do* handle this correctly, they usually need to be 458 models that *do* handle this correctly, they usually need to be loaded
265 loaded before the process exits (i.e. before you fork in the first 459 before the process exits (i.e. before you fork in the first place).
266 place). 460 AnyEvent's pure perl event loop handles all cases correctly regardless
461 of when you start the watcher.
267 462
268 This means you cannot create a child watcher as the very first thing in 463 This means you cannot create a child watcher as the very first thing in
269 an AnyEvent program, you *have* to create at least one watcher before 464 an AnyEvent program, you *have* to create at least one watcher before
270 you "fork" the child (alternatively, you can call "AnyEvent::detect"). 465 you "fork" the child (alternatively, you can call "AnyEvent::detect").
271 466
467 As most event loops do not support waiting for child events, they will
468 be emulated by AnyEvent in most cases, in which case the latency and
469 race problems mentioned in the description of signal watchers apply.
470
272 Example: fork a process and wait for it 471 Example: fork a process and wait for it
273 472
274 my $done = AnyEvent->condvar; 473 my $done = AnyEvent->condvar;
275 474
276 AnyEvent::detect; # force event module to be initialised 475 # this forks and immediately calls exit in the child. this
277 476 # normally has all sorts of bad consequences for your parent,
477 # so take this as an example only. always fork and exec,
478 # or call POSIX::_exit, in real code.
278 my $pid = fork or exit 5; 479 my $pid = fork or exit 5;
279 480
280 my $w = AnyEvent->child ( 481 my $w = AnyEvent->child (
281 pid => $pid, 482 pid => $pid,
282 cb => sub { 483 cb => sub {
283 my ($pid, $status) = @_; 484 my ($pid, $status) = @_;
284 warn "pid $pid exited with status $status"; 485 warn "pid $pid exited with status $status";
285 $done->broadcast; 486 $done->send;
286 }, 487 },
287 ); 488 );
288 489
289 # do something else, then wait for process exit 490 # do something else, then wait for process exit
290 $done->wait; 491 $done->recv;
492
493 IDLE WATCHERS
494 $w = AnyEvent->idle (cb => <callback>);
495
496 This will repeatedly invoke the callback after the process becomes idle,
497 until either the watcher is destroyed or new events have been detected.
498
499 Idle watchers are useful when there is a need to do something, but it is
500 not so important (or wise) to do it instantly. The callback will be
501 invoked only when there is "nothing better to do", which is usually
502 defined as "all outstanding events have been handled and no new events
503 have been detected". That means that idle watchers ideally get invoked
504 when the event loop has just polled for new events but none have been
505 detected. Instead of blocking to wait for more events, the idle watchers
506 will be invoked.
507
508 Unfortunately, most event loops do not really support idle watchers
509 (only EV, Event and Glib do it in a usable fashion) - for the rest,
510 AnyEvent will simply call the callback "from time to time".
511
512 Example: read lines from STDIN, but only process them when the program
513 is otherwise idle:
514
515 my @lines; # read data
516 my $idle_w;
517 my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
518 push @lines, scalar <STDIN>;
519
520 # start an idle watcher, if not already done
521 $idle_w ||= AnyEvent->idle (cb => sub {
522 # handle only one line, when there are lines left
523 if (my $line = shift @lines) {
524 print "handled when idle: $line";
525 } else {
526 # otherwise disable the idle watcher again
527 undef $idle_w;
528 }
529 });
530 });
291 531
292 CONDITION VARIABLES 532 CONDITION VARIABLES
533 $cv = AnyEvent->condvar;
534
535 $cv->send (<list>);
536 my @res = $cv->recv;
537
538 If you are familiar with some event loops you will know that all of them
539 require you to run some blocking "loop", "run" or similar function that
540 will actively watch for new events and call your callbacks.
541
542 AnyEvent is slightly different: it expects somebody else to run the
543 event loop and will only block when necessary (usually when told by the
544 user).
545
546 The tool to do that is called a "condition variable", so called because
547 they represent a condition that must become true.
548
549 Now is probably a good time to look at the examples further below.
550
293 Condition variables can be created by calling the "AnyEvent->condvar" 551 Condition variables can be created by calling the "AnyEvent->condvar"
294 method without any arguments. 552 method, usually without arguments. The only argument pair allowed is
553 "cb", which specifies a callback to be called when the condition
554 variable becomes true, with the condition variable as the first argument
555 (but not the results).
295 556
296 A condition variable waits for a condition - precisely that the 557 After creation, the condition variable is "false" until it becomes
297 "->broadcast" method has been called. 558 "true" by calling the "send" method (or calling the condition variable
559 as if it were a callback, read about the caveats in the description for
560 the "->send" method).
298 561
299 They are very useful to signal that a condition has been fulfilled, for 562 Since condition variables are the most complex part of the AnyEvent API,
563 here are some different mental models of what they are - pick the ones
564 you can connect to:
565
566 * Condition variables are like callbacks - you can call them (and pass
567 them instead of callbacks). Unlike callbacks however, you can also
568 wait for them to be called.
569
570 * Condition variables are signals - one side can emit or send them,
571 the other side can wait for them, or install a handler that is
572 called when the signal fires.
573
574 * Condition variables are like "Merge Points" - points in your program
575 where you merge multiple independent results/control flows into one.
576
577 * Condition variables represent a transaction - functions that start
578 some kind of transaction can return them, leaving the caller the
579 choice between waiting in a blocking fashion, or setting a callback.
580
581 * Condition variables represent future values, or promises to deliver
582 some result, long before the result is available.
583
584 Condition variables are very useful to signal that something has
300 example, if you write a module that does asynchronous http requests, 585 finished, for example, if you write a module that does asynchronous http
301 then a condition variable would be the ideal candidate to signal the 586 requests, then a condition variable would be the ideal candidate to
302 availability of results. 587 signal the availability of results. The user can either act when the
588 callback is called or can synchronously "->recv" for the results.
303 589
304 You can also use condition variables to block your main program until an 590 You can also use them to simulate traditional event loops - for example,
305 event occurs - for example, you could "->wait" in your main program 591 you can block your main program until an event occurs - for example, you
306 until the user clicks the Quit button in your app, which would 592 could "->recv" in your main program until the user clicks the Quit
307 "->broadcast" the "quit" event. 593 button of your app, which would "->send" the "quit" event.
308 594
309 Note that condition variables recurse into the event loop - if you have 595 Note that condition variables recurse into the event loop - if you have
310 two pirces of code that call "->wait" in a round-robbin fashion, you 596 two pieces of code that call "->recv" in a round-robin fashion, you
311 lose. Therefore, condition variables are good to export to your caller, 597 lose. Therefore, condition variables are good to export to your caller,
312 but you should avoid making a blocking wait yourself, at least in 598 but you should avoid making a blocking wait yourself, at least in
313 callbacks, as this asks for trouble. 599 callbacks, as this asks for trouble.
314 600
315 This object has two methods: 601 Condition variables are represented by hash refs in perl, and the keys
602 used by AnyEvent itself are all named "_ae_XXX" to make subclassing easy
603 (it is often useful to build your own transaction class on top of
604 AnyEvent). To subclass, use "AnyEvent::CondVar" as base class and call
605 its "new" method in your own "new" method.
316 606
317 $cv->wait 607 There are two "sides" to a condition variable - the "producer side"
318 Wait (blocking if necessary) until the "->broadcast" method has been 608 which eventually calls "-> send", and the "consumer side", which waits
319 called on c<$cv>, while servicing other watchers normally. 609 for the send to occur.
320 610
321 You can only wait once on a condition - additional calls will return 611 Example: wait for a timer.
322 immediately.
323 612
324 Not all event models support a blocking wait - some die in that case 613 # condition: "wait till the timer is fired"
325 (programs might want to do that to stay interactive), so *if you are
326 using this from a module, never require a blocking wait*, but let
327 the caller decide whether the call will block or not (for example,
328 by coupling condition variables with some kind of request results
329 and supporting callbacks so the caller knows that getting the result
330 will not block, while still suppporting blocking waits if the caller
331 so desires).
332
333 Another reason *never* to "->wait" in a module is that you cannot
334 sensibly have two "->wait"'s in parallel, as that would require
335 multiple interpreters or coroutines/threads, none of which
336 "AnyEvent" can supply (the coroutine-aware backends
337 AnyEvent::Impl::CoroEV and AnyEvent::Impl::CoroEvent explicitly
338 support concurrent "->wait"'s from different coroutines, however).
339
340 $cv->broadcast
341 Flag the condition as ready - a running "->wait" and all further
342 calls to "wait" will (eventually) return after this method has been
343 called. If nobody is waiting the broadcast will be remembered..
344
345 Example:
346
347 # wait till the result is ready
348 my $result_ready = AnyEvent->condvar; 614 my $timer_fired = AnyEvent->condvar;
349 615
350 # do something such as adding a timer 616 # create the timer - we could wait for, say
351 # or socket watcher the calls $result_ready->broadcast 617 # a handle becomign ready, or even an
352 # when the "result" is ready. 618 # AnyEvent::HTTP request to finish, but
353 # in this case, we simply use a timer: 619 # in this case, we simply use a timer:
354 my $w = AnyEvent->timer ( 620 my $w = AnyEvent->timer (
355 after => 1, 621 after => 1,
356 cb => sub { $result_ready->broadcast }, 622 cb => sub { $timer_fired->send },
357 ); 623 );
358 624
359 # this "blocks" (while handling events) till the watcher 625 # this "blocks" (while handling events) till the callback
360 # calls broadcast 626 # calls ->send
361 $result_ready->wait; 627 $timer_fired->recv;
628
629 Example: wait for a timer, but take advantage of the fact that condition
630 variables are also callable directly.
631
632 my $done = AnyEvent->condvar;
633 my $delay = AnyEvent->timer (after => 5, cb => $done);
634 $done->recv;
635
636 Example: Imagine an API that returns a condvar and doesn't support
637 callbacks. This is how you make a synchronous call, for example from the
638 main program:
639
640 use AnyEvent::CouchDB;
641
642 ...
643
644 my @info = $couchdb->info->recv;
645
646 And this is how you would just set a callback to be called whenever the
647 results are available:
648
649 $couchdb->info->cb (sub {
650 my @info = $_[0]->recv;
651 });
652
653 METHODS FOR PRODUCERS
654 These methods should only be used by the producing side, i.e. the
655 code/module that eventually sends the signal. Note that it is also the
656 producer side which creates the condvar in most cases, but it isn't
657 uncommon for the consumer to create it as well.
658
659 $cv->send (...)
660 Flag the condition as ready - a running "->recv" and all further
661 calls to "recv" will (eventually) return after this method has been
662 called. If nobody is waiting the send will be remembered.
663
664 If a callback has been set on the condition variable, it is called
665 immediately from within send.
666
667 Any arguments passed to the "send" call will be returned by all
668 future "->recv" calls.
669
670 Condition variables are overloaded so one can call them directly (as
671 if they were a code reference). Calling them directly is the same as
672 calling "send".
673
674 $cv->croak ($error)
675 Similar to send, but causes all calls to "->recv" to invoke
676 "Carp::croak" with the given error message/object/scalar.
677
678 This can be used to signal any errors to the condition variable
679 user/consumer. Doing it this way instead of calling "croak" directly
680 delays the error detection, but has the overwhelming advantage that
681 it diagnoses the error at the place where the result is expected,
682 and not deep in some event callback with no connection to the actual
683 code causing the problem.
684
685 $cv->begin ([group callback])
686 $cv->end
687 These two methods can be used to combine many transactions/events
688 into one. For example, a function that pings many hosts in parallel
689 might want to use a condition variable for the whole process.
690
691 Every call to "->begin" will increment a counter, and every call to
692 "->end" will decrement it. If the counter reaches 0 in "->end", the
693 (last) callback passed to "begin" will be executed, passing the
694 condvar as first argument. That callback is *supposed* to call
695 "->send", but that is not required. If no group callback was set,
696 "send" will be called without any arguments.
697
698 You can think of "$cv->send" giving you an OR condition (one call
699 sends), while "$cv->begin" and "$cv->end" giving you an AND
700 condition (all "begin" calls must be "end"'ed before the condvar
701 sends).
702
703 Let's start with a simple example: you have two I/O watchers (for
704 example, STDOUT and STDERR for a program), and you want to wait for
705 both streams to close before activating a condvar:
706
707 my $cv = AnyEvent->condvar;
708
709 $cv->begin; # first watcher
710 my $w1 = AnyEvent->io (fh => $fh1, cb => sub {
711 defined sysread $fh1, my $buf, 4096
712 or $cv->end;
713 });
714
715 $cv->begin; # second watcher
716 my $w2 = AnyEvent->io (fh => $fh2, cb => sub {
717 defined sysread $fh2, my $buf, 4096
718 or $cv->end;
719 });
720
721 $cv->recv;
722
723 This works because for every event source (EOF on file handle),
724 there is one call to "begin", so the condvar waits for all calls to
725 "end" before sending.
726
727 The ping example mentioned above is slightly more complicated, as
728 the there are results to be passed back, and the number of tasks
729 that are begun can potentially be zero:
730
731 my $cv = AnyEvent->condvar;
732
733 my %result;
734 $cv->begin (sub { shift->send (\%result) });
735
736 for my $host (@list_of_hosts) {
737 $cv->begin;
738 ping_host_then_call_callback $host, sub {
739 $result{$host} = ...;
740 $cv->end;
741 };
742 }
743
744 $cv->end;
745
746 ...
747
748 my $results = $cv->recv;
749
750 This code fragment supposedly pings a number of hosts and calls
751 "send" after results for all then have have been gathered - in any
752 order. To achieve this, the code issues a call to "begin" when it
753 starts each ping request and calls "end" when it has received some
754 result for it. Since "begin" and "end" only maintain a counter, the
755 order in which results arrive is not relevant.
756
757 There is an additional bracketing call to "begin" and "end" outside
758 the loop, which serves two important purposes: first, it sets the
759 callback to be called once the counter reaches 0, and second, it
760 ensures that "send" is called even when "no" hosts are being pinged
761 (the loop doesn't execute once).
762
763 This is the general pattern when you "fan out" into multiple (but
764 potentially zero) subrequests: use an outer "begin"/"end" pair to
765 set the callback and ensure "end" is called at least once, and then,
766 for each subrequest you start, call "begin" and for each subrequest
767 you finish, call "end".
768
769 METHODS FOR CONSUMERS
770 These methods should only be used by the consuming side, i.e. the code
771 awaits the condition.
772
773 $cv->recv
774 Wait (blocking if necessary) until the "->send" or "->croak" methods
775 have been called on $cv, while servicing other watchers normally.
776
777 You can only wait once on a condition - additional calls are valid
778 but will return immediately.
779
780 If an error condition has been set by calling "->croak", then this
781 function will call "croak".
782
783 In list context, all parameters passed to "send" will be returned,
784 in scalar context only the first one will be returned.
785
786 Note that doing a blocking wait in a callback is not supported by
787 any event loop, that is, recursive invocation of a blocking "->recv"
788 is not allowed and the "recv" call will "croak" if such a condition
789 is detected. This requirement can be dropped by relying on
790 Coro::AnyEvent , which allows you to do a blocking "->recv" from any
791 thread that doesn't run the event loop itself. Coro::AnyEvent is
792 loaded automatically when Coro is used with AnyEvent, so code does
793 not need to do anything special to take advantage of that: any code
794 that would normally block your program because it calls "recv", be
795 executed in an "async" thread instead without blocking other
796 threads.
797
798 Not all event models support a blocking wait - some die in that case
799 (programs might want to do that to stay interactive), so *if you are
800 using this from a module, never require a blocking wait*. Instead,
801 let the caller decide whether the call will block or not (for
802 example, by coupling condition variables with some kind of request
803 results and supporting callbacks so the caller knows that getting
804 the result will not block, while still supporting blocking waits if
805 the caller so desires).
806
807 You can ensure that "->recv" never blocks by setting a callback and
808 only calling "->recv" from within that callback (or at a later
809 time). This will work even when the event loop does not support
810 blocking waits otherwise.
811
812 $bool = $cv->ready
813 Returns true when the condition is "true", i.e. whether "send" or
814 "croak" have been called.
815
816 $cb = $cv->cb ($cb->($cv))
817 This is a mutator function that returns the callback set and
818 optionally replaces it before doing so.
819
820 The callback will be called when the condition becomes "true", i.e.
821 when "send" or "croak" are called, with the only argument being the
822 condition variable itself. If the condition is already true, the
823 callback is called immediately when it is set. Calling "recv" inside
824 the callback or at any later time is guaranteed not to block.
825
826SUPPORTED EVENT LOOPS/BACKENDS
827 The available backend classes are (every class has its own manpage):
828
829 Backends that are autoprobed when no other event loop can be found.
830 EV is the preferred backend when no other event loop seems to be in
831 use. If EV is not installed, then AnyEvent will fall back to its own
832 pure-perl implementation, which is available everywhere as it comes
833 with AnyEvent itself.
834
835 AnyEvent::Impl::EV based on EV (interface to libev, best choice).
836 AnyEvent::Impl::Perl pure-perl AnyEvent::Loop, fast and portable.
837
838 Backends that are transparently being picked up when they are used.
839 These will be used if they are already loaded when the first watcher
840 is created, in which case it is assumed that the application is
841 using them. This means that AnyEvent will automatically pick the
842 right backend when the main program loads an event module before
843 anything starts to create watchers. Nothing special needs to be done
844 by the main program.
845
846 AnyEvent::Impl::Event based on Event, very stable, few glitches.
847 AnyEvent::Impl::Glib based on Glib, slow but very stable.
848 AnyEvent::Impl::Tk based on Tk, very broken.
849 AnyEvent::Impl::UV based on UV, innovated square wheels.
850 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
851 AnyEvent::Impl::POE based on POE, very slow, some limitations.
852 AnyEvent::Impl::Irssi used when running within irssi.
853 AnyEvent::Impl::IOAsync based on IO::Async.
854 AnyEvent::Impl::Cocoa based on Cocoa::EventLoop.
855 AnyEvent::Impl::FLTK based on FLTK (fltk 2 binding).
856
857 Backends with special needs.
858 Qt requires the Qt::Application to be instantiated first, but will
859 otherwise be picked up automatically. As long as the main program
860 instantiates the application before any AnyEvent watchers are
861 created, everything should just work.
862
863 AnyEvent::Impl::Qt based on Qt.
864
865 Event loops that are indirectly supported via other backends.
866 Some event loops can be supported via other modules:
867
868 There is no direct support for WxWidgets (Wx) or Prima.
869
870 WxWidgets has no support for watching file handles. However, you can
871 use WxWidgets through the POE adaptor, as POE has a Wx backend that
872 simply polls 20 times per second, which was considered to be too
873 horrible to even consider for AnyEvent.
874
875 Prima is not supported as nobody seems to be using it, but it has a
876 POE backend, so it can be supported through POE.
877
878 AnyEvent knows about both Prima and Wx, however, and will try to
879 load POE when detecting them, in the hope that POE will pick them
880 up, in which case everything will be automatic.
362 881
363GLOBAL VARIABLES AND FUNCTIONS 882GLOBAL VARIABLES AND FUNCTIONS
883 These are not normally required to use AnyEvent, but can be useful to
884 write AnyEvent extension modules.
885
364 $AnyEvent::MODEL 886 $AnyEvent::MODEL
365 Contains "undef" until the first watcher is being created. Then it 887 Contains "undef" until the first watcher is being created, before
888 the backend has been autodetected.
889
366 contains the event model that is being used, which is the name of 890 Afterwards it contains the event model that is being used, which is
367 the Perl class implementing the model. This class is usually one of 891 the name of the Perl class implementing the model. This class is
368 the "AnyEvent::Impl:xxx" modules, but can be any other class in the 892 usually one of the "AnyEvent::Impl::xxx" modules, but can be any
369 case AnyEvent has been extended at runtime (e.g. in *rxvt-unicode*). 893 other class in the case AnyEvent has been extended at runtime (e.g.
370 894 in *rxvt-unicode* it will be "urxvt::anyevent").
371 The known classes so far are:
372
373 AnyEvent::Impl::CoroEV based on Coro::EV, best choice.
374 AnyEvent::Impl::CoroEvent based on Coro::Event, second best choice.
375 AnyEvent::Impl::EV based on EV (an interface to libev, best choice).
376 AnyEvent::Impl::Event based on Event, second best choice.
377 AnyEvent::Impl::Glib based on Glib, third-best choice.
378 AnyEvent::Impl::Perl pure-perl implementation, inefficient but portable.
379 AnyEvent::Impl::Tk based on Tk, very bad choice.
380 AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).
381 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
382 AnyEvent::Impl::POE based on POE, not generic enough for full support.
383
384 There is no support for WxWidgets, as WxWidgets has no support for
385 watching file handles. However, you can use WxWidgets through the
386 POE Adaptor, as POE has a Wx backend that simply polls 20 times per
387 second, which was considered to be too horrible to even consider for
388 AnyEvent. Likewise, other POE backends can be used by AnyEvent by
389 using it's adaptor.
390
391 AnyEvent knows about Prima and Wx and will try to use POE when
392 autodetecting them.
393 895
394 AnyEvent::detect 896 AnyEvent::detect
395 Returns $AnyEvent::MODEL, forcing autodetection of the event model 897 Returns $AnyEvent::MODEL, forcing autodetection of the event model
396 if necessary. You should only call this function right before you 898 if necessary. You should only call this function right before you
397 would have created an AnyEvent watcher anyway, that is, as late as 899 would have created an AnyEvent watcher anyway, that is, as late as
398 possible at runtime. 900 possible at runtime, and not e.g. during initialisation of your
901 module.
902
903 The effect of calling this function is as if a watcher had been
904 created (specifically, actions that happen "when the first watcher
905 is created" happen when calling detetc as well).
906
907 If you need to do some initialisation before AnyEvent watchers are
908 created, use "post_detect".
909
910 $guard = AnyEvent::post_detect { BLOCK }
911 Arranges for the code block to be executed as soon as the event
912 model is autodetected (or immediately if that has already happened).
913
914 The block will be executed *after* the actual backend has been
915 detected ($AnyEvent::MODEL is set), but *before* any watchers have
916 been created, so it is possible to e.g. patch @AnyEvent::ISA or do
917 other initialisations - see the sources of AnyEvent::Strict or
918 AnyEvent::AIO to see how this is used.
919
920 The most common usage is to create some global watchers, without
921 forcing event module detection too early, for example, AnyEvent::AIO
922 creates and installs the global IO::AIO watcher in a "post_detect"
923 block to avoid autodetecting the event module at load time.
924
925 If called in scalar or list context, then it creates and returns an
926 object that automatically removes the callback again when it is
927 destroyed (or "undef" when the hook was immediately executed). See
928 AnyEvent::AIO for a case where this is useful.
929
930 Example: Create a watcher for the IO::AIO module and store it in
931 $WATCHER, but do so only do so after the event loop is initialised.
932
933 our WATCHER;
934
935 my $guard = AnyEvent::post_detect {
936 $WATCHER = AnyEvent->io (fh => IO::AIO::poll_fileno, poll => 'r', cb => \&IO::AIO::poll_cb);
937 };
938
939 # the ||= is important in case post_detect immediately runs the block,
940 # as to not clobber the newly-created watcher. assigning both watcher and
941 # post_detect guard to the same variable has the advantage of users being
942 # able to just C<undef $WATCHER> if the watcher causes them grief.
943
944 $WATCHER ||= $guard;
945
946 @AnyEvent::post_detect
947 If there are any code references in this array (you can "push" to it
948 before or after loading AnyEvent), then they will be called directly
949 after the event loop has been chosen.
950
951 You should check $AnyEvent::MODEL before adding to this array,
952 though: if it is defined then the event loop has already been
953 detected, and the array will be ignored.
954
955 Best use "AnyEvent::post_detect { BLOCK }" when your application
956 allows it, as it takes care of these details.
957
958 This variable is mainly useful for modules that can do something
959 useful when AnyEvent is used and thus want to know when it is
960 initialised, but do not need to even load it by default. This array
961 provides the means to hook into AnyEvent passively, without loading
962 it.
963
964 Example: To load Coro::AnyEvent whenever Coro and AnyEvent are used
965 together, you could put this into Coro (this is the actual code used
966 by Coro to accomplish this):
967
968 if (defined $AnyEvent::MODEL) {
969 # AnyEvent already initialised, so load Coro::AnyEvent
970 require Coro::AnyEvent;
971 } else {
972 # AnyEvent not yet initialised, so make sure to load Coro::AnyEvent
973 # as soon as it is
974 push @AnyEvent::post_detect, sub { require Coro::AnyEvent };
975 }
976
977 AnyEvent::postpone { BLOCK }
978 Arranges for the block to be executed as soon as possible, but not
979 before the call itself returns. In practise, the block will be
980 executed just before the event loop polls for new events, or shortly
981 afterwards.
982
983 This function never returns anything (to make the "return postpone {
984 ... }" idiom more useful.
985
986 To understand the usefulness of this function, consider a function
987 that asynchronously does something for you and returns some
988 transaction object or guard to let you cancel the operation. For
989 example, "AnyEvent::Socket::tcp_connect":
990
991 # start a connection attempt unless one is active
992 $self->{connect_guard} ||= AnyEvent::Socket::tcp_connect "www.example.net", 80, sub {
993 delete $self->{connect_guard};
994 ...
995 };
996
997 Imagine that this function could instantly call the callback, for
998 example, because it detects an obvious error such as a negative port
999 number. Invoking the callback before the function returns causes
1000 problems however: the callback will be called and will try to delete
1001 the guard object. But since the function hasn't returned yet, there
1002 is nothing to delete. When the function eventually returns it will
1003 assign the guard object to "$self->{connect_guard}", where it will
1004 likely never be deleted, so the program thinks it is still trying to
1005 connect.
1006
1007 This is where "AnyEvent::postpone" should be used. Instead of
1008 calling the callback directly on error:
1009
1010 $cb->(undef), return # signal error to callback, BAD!
1011 if $some_error_condition;
1012
1013 It should use "postpone":
1014
1015 AnyEvent::postpone { $cb->(undef) }, return # signal error to callback, later
1016 if $some_error_condition;
1017
1018 AnyEvent::log $level, $msg[, @args]
1019 Log the given $msg at the given $level.
1020
1021 If AnyEvent::Log is not loaded then this function makes a simple
1022 test to see whether the message will be logged. If the test succeeds
1023 it will load AnyEvent::Log and call "AnyEvent::Log::log" -
1024 consequently, look at the AnyEvent::Log documentation for details.
1025
1026 If the test fails it will simply return. Right now this happens when
1027 a numerical loglevel is used and it is larger than the level
1028 specified via $ENV{PERL_ANYEVENT_VERBOSE}.
1029
1030 If you want to sprinkle loads of logging calls around your code,
1031 consider creating a logger callback with the "AnyEvent::Log::logger"
1032 function, which can reduce typing, codesize and can reduce the
1033 logging overhead enourmously.
1034
1035 AnyEvent::fh_block $filehandle
1036 AnyEvent::fh_unblock $filehandle
1037 Sets blocking or non-blocking behaviour for the given filehandle.
399 1038
400WHAT TO DO IN A MODULE 1039WHAT TO DO IN A MODULE
401 As a module author, you should "use AnyEvent" and call AnyEvent methods 1040 As a module author, you should "use AnyEvent" and call AnyEvent methods
402 freely, but you should not load a specific event module or rely on it. 1041 freely, but you should not load a specific event module or rely on it.
403 1042
404 Be careful when you create watchers in the module body - AnyEvent will 1043 Be careful when you create watchers in the module body - AnyEvent will
405 decide which event module to use as soon as the first method is called, 1044 decide which event module to use as soon as the first method is called,
406 so by calling AnyEvent in your module body you force the user of your 1045 so by calling AnyEvent in your module body you force the user of your
407 module to load the event module first. 1046 module to load the event module first.
408 1047
409 Never call "->wait" on a condition variable unless you *know* that the 1048 Never call "->recv" on a condition variable unless you *know* that the
410 "->broadcast" method has been called on it already. This is because it 1049 "->send" method has been called on it already. This is because it will
411 will stall the whole program, and the whole point of using events is to 1050 stall the whole program, and the whole point of using events is to stay
412 stay interactive. 1051 interactive.
413 1052
414 It is fine, however, to call "->wait" when the user of your module 1053 It is fine, however, to call "->recv" when the user of your module
415 requests it (i.e. if you create a http request object ad have a method 1054 requests it (i.e. if you create a http request object ad have a method
416 called "results" that returns the results, it should call "->wait" 1055 called "results" that returns the results, it may call "->recv" freely,
417 freely, as the user of your module knows what she is doing. always). 1056 as the user of your module knows what she is doing. Always).
418 1057
419WHAT TO DO IN THE MAIN PROGRAM 1058WHAT TO DO IN THE MAIN PROGRAM
420 There will always be a single main program - the only place that should 1059 There will always be a single main program - the only place that should
421 dictate which event model to use. 1060 dictate which event model to use.
422 1061
423 If it doesn't care, it can just "use AnyEvent" and use it itself, or not 1062 If the program is not event-based, it need not do anything special, even
424 do anything special (it does not need to be event-based) and let 1063 when it depends on a module that uses an AnyEvent. If the program itself
425 AnyEvent decide which implementation to chose if some module relies on 1064 uses AnyEvent, but does not care which event loop is used, all it needs
426 it. 1065 to do is "use AnyEvent". In either case, AnyEvent will choose the best
1066 available loop implementation.
427 1067
428 If the main program relies on a specific event model. For example, in 1068 If the main program relies on a specific event model - for example, in
429 Gtk2 programs you have to rely on the Glib module. You should load the 1069 Gtk2 programs you have to rely on the Glib module - you should load the
430 event module before loading AnyEvent or any module that uses it: 1070 event module before loading AnyEvent or any module that uses it:
431 generally speaking, you should load it as early as possible. The reason 1071 generally speaking, you should load it as early as possible. The reason
432 is that modules might create watchers when they are loaded, and AnyEvent 1072 is that modules might create watchers when they are loaded, and AnyEvent
433 will decide on the event model to use as soon as it creates watchers, 1073 will decide on the event model to use as soon as it creates watchers,
434 and it might chose the wrong one unless you load the correct one 1074 and it might choose the wrong one unless you load the correct one
435 yourself. 1075 yourself.
436 1076
437 You can chose to use a rather inefficient pure-perl implementation by 1077 You can chose to use a pure-perl implementation by loading the
438 loading the "AnyEvent::Impl::Perl" module, which gives you similar 1078 "AnyEvent::Loop" module, which gives you similar behaviour everywhere,
439 behaviour everywhere, but letting AnyEvent chose is generally better. 1079 but letting AnyEvent chose the model is generally better.
1080
1081 MAINLOOP EMULATION
1082 Sometimes (often for short test scripts, or even standalone programs who
1083 only want to use AnyEvent), you do not want to run a specific event
1084 loop.
1085
1086 In that case, you can use a condition variable like this:
1087
1088 AnyEvent->condvar->recv;
1089
1090 This has the effect of entering the event loop and looping forever.
1091
1092 Note that usually your program has some exit condition, in which case it
1093 is better to use the "traditional" approach of storing a condition
1094 variable somewhere, waiting for it, and sending it when the program
1095 should exit cleanly.
440 1096
441OTHER MODULES 1097OTHER MODULES
442 The following is a non-exhaustive list of additional modules that use 1098 The following is a non-exhaustive list of additional modules that use
443 AnyEvent and can therefore be mixed easily with other AnyEvent modules 1099 AnyEvent as a client and can therefore be mixed easily with other
444 in the same program. Some of the modules come with AnyEvent, some are 1100 AnyEvent modules and other event loops in the same program. Some of the
445 available via CPAN. 1101 modules come as part of AnyEvent, the others are available via CPAN (see
1102 <http://search.cpan.org/search?m=module&q=anyevent%3A%3A*> for a longer
1103 non-exhaustive list), and the list is heavily biased towards modules of
1104 the AnyEvent author himself :)
446 1105
447 AnyEvent::Util 1106 AnyEvent::Util (part of the AnyEvent distribution)
448 Contains various utility functions that replace often-used but 1107 Contains various utility functions that replace often-used blocking
449 blocking functions such as "inet_aton" by event-/callback-based 1108 functions such as "inet_aton" with event/callback-based versions.
450 versions.
451 1109
452 AnyEvent::Handle 1110 AnyEvent::Socket (part of the AnyEvent distribution)
1111 Provides various utility functions for (internet protocol) sockets,
1112 addresses and name resolution. Also functions to create non-blocking
1113 tcp connections or tcp servers, with IPv6 and SRV record support and
1114 more.
1115
1116 AnyEvent::Handle (part of the AnyEvent distribution)
453 Provide read and write buffers and manages watchers for reads and 1117 Provide read and write buffers, manages watchers for reads and
1118 writes, supports raw and formatted I/O, I/O queued and fully
1119 transparent and non-blocking SSL/TLS (via AnyEvent::TLS).
1120
1121 AnyEvent::DNS (part of the AnyEvent distribution)
1122 Provides rich asynchronous DNS resolver capabilities.
1123
1124 AnyEvent::HTTP, AnyEvent::IRC, AnyEvent::XMPP, AnyEvent::GPSD,
1125 AnyEvent::IGS, AnyEvent::FCP
1126 Implement event-based interfaces to the protocols of the same name
1127 (for the curious, IGS is the International Go Server and FCP is the
1128 Freenet Client Protocol).
1129
1130 AnyEvent::AIO (part of the AnyEvent distribution)
1131 Truly asynchronous (as opposed to non-blocking) I/O, should be in
1132 the toolbox of every event programmer. AnyEvent::AIO transparently
1133 fuses IO::AIO and AnyEvent together, giving AnyEvent access to
1134 event-based file I/O, and much more.
1135
1136 AnyEvent::Fork, AnyEvent::Fork::RPC, AnyEvent::Fork::Pool,
1137 AnyEvent::Fork::Remote
1138 These let you safely fork new subprocesses, either locally or
1139 remotely (e.g.v ia ssh), using some RPC protocol or not, without the
1140 limitations normally imposed by fork (AnyEvent works fine for
1141 example). Dynamically-resized worker pools are obviously included as
454 writes. 1142 well.
455 1143
456 AnyEvent::Socket 1144 And they are quite tiny and fast as well - "abusing" AnyEvent::Fork
457 Provides a means to do non-blocking connects, accepts etc. 1145 just to exec external programs can easily beat using "fork" and
1146 "exec" (or even "system") in most programs.
458 1147
459 AnyEvent::HTTPD 1148 AnyEvent::Filesys::Notify
460 Provides a simple web application server framework. 1149 AnyEvent is good for non-blocking stuff, but it can't detect file or
1150 path changes (e.g. "watch this directory for new files", "watch this
1151 file for changes"). The AnyEvent::Filesys::Notify module promises to
1152 do just that in a portbale fashion, supporting inotify on GNU/Linux
1153 and some weird, without doubt broken, stuff on OS X to monitor
1154 files. It can fall back to blocking scans at regular intervals
1155 transparently on other platforms, so it's about as portable as it
1156 gets.
461 1157
1158 (I haven't used it myself, but it seems the biggest problem with it
1159 is it quite bad performance).
1160
462 AnyEvent::DNS 1161 AnyEvent::DBI
463 Provides asynchronous DNS resolver capabilities, beyond what 1162 Executes DBI requests asynchronously in a proxy process for you,
464 AnyEvent::Util offers. 1163 notifying you in an event-based way when the operation is finished.
465 1164
466 AnyEvent::FastPing 1165 AnyEvent::FastPing
467 The fastest ping in the west. 1166 The fastest ping in the west.
468 1167
469 Net::IRC3
470 AnyEvent based IRC client module family.
471
472 Net::XMPP2
473 AnyEvent based XMPP (Jabber protocol) module family.
474
475 Net::FCP
476 AnyEvent-based implementation of the Freenet Client Protocol,
477 birthplace of AnyEvent.
478
479 Event::ExecFlow
480 High level API for event-based execution flow control.
481
482 Coro 1168 Coro
483 Has special support for AnyEvent. 1169 Has special support for AnyEvent via Coro::AnyEvent, which allows
1170 you to simply invert the flow control - don't call us, we will call
1171 you:
484 1172
485 IO::Lambda 1173 async {
486 The lambda approach to I/O - don't ask, look there. Can use 1174 Coro::AnyEvent::sleep 5; # creates a 5s timer and waits for it
1175 print "5 seconds later!\n";
1176
1177 Coro::AnyEvent::readable *STDIN; # uses an I/O watcher
1178 my $line = <STDIN>; # works for ttys
1179
1180 AnyEvent::HTTP::http_get "url", Coro::rouse_cb;
1181 my ($body, $hdr) = Coro::rouse_wait;
1182 };
1183
1184SIMPLIFIED AE API
1185 Starting with version 5.0, AnyEvent officially supports a second, much
1186 simpler, API that is designed to reduce the calling, typing and memory
1187 overhead by using function call syntax and a fixed number of parameters.
1188
1189 See the AE manpage for details.
1190
1191ERROR AND EXCEPTION HANDLING
1192 In general, AnyEvent does not do any error handling - it relies on the
1193 caller to do that if required. The AnyEvent::Strict module (see also the
1194 "PERL_ANYEVENT_STRICT" environment variable, below) provides strict
1195 checking of all AnyEvent methods, however, which is highly useful during
1196 development.
1197
1198 As for exception handling (i.e. runtime errors and exceptions thrown
1199 while executing a callback), this is not only highly event-loop
1200 specific, but also not in any way wrapped by this module, as this is the
1201 job of the main program.
1202
1203 The pure perl event loop simply re-throws the exception (usually within
1204 "condvar->recv"), the Event and EV modules call "$Event/EV::DIED->()",
1205 Glib uses "install_exception_handler" and so on.
1206
1207ENVIRONMENT VARIABLES
1208 AnyEvent supports a number of environment variables that tune the
1209 runtime behaviour. They are usually evaluated when AnyEvent is loaded,
1210 initialised, or a submodule that uses them is loaded. Many of them also
1211 cause AnyEvent to load additional modules - for example,
1212 "PERL_ANYEVENT_DEBUG_WRAP" causes the AnyEvent::Debug module to be
1213 loaded.
1214
1215 All the environment variables documented here start with
1216 "PERL_ANYEVENT_", which is what AnyEvent considers its own namespace.
1217 Other modules are encouraged (but by no means required) to use
1218 "PERL_ANYEVENT_SUBMODULE" if they have registered the
1219 AnyEvent::Submodule namespace on CPAN, for any submodule. For example,
1220 AnyEvent::HTTP could be expected to use "PERL_ANYEVENT_HTTP_PROXY" (it
1221 should not access env variables starting with "AE_", see below).
1222
1223 All variables can also be set via the "AE_" prefix, that is, instead of
1224 setting "PERL_ANYEVENT_VERBOSE" you can also set "AE_VERBOSE". In case
1225 there is a clash btween anyevent and another program that uses
1226 "AE_something" you can set the corresponding "PERL_ANYEVENT_something"
1227 variable to the empty string, as those variables take precedence.
1228
1229 When AnyEvent is first loaded, it copies all "AE_xxx" env variables to
1230 their "PERL_ANYEVENT_xxx" counterpart unless that variable already
1231 exists. If taint mode is on, then AnyEvent will remove *all* environment
1232 variables starting with "PERL_ANYEVENT_" from %ENV (or replace them with
1233 "undef" or the empty string, if the corresaponding "AE_" variable is
1234 set).
1235
1236 The exact algorithm is currently:
1237
1238 1. if taint mode enabled, delete all PERL_ANYEVENT_xyz variables from %ENV
1239 2. copy over AE_xyz to PERL_ANYEVENT_xyz unless the latter alraedy exists
1240 3. if taint mode enabled, set all PERL_ANYEVENT_xyz variables to undef.
1241
1242 This ensures that child processes will not see the "AE_" variables.
1243
1244 The following environment variables are currently known to AnyEvent:
1245
1246 "PERL_ANYEVENT_VERBOSE"
1247 By default, AnyEvent will log messages with loglevel 4 ("error") or
1248 higher (see AnyEvent::Log). You can set this environment variable to
1249 a numerical loglevel to make AnyEvent more (or less) talkative.
1250
1251 If you want to do more than just set the global logging level you
1252 should have a look at "PERL_ANYEVENT_LOG", which allows much more
1253 complex specifications.
1254
1255 When set to 0 ("off"), then no messages whatsoever will be logged
1256 with everything else at defaults.
1257
1258 When set to 5 or higher ("warn"), AnyEvent warns about unexpected
1259 conditions, such as not being able to load the event model specified
1260 by "PERL_ANYEVENT_MODEL", or a guard callback throwing an exception
1261 - this is the minimum recommended level for use during development.
1262
1263 When set to 7 or higher (info), AnyEvent reports which event model
1264 it chooses.
1265
1266 When set to 8 or higher (debug), then AnyEvent will report extra
1267 information on which optional modules it loads and how it implements
1268 certain features.
1269
1270 "PERL_ANYEVENT_LOG"
1271 Accepts rather complex logging specifications. For example, you
1272 could log all "debug" messages of some module to stderr, warnings
1273 and above to stderr, and errors and above to syslog, with:
1274
1275 PERL_ANYEVENT_LOG=Some::Module=debug,+log:filter=warn,+%syslog:%syslog=error,syslog
1276
1277 For the rather extensive details, see AnyEvent::Log.
1278
1279 This variable is evaluated when AnyEvent (or AnyEvent::Log) is
1280 loaded, so will take effect even before AnyEvent has initialised
1281 itself.
1282
1283 Note that specifying this environment variable causes the
1284 AnyEvent::Log module to be loaded, while "PERL_ANYEVENT_VERBOSE"
1285 does not, so only using the latter saves a few hundred kB of memory
1286 unless a module explicitly needs the extra features of
487 AnyEvent. 1287 AnyEvent::Log.
488 1288
489 IO::AIO 1289 "PERL_ANYEVENT_STRICT"
490 Truly asynchronous I/O, should be in the toolbox of every event 1290 AnyEvent does not do much argument checking by default, as thorough
491 programmer. Can be trivially made to use AnyEvent. 1291 argument checking is very costly. Setting this variable to a true
1292 value will cause AnyEvent to load "AnyEvent::Strict" and then to
1293 thoroughly check the arguments passed to most method calls. If it
1294 finds any problems, it will croak.
492 1295
493 BDB Truly asynchronous Berkeley DB access. Can be trivially made to use 1296 In other words, enables "strict" mode.
494 AnyEvent. 1297
1298 Unlike "use strict" (or its modern cousin, "use common::sense", it
1299 is definitely recommended to keep it off in production. Keeping
1300 "PERL_ANYEVENT_STRICT=1" in your environment while developing
1301 programs can be very useful, however.
1302
1303 "PERL_ANYEVENT_DEBUG_SHELL"
1304 If this env variable is nonempty, then its contents will be
1305 interpreted by "AnyEvent::Socket::parse_hostport" and
1306 "AnyEvent::Debug::shell" (after replacing every occurance of $$ by
1307 the process pid). The shell object is saved in
1308 $AnyEvent::Debug::SHELL.
1309
1310 This happens when the first watcher is created.
1311
1312 For example, to bind a debug shell on a unix domain socket in
1313 /tmp/debug<pid>.sock, you could use this:
1314
1315 PERL_ANYEVENT_DEBUG_SHELL=/tmp/debug\$\$.sock perlprog
1316 # connect with e.g.: socat readline /tmp/debug123.sock
1317
1318 Or to bind to tcp port 4545 on localhost:
1319
1320 PERL_ANYEVENT_DEBUG_SHELL=127.0.0.1:4545 perlprog
1321 # connect with e.g.: telnet localhost 4545
1322
1323 Note that creating sockets in /tmp or on localhost is very unsafe on
1324 multiuser systems.
1325
1326 "PERL_ANYEVENT_DEBUG_WRAP"
1327 Can be set to 0, 1 or 2 and enables wrapping of all watchers for
1328 debugging purposes. See "AnyEvent::Debug::wrap" for details.
1329
1330 "PERL_ANYEVENT_MODEL"
1331 This can be used to specify the event model to be used by AnyEvent,
1332 before auto detection and -probing kicks in.
1333
1334 It normally is a string consisting entirely of ASCII letters (e.g.
1335 "EV" or "IOAsync"). The string "AnyEvent::Impl::" gets prepended and
1336 the resulting module name is loaded and - if the load was successful
1337 - used as event model backend. If it fails to load then AnyEvent
1338 will proceed with auto detection and -probing.
1339
1340 If the string ends with "::" instead (e.g. "AnyEvent::Impl::EV::")
1341 then nothing gets prepended and the module name is used as-is (hint:
1342 "::" at the end of a string designates a module name and quotes it
1343 appropriately).
1344
1345 For example, to force the pure perl model (AnyEvent::Loop::Perl) you
1346 could start your program like this:
1347
1348 PERL_ANYEVENT_MODEL=Perl perl ...
1349
1350 "PERL_ANYEVENT_IO_MODEL"
1351 The current file I/O model - see AnyEvent::IO for more info.
1352
1353 At the moment, only "Perl" (small, pure-perl, synchronous) and
1354 "IOAIO" (truly asynchronous) are supported. The default is "IOAIO"
1355 if AnyEvent::AIO can be loaded, otherwise it is "Perl".
1356
1357 "PERL_ANYEVENT_PROTOCOLS"
1358 Used by both AnyEvent::DNS and AnyEvent::Socket to determine
1359 preferences for IPv4 or IPv6. The default is unspecified (and might
1360 change, or be the result of auto probing).
1361
1362 Must be set to a comma-separated list of protocols or address
1363 families, current supported: "ipv4" and "ipv6". Only protocols
1364 mentioned will be used, and preference will be given to protocols
1365 mentioned earlier in the list.
1366
1367 This variable can effectively be used for denial-of-service attacks
1368 against local programs (e.g. when setuid), although the impact is
1369 likely small, as the program has to handle connection and other
1370 failures anyways.
1371
1372 Examples: "PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6" - prefer IPv4 over
1373 IPv6, but support both and try to use both.
1374 "PERL_ANYEVENT_PROTOCOLS=ipv4" - only support IPv4, never try to
1375 resolve or contact IPv6 addresses.
1376 "PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4" support either IPv4 or IPv6, but
1377 prefer IPv6 over IPv4.
1378
1379 "PERL_ANYEVENT_HOSTS"
1380 This variable, if specified, overrides the /etc/hosts file used by
1381 AnyEvent::Socket"::resolve_sockaddr", i.e. hosts aliases will be
1382 read from that file instead.
1383
1384 "PERL_ANYEVENT_EDNS0"
1385 Used by AnyEvent::DNS to decide whether to use the EDNS0 extension
1386 for DNS. This extension is generally useful to reduce DNS traffic,
1387 especially when DNSSEC is involved, but some (broken) firewalls drop
1388 such DNS packets, which is why it is off by default.
1389
1390 Setting this variable to 1 will cause AnyEvent::DNS to announce
1391 EDNS0 in its DNS requests.
1392
1393 "PERL_ANYEVENT_MAX_FORKS"
1394 The maximum number of child processes that
1395 "AnyEvent::Util::fork_call" will create in parallel.
1396
1397 "PERL_ANYEVENT_MAX_OUTSTANDING_DNS"
1398 The default value for the "max_outstanding" parameter for the
1399 default DNS resolver - this is the maximum number of parallel DNS
1400 requests that are sent to the DNS server.
1401
1402 "PERL_ANYEVENT_MAX_SIGNAL_LATENCY"
1403 Perl has inherently racy signal handling (you can basically choose
1404 between losing signals and memory corruption) - pure perl event
1405 loops (including "AnyEvent::Loop", when "Async::Interrupt" isn't
1406 available) therefore have to poll regularly to avoid losing signals.
1407
1408 Some event loops are racy, but don't poll regularly, and some event
1409 loops are written in C but are still racy. For those event loops,
1410 AnyEvent installs a timer that regularly wakes up the event loop.
1411
1412 By default, the interval for this timer is 10 seconds, but you can
1413 override this delay with this environment variable (or by setting
1414 the $AnyEvent::MAX_SIGNAL_LATENCY variable before creating signal
1415 watchers).
1416
1417 Lower values increase CPU (and energy) usage, higher values can
1418 introduce long delays when reaping children or waiting for signals.
1419
1420 The AnyEvent::Async module, if available, will be used to avoid this
1421 polling (with most event loops).
1422
1423 "PERL_ANYEVENT_RESOLV_CONF"
1424 The absolute path to a resolv.conf-style file to use instead of
1425 /etc/resolv.conf (or the OS-specific configuration) in the default
1426 resolver, or the empty string to select the default configuration.
1427
1428 "PERL_ANYEVENT_CA_FILE", "PERL_ANYEVENT_CA_PATH".
1429 When neither "ca_file" nor "ca_path" was specified during
1430 AnyEvent::TLS context creation, and either of these environment
1431 variables are nonempty, they will be used to specify CA certificate
1432 locations instead of a system-dependent default.
1433
1434 "PERL_ANYEVENT_AVOID_GUARD" and "PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT"
1435 When these are set to 1, then the respective modules are not loaded.
1436 Mostly good for testing AnyEvent itself.
495 1437
496SUPPLYING YOUR OWN EVENT MODEL INTERFACE 1438SUPPLYING YOUR OWN EVENT MODEL INTERFACE
497 This is an advanced topic that you do not normally need to use AnyEvent 1439 This is an advanced topic that you do not normally need to use AnyEvent
498 in a module. This section is only of use to event loop authors who want 1440 in a module. This section is only of use to event loop authors who want
499 to provide AnyEvent compatibility. 1441 to provide AnyEvent compatibility.
… …
533 1475
534 *rxvt-unicode* also cheats a bit by not providing blocking access to 1476 *rxvt-unicode* also cheats a bit by not providing blocking access to
535 condition variables: code blocking while waiting for a condition will 1477 condition variables: code blocking while waiting for a condition will
536 "die". This still works with most modules/usages, and blocking calls 1478 "die". This still works with most modules/usages, and blocking calls
537 must not be done in an interactive application, so it makes sense. 1479 must not be done in an interactive application, so it makes sense.
538
539ENVIRONMENT VARIABLES
540 The following environment variables are used by this module:
541
542 "PERL_ANYEVENT_VERBOSE"
543 By default, AnyEvent will be completely silent except in fatal
544 conditions. You can set this environment variable to make AnyEvent
545 more talkative.
546
547 When set to 1 or higher, causes AnyEvent to warn about unexpected
548 conditions, such as not being able to load the event model specified
549 by "PERL_ANYEVENT_MODEL".
550
551 When set to 2 or higher, cause AnyEvent to report to STDERR which
552 event model it chooses.
553
554 "PERL_ANYEVENT_MODEL"
555 This can be used to specify the event model to be used by AnyEvent,
556 before autodetection and -probing kicks in. It must be a string
557 consisting entirely of ASCII letters. The string "AnyEvent::Impl::"
558 gets prepended and the resulting module name is loaded and if the
559 load was successful, used as event model. If it fails to load
560 AnyEvent will proceed with autodetection and -probing.
561
562 This functionality might change in future versions.
563
564 For example, to force the pure perl model (AnyEvent::Impl::Perl) you
565 could start your program like this:
566
567 PERL_ANYEVENT_MODEL=Perl perl ...
568 1480
569EXAMPLE PROGRAM 1481EXAMPLE PROGRAM
570 The following program uses an I/O watcher to read data from STDIN, a 1482 The following program uses an I/O watcher to read data from STDIN, a
571 timer to display a message once per second, and a condition variable to 1483 timer to display a message once per second, and a condition variable to
572 quit the program when the user enters quit: 1484 quit the program when the user enters quit:
… …
580 poll => 'r', 1492 poll => 'r',
581 cb => sub { 1493 cb => sub {
582 warn "io event <$_[0]>\n"; # will always output <r> 1494 warn "io event <$_[0]>\n"; # will always output <r>
583 chomp (my $input = <STDIN>); # read a line 1495 chomp (my $input = <STDIN>); # read a line
584 warn "read: $input\n"; # output what has been read 1496 warn "read: $input\n"; # output what has been read
585 $cv->broadcast if $input =~ /^q/i; # quit program if /^q/i 1497 $cv->send if $input =~ /^q/i; # quit program if /^q/i
586 }, 1498 },
587 ); 1499 );
588 1500
589 my $time_watcher; # can only be used once
590
591 sub new_timer {
592 $timer = AnyEvent->timer (after => 1, cb => sub { 1501 my $time_watcher = AnyEvent->timer (after => 1, interval => 1, cb => sub {
593 warn "timeout\n"; # print 'timeout' about every second 1502 warn "timeout\n"; # print 'timeout' at most every second
594 &new_timer; # and restart the time
595 });
596 } 1503 });
597 1504
598 new_timer; # create first timer
599
600 $cv->wait; # wait until user enters /^q/i 1505 $cv->recv; # wait until user enters /^q/i
601 1506
602REAL-WORLD EXAMPLE 1507REAL-WORLD EXAMPLE
603 Consider the Net::FCP module. It features (among others) the following 1508 Consider the Net::FCP module. It features (among others) the following
604 API calls, which are to freenet what HTTP GET requests are to http: 1509 API calls, which are to freenet what HTTP GET requests are to http:
605 1510
… …
654 syswrite $txn->{fh}, $txn->{request} 1559 syswrite $txn->{fh}, $txn->{request}
655 or die "connection or write error"; 1560 or die "connection or write error";
656 $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r }); 1561 $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
657 1562
658 Again, "fh_ready_r" waits till all data has arrived, and then stores the 1563 Again, "fh_ready_r" waits till all data has arrived, and then stores the
659 result and signals any possible waiters that the request ahs finished: 1564 result and signals any possible waiters that the request has finished:
660 1565
661 sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf}; 1566 sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
662 1567
663 if (end-of-file or data complete) { 1568 if (end-of-file or data complete) {
664 $txn->{result} = $txn->{buf}; 1569 $txn->{result} = $txn->{buf};
665 $txn->{finished}->broadcast; 1570 $txn->{finished}->send;
666 $txb->{cb}->($txn) of $txn->{cb}; # also call callback 1571 $txb->{cb}->($txn) of $txn->{cb}; # also call callback
667 } 1572 }
668 1573
669 The "result" method, finally, just waits for the finished signal (if the 1574 The "result" method, finally, just waits for the finished signal (if the
670 request was already finished, it doesn't wait, of course, and returns 1575 request was already finished, it doesn't wait, of course, and returns
671 the data: 1576 the data:
672 1577
673 $txn->{finished}->wait; 1578 $txn->{finished}->recv;
674 return $txn->{result}; 1579 return $txn->{result};
675 1580
676 The actual code goes further and collects all errors ("die"s, 1581 The actual code goes further and collects all errors ("die"s,
677 exceptions) that occured during request processing. The "result" method 1582 exceptions) that occurred during request processing. The "result" method
678 detects whether an exception as thrown (it is stored inside the $txn 1583 detects whether an exception as thrown (it is stored inside the $txn
679 object) and just throws the exception, which means connection errors and 1584 object) and just throws the exception, which means connection errors and
680 other problems get reported tot he code that tries to use the result, 1585 other problems get reported to the code that tries to use the result,
681 not in a random callback. 1586 not in a random callback.
682 1587
683 All of this enables the following usage styles: 1588 All of this enables the following usage styles:
684 1589
685 1. Blocking: 1590 1. Blocking:
… …
713 1618
714 my $quit = AnyEvent->condvar; 1619 my $quit = AnyEvent->condvar;
715 1620
716 $fcp->txn_client_get ($url)->cb (sub { 1621 $fcp->txn_client_get ($url)->cb (sub {
717 ... 1622 ...
718 $quit->broadcast; 1623 $quit->send;
719 }); 1624 });
720 1625
721 $quit->wait; 1626 $quit->recv;
722 1627
723BENCHMARKS 1628BENCHMARKS
724 To give you an idea of the performance and overheads that AnyEvent adds 1629 To give you an idea of the performance and overheads that AnyEvent adds
725 over the event loops themselves and to give you an impression of the 1630 over the event loops themselves and to give you an impression of the
726 speed of various event loops I prepared some benchmarks. 1631 speed of various event loops I prepared some benchmarks.
727 1632
728 BENCHMARKING ANYEVENT OVERHEAD 1633 BENCHMARKING ANYEVENT OVERHEAD
729 Here is a benchmark of various supported event models used natively and 1634 Here is a benchmark of various supported event models used natively and
730 through anyevent. The benchmark creates a lot of timers (with a zero 1635 through AnyEvent. The benchmark creates a lot of timers (with a zero
731 timeout) and I/O watchers (watching STDOUT, a pty, to become writable, 1636 timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
732 which it is), lets them fire exactly once and destroys them again. 1637 which it is), lets them fire exactly once and destroys them again.
733 1638
734 Source code for this benchmark is found as eg/bench in the AnyEvent 1639 Source code for this benchmark is found as eg/bench in the AnyEvent
735 distribution. 1640 distribution. It uses the AE interface, which makes a real difference
1641 for the EV and Perl backends only.
736 1642
737 Explanation of the columns 1643 Explanation of the columns
738 *watcher* is the number of event watchers created/destroyed. Since 1644 *watcher* is the number of event watchers created/destroyed. Since
739 different event models feature vastly different performances, each event 1645 different event models feature vastly different performances, each event
740 loop was given a number of watchers so that overall runtime is 1646 loop was given a number of watchers so that overall runtime is
… …
751 between all watchers, to avoid adding memory overhead. That means 1657 between all watchers, to avoid adding memory overhead. That means
752 closure creation and memory usage is not included in the figures. 1658 closure creation and memory usage is not included in the figures.
753 1659
754 *invoke* is the time, in microseconds, used to invoke a simple callback. 1660 *invoke* is the time, in microseconds, used to invoke a simple callback.
755 The callback simply counts down a Perl variable and after it was invoked 1661 The callback simply counts down a Perl variable and after it was invoked
756 "watcher" times, it would "->broadcast" a condvar once to signal the end 1662 "watcher" times, it would "->send" a condvar once to signal the end of
757 of this phase. 1663 this phase.
758 1664
759 *destroy* is the time, in microseconds, that it takes to destroy a 1665 *destroy* is the time, in microseconds, that it takes to destroy a
760 single watcher. 1666 single watcher.
761 1667
762 Results 1668 Results
763 name watchers bytes create invoke destroy comment 1669 name watchers bytes create invoke destroy comment
764 EV/EV 400000 244 0.56 0.46 0.31 EV native interface 1670 EV/EV 100000 223 0.47 0.43 0.27 EV native interface
765 EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers 1671 EV/Any 100000 223 0.48 0.42 0.26 EV + AnyEvent watchers
766 CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal 1672 Coro::EV/Any 100000 223 0.47 0.42 0.26 coroutines + Coro::Signal
767 Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation 1673 Perl/Any 100000 431 2.70 0.74 0.92 pure perl implementation
768 Event/Event 16000 516 31.88 31.30 0.85 Event native interface 1674 Event/Event 16000 516 31.16 31.84 0.82 Event native interface
769 Event/Any 16000 590 35.75 31.42 1.08 Event + AnyEvent watchers 1675 Event/Any 16000 1203 42.61 34.79 1.80 Event + AnyEvent watchers
1676 IOAsync/Any 16000 1911 41.92 27.45 16.81 via IO::Async::Loop::IO_Poll
1677 IOAsync/Any 16000 1726 40.69 26.37 15.25 via IO::Async::Loop::Epoll
770 Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour 1678 Glib/Any 16000 1118 89.00 12.57 51.17 quadratic behaviour
771 Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers 1679 Tk/Any 2000 1346 20.96 10.75 8.00 SEGV with >> 2000 watchers
772 POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event 1680 POE/Any 2000 6951 108.97 795.32 14.24 via POE::Loop::Event
773 POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select 1681 POE/Any 2000 6648 94.79 774.40 575.51 via POE::Loop::Select
774 1682
775 Discussion 1683 Discussion
776 The benchmark does *not* measure scalability of the event loop very 1684 The benchmark does *not* measure scalability of the event loop very
777 well. For example, a select-based event loop (such as the pure perl one) 1685 well. For example, a select-based event loop (such as the pure perl one)
778 can never compete with an event loop that uses epoll when the number of 1686 can never compete with an event loop that uses epoll when the number of
… …
789 benchmark machine, handling an event takes roughly 1600 CPU cycles with 1697 benchmark machine, handling an event takes roughly 1600 CPU cycles with
790 EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 1698 EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000
791 CPU cycles with POE. 1699 CPU cycles with POE.
792 1700
793 "EV" is the sole leader regarding speed and memory use, which are both 1701 "EV" is the sole leader regarding speed and memory use, which are both
794 maximal/minimal, respectively. Even when going through AnyEvent, it uses 1702 maximal/minimal, respectively. When using the AE API there is zero
1703 overhead (when going through the AnyEvent API create is about 5-6 times
1704 slower, with other times being equal, so still uses far less memory than
795 far less memory than any other event loop and is still faster than Event 1705 any other event loop and is still faster than Event natively).
796 natively.
797 1706
798 The pure perl implementation is hit in a few sweet spots (both the 1707 The pure perl implementation is hit in a few sweet spots (both the
799 constant timeout and the use of a single fd hit optimisations in the 1708 constant timeout and the use of a single fd hit optimisations in the
800 perl interpreter and the backend itself). Nevertheless this shows that 1709 perl interpreter and the backend itself). Nevertheless this shows that
801 it adds very little overhead in itself. Like any select-based backend 1710 it adds very little overhead in itself. Like any select-based backend
… …
804 benchmark. 1713 benchmark.
805 1714
806 The "Event" module has a relatively high setup and callback invocation 1715 The "Event" module has a relatively high setup and callback invocation
807 cost, but overall scores in on the third place. 1716 cost, but overall scores in on the third place.
808 1717
1718 "IO::Async" performs admirably well, about on par with "Event", even
1719 when using its pure perl backend.
1720
809 "Glib"'s memory usage is quite a bit higher, but it features a faster 1721 "Glib"'s memory usage is quite a bit higher, but it features a faster
810 callback invocation and overall ends up in the same class as "Event". 1722 callback invocation and overall ends up in the same class as "Event".
811 However, Glib scales extremely badly, doubling the number of watchers 1723 However, Glib scales extremely badly, doubling the number of watchers
812 increases the processing time by more than a factor of four, making it 1724 increases the processing time by more than a factor of four, making it
813 completely unusable when using larger numbers of watchers (note that 1725 completely unusable when using larger numbers of watchers (note that
… …
823 the figures above). 1735 the figures above).
824 1736
825 "POE", regardless of underlying event loop (whether using its pure perl 1737 "POE", regardless of underlying event loop (whether using its pure perl
826 select-based backend or the Event module, the POE-EV backend couldn't be 1738 select-based backend or the Event module, the POE-EV backend couldn't be
827 tested because it wasn't working) shows abysmal performance and memory 1739 tested because it wasn't working) shows abysmal performance and memory
828 usage: Watchers use almost 30 times as much memory as EV watchers, and 1740 usage with AnyEvent: Watchers use almost 30 times as much memory as EV
829 10 times as much memory as Event (the high memory requirements are 1741 watchers, and 10 times as much memory as Event (the high memory
830 caused by requiring a session for each watcher). Watcher invocation 1742 requirements are caused by requiring a session for each watcher).
831 speed is almost 900 times slower than with AnyEvent's pure perl 1743 Watcher invocation speed is almost 900 times slower than with AnyEvent's
1744 pure perl implementation.
1745
832 implementation. The design of the POE adaptor class in AnyEvent can not 1746 The design of the POE adaptor class in AnyEvent can not really account
833 really account for this, as session creation overhead is small compared 1747 for the performance issues, though, as session creation overhead is
834 to execution of the state machine, which is coded pretty optimally 1748 small compared to execution of the state machine, which is coded pretty
835 within AnyEvent::Impl::POE. POE simply seems to be abysmally slow. 1749 optimally within AnyEvent::Impl::POE (and while everybody agrees that
1750 using multiple sessions is not a good approach, especially regarding
1751 memory usage, even the author of POE could not come up with a faster
1752 design).
836 1753
837 Summary 1754 Summary
838 * Using EV through AnyEvent is faster than any other event loop (even 1755 * Using EV through AnyEvent is faster than any other event loop (even
839 when used without AnyEvent), but most event loops have acceptable 1756 when used without AnyEvent), but most event loops have acceptable
840 performance with or without AnyEvent. 1757 performance with or without AnyEvent.
841 1758
842 * The overhead AnyEvent adds is usually much smaller than the overhead 1759 * The overhead AnyEvent adds is usually much smaller than the overhead
843 of the actual event loop, only with extremely fast event loops such 1760 of the actual event loop, only with extremely fast event loops such
844 as EV adds AnyEvent significant overhead. 1761 as EV does AnyEvent add significant overhead.
845 1762
846 * You should avoid POE like the plague if you want performance or 1763 * You should avoid POE like the plague if you want performance or
847 reasonable memory usage. 1764 reasonable memory usage.
848 1765
849 BENCHMARKING THE LARGE SERVER CASE 1766 BENCHMARKING THE LARGE SERVER CASE
850 This benchmark atcually benchmarks the event loop itself. It works by 1767 This benchmark actually benchmarks the event loop itself. It works by
851 creating a number of "servers": each server consists of a socketpair, a 1768 creating a number of "servers": each server consists of a socket pair, a
852 timeout watcher that gets reset on activity (but never fires), and an 1769 timeout watcher that gets reset on activity (but never fires), and an
853 I/O watcher waiting for input on one side of the socket. Each time the 1770 I/O watcher waiting for input on one side of the socket. Each time the
854 socket watcher reads a byte it will write that byte to a random other 1771 socket watcher reads a byte it will write that byte to a random other
855 "server". 1772 "server".
856 1773
857 The effect is that there will be a lot of I/O watchers, only part of 1774 The effect is that there will be a lot of I/O watchers, only part of
858 which are active at any one point (so there is a constant number of 1775 which are active at any one point (so there is a constant number of
859 active fds for each loop iterstaion, but which fds these are is random). 1776 active fds for each loop iteration, but which fds these are is random).
860 The timeout is reset each time something is read because that reflects 1777 The timeout is reset each time something is read because that reflects
861 how most timeouts work (and puts extra pressure on the event loops). 1778 how most timeouts work (and puts extra pressure on the event loops).
862 1779
863 In this benchmark, we use 10000 socketpairs (20000 sockets), of which 1780 In this benchmark, we use 10000 socket pairs (20000 sockets), of which
864 100 (1%) are active. This mirrors the activity of large servers with 1781 100 (1%) are active. This mirrors the activity of large servers with
865 many connections, most of which are idle at any one point in time. 1782 many connections, most of which are idle at any one point in time.
866 1783
867 Source code for this benchmark is found as eg/bench2 in the AnyEvent 1784 Source code for this benchmark is found as eg/bench2 in the AnyEvent
868 distribution. 1785 distribution. It uses the AE interface, which makes a real difference
1786 for the EV and Perl backends only.
869 1787
870 Explanation of the columns 1788 Explanation of the columns
871 *sockets* is the number of sockets, and twice the number of "servers" 1789 *sockets* is the number of sockets, and twice the number of "servers"
872 (as each server has a read and write socket end). 1790 (as each server has a read and write socket end).
873 1791
874 *create* is the time it takes to create a socketpair (which is 1792 *create* is the time it takes to create a socket pair (which is
875 nontrivial) and two watchers: an I/O watcher and a timeout watcher. 1793 nontrivial) and two watchers: an I/O watcher and a timeout watcher.
876 1794
877 *request*, the most important value, is the time it takes to handle a 1795 *request*, the most important value, is the time it takes to handle a
878 single "request", that is, reading the token from the pipe and 1796 single "request", that is, reading the token from the pipe and
879 forwarding it to another server. This includes deleting the old timeout 1797 forwarding it to another server. This includes deleting the old timeout
880 and creating a new one that moves the timeout into the future. 1798 and creating a new one that moves the timeout into the future.
881 1799
882 Results 1800 Results
883 name sockets create request 1801 name sockets create request
884 EV 20000 69.01 11.16 1802 EV 20000 62.66 7.99
885 Perl 20000 73.32 35.87 1803 Perl 20000 68.32 32.64
886 Event 20000 212.62 257.32 1804 IOAsync 20000 174.06 101.15 epoll
887 Glib 20000 651.16 1896.30 1805 IOAsync 20000 174.67 610.84 poll
1806 Event 20000 202.69 242.91
1807 Glib 20000 557.01 1689.52
888 POE 20000 349.67 12317.24 uses POE::Loop::Event 1808 POE 20000 341.54 12086.32 uses POE::Loop::Event
889 1809
890 Discussion 1810 Discussion
891 This benchmark *does* measure scalability and overall performance of the 1811 This benchmark *does* measure scalability and overall performance of the
892 particular event loop. 1812 particular event loop.
893 1813
894 EV is again fastest. Since it is using epoll on my system, the setup 1814 EV is again fastest. Since it is using epoll on my system, the setup
895 time is relatively high, though. 1815 time is relatively high, though.
896 1816
897 Perl surprisingly comes second. It is much faster than the C-based event 1817 Perl surprisingly comes second. It is much faster than the C-based event
898 loops Event and Glib. 1818 loops Event and Glib.
1819
1820 IO::Async performs very well when using its epoll backend, and still
1821 quite good compared to Glib when using its pure perl backend.
899 1822
900 Event suffers from high setup time as well (look at its code and you 1823 Event suffers from high setup time as well (look at its code and you
901 will understand why). Callback invocation also has a high overhead 1824 will understand why). Callback invocation also has a high overhead
902 compared to the "$_->() for .."-style loop that the Perl event loop 1825 compared to the "$_->() for .."-style loop that the Perl event loop
903 uses. Event uses select or poll in basically all documented 1826 uses. Event uses select or poll in basically all documented
… …
909 POE is still completely out of the picture, taking over 1000 times as 1832 POE is still completely out of the picture, taking over 1000 times as
910 long as EV, and over 100 times as long as the Perl implementation, even 1833 long as EV, and over 100 times as long as the Perl implementation, even
911 though it uses a C-based event loop in this case. 1834 though it uses a C-based event loop in this case.
912 1835
913 Summary 1836 Summary
914 * The pure perl implementation performs extremely well, considering 1837 * The pure perl implementation performs extremely well.
915 that it uses select.
916 1838
917 * Avoid Glib or POE in large projects where performance matters. 1839 * Avoid Glib or POE in large projects where performance matters.
918 1840
919 BENCHMARKING SMALL SERVERS 1841 BENCHMARKING SMALL SERVERS
920 While event loops should scale (and select-based ones do not...) even to 1842 While event loops should scale (and select-based ones do not...) even to
… …
944 and speed most when you have lots of watchers, not when you only have a 1866 and speed most when you have lots of watchers, not when you only have a
945 few of them). 1867 few of them).
946 1868
947 EV is again fastest. 1869 EV is again fastest.
948 1870
949 Perl again comes second. It is noticably faster than the C-based event 1871 Perl again comes second. It is noticeably faster than the C-based event
950 loops Event and Glib, although the difference is too small to really 1872 loops Event and Glib, although the difference is too small to really
951 matter. 1873 matter.
952 1874
953 POE also performs much better in this case, but is is still far behind 1875 POE also performs much better in this case, but is is still far behind
954 the others. 1876 the others.
955 1877
956 Summary 1878 Summary
957 * C-based event loops perform very well with small number of watchers, 1879 * C-based event loops perform very well with small number of watchers,
958 as the management overhead dominates. 1880 as the management overhead dominates.
959 1881
1882 THE IO::Lambda BENCHMARK
1883 Recently I was told about the benchmark in the IO::Lambda manpage, which
1884 could be misinterpreted to make AnyEvent look bad. In fact, the
1885 benchmark simply compares IO::Lambda with POE, and IO::Lambda looks
1886 better (which shouldn't come as a surprise to anybody). As such, the
1887 benchmark is fine, and mostly shows that the AnyEvent backend from
1888 IO::Lambda isn't very optimal. But how would AnyEvent compare when used
1889 without the extra baggage? To explore this, I wrote the equivalent
1890 benchmark for AnyEvent.
1891
1892 The benchmark itself creates an echo-server, and then, for 500 times,
1893 connects to the echo server, sends a line, waits for the reply, and then
1894 creates the next connection. This is a rather bad benchmark, as it
1895 doesn't test the efficiency of the framework or much non-blocking I/O,
1896 but it is a benchmark nevertheless.
1897
1898 name runtime
1899 Lambda/select 0.330 sec
1900 + optimized 0.122 sec
1901 Lambda/AnyEvent 0.327 sec
1902 + optimized 0.138 sec
1903 Raw sockets/select 0.077 sec
1904 POE/select, components 0.662 sec
1905 POE/select, raw sockets 0.226 sec
1906 POE/select, optimized 0.404 sec
1907
1908 AnyEvent/select/nb 0.085 sec
1909 AnyEvent/EV/nb 0.068 sec
1910 +state machine 0.134 sec
1911
1912 The benchmark is also a bit unfair (my fault): the IO::Lambda/POE
1913 benchmarks actually make blocking connects and use 100% blocking I/O,
1914 defeating the purpose of an event-based solution. All of the newly
1915 written AnyEvent benchmarks use 100% non-blocking connects (using
1916 AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS
1917 resolver), so AnyEvent is at a disadvantage here, as non-blocking
1918 connects generally require a lot more bookkeeping and event handling
1919 than blocking connects (which involve a single syscall only).
1920
1921 The last AnyEvent benchmark additionally uses AnyEvent::Handle, which
1922 offers similar expressive power as POE and IO::Lambda, using
1923 conventional Perl syntax. This means that both the echo server and the
1924 client are 100% non-blocking, further placing it at a disadvantage.
1925
1926 As you can see, the AnyEvent + EV combination even beats the
1927 hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl
1928 backend easily beats IO::Lambda and POE.
1929
1930 And even the 100% non-blocking version written using the high-level (and
1931 slow :) AnyEvent::Handle abstraction beats both POE and IO::Lambda
1932 higher level ("unoptimised") abstractions by a large margin, even though
1933 it does all of DNS, tcp-connect and socket I/O in a non-blocking way.
1934
1935 The two AnyEvent benchmarks programs can be found as eg/ae0.pl and
1936 eg/ae2.pl in the AnyEvent distribution, the remaining benchmarks are
1937 part of the IO::Lambda distribution and were used without any changes.
1938
1939SIGNALS
1940 AnyEvent currently installs handlers for these signals:
1941
1942 SIGCHLD
1943 A handler for "SIGCHLD" is installed by AnyEvent's child watcher
1944 emulation for event loops that do not support them natively. Also,
1945 some event loops install a similar handler.
1946
1947 Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE,
1948 then AnyEvent will reset it to default, to avoid losing child exit
1949 statuses.
1950
1951 SIGPIPE
1952 A no-op handler is installed for "SIGPIPE" when $SIG{PIPE} is
1953 "undef" when AnyEvent gets loaded.
1954
1955 The rationale for this is that AnyEvent users usually do not really
1956 depend on SIGPIPE delivery (which is purely an optimisation for
1957 shell use, or badly-written programs), but "SIGPIPE" can cause
1958 spurious and rare program exits as a lot of people do not expect
1959 "SIGPIPE" when writing to some random socket.
1960
1961 The rationale for installing a no-op handler as opposed to ignoring
1962 it is that this way, the handler will be restored to defaults on
1963 exec.
1964
1965 Feel free to install your own handler, or reset it to defaults.
1966
1967RECOMMENDED/OPTIONAL MODULES
1968 One of AnyEvent's main goals is to be 100% Pure-Perl(tm): only perl (and
1969 its built-in modules) are required to use it.
1970
1971 That does not mean that AnyEvent won't take advantage of some additional
1972 modules if they are installed.
1973
1974 This section explains which additional modules will be used, and how
1975 they affect AnyEvent's operation.
1976
1977 Async::Interrupt
1978 This slightly arcane module is used to implement fast signal
1979 handling: To my knowledge, there is no way to do completely
1980 race-free and quick signal handling in pure perl. To ensure that
1981 signals still get delivered, AnyEvent will start an interval timer
1982 to wake up perl (and catch the signals) with some delay (default is
1983 10 seconds, look for $AnyEvent::MAX_SIGNAL_LATENCY).
1984
1985 If this module is available, then it will be used to implement
1986 signal catching, which means that signals will not be delayed, and
1987 the event loop will not be interrupted regularly, which is more
1988 efficient (and good for battery life on laptops).
1989
1990 This affects not just the pure-perl event loop, but also other event
1991 loops that have no signal handling on their own (e.g. Glib, Tk, Qt).
1992
1993 Some event loops (POE, Event, Event::Lib) offer signal watchers
1994 natively, and either employ their own workarounds (POE) or use
1995 AnyEvent's workaround (using $AnyEvent::MAX_SIGNAL_LATENCY).
1996 Installing Async::Interrupt does nothing for those backends.
1997
1998 EV This module isn't really "optional", as it is simply one of the
1999 backend event loops that AnyEvent can use. However, it is simply the
2000 best event loop available in terms of features, speed and stability:
2001 It supports the AnyEvent API optimally, implements all the watcher
2002 types in XS, does automatic timer adjustments even when no monotonic
2003 clock is available, can take avdantage of advanced kernel interfaces
2004 such as "epoll" and "kqueue", and is the fastest backend *by far*.
2005 You can even embed Glib/Gtk2 in it (or vice versa, see EV::Glib and
2006 Glib::EV).
2007
2008 If you only use backends that rely on another event loop (e.g.
2009 "Tk"), then this module will do nothing for you.
2010
2011 Guard
2012 The guard module, when used, will be used to implement
2013 "AnyEvent::Util::guard". This speeds up guards considerably (and
2014 uses a lot less memory), but otherwise doesn't affect guard
2015 operation much. It is purely used for performance.
2016
2017 JSON and JSON::XS
2018 One of these modules is required when you want to read or write JSON
2019 data via AnyEvent::Handle. JSON is also written in pure-perl, but
2020 can take advantage of the ultra-high-speed JSON::XS module when it
2021 is installed.
2022
2023 Net::SSLeay
2024 Implementing TLS/SSL in Perl is certainly interesting, but not very
2025 worthwhile: If this module is installed, then AnyEvent::Handle (with
2026 the help of AnyEvent::TLS), gains the ability to do TLS/SSL.
2027
2028 Time::HiRes
2029 This module is part of perl since release 5.008. It will be used
2030 when the chosen event library does not come with a timing source of
2031 its own. The pure-perl event loop (AnyEvent::Loop) will additionally
2032 load it to try to use a monotonic clock for timing stability.
2033
2034 AnyEvent::AIO (and IO::AIO)
2035 The default implementation of AnyEvent::IO is to do I/O
2036 synchronously, stopping programs while they access the disk, which
2037 is fine for a lot of programs.
2038
2039 Installing AnyEvent::AIO (and its IO::AIO dependency) makes it
2040 switch to a true asynchronous implementation, so event processing
2041 can continue even while waiting for disk I/O.
2042
960FORK 2043FORK
961 Most event libraries are not fork-safe. The ones who are usually are 2044 Most event libraries are not fork-safe. The ones who are usually are
962 because they are so inefficient. Only EV is fully fork-aware. 2045 because they rely on inefficient but fork-safe "select" or "poll" calls
2046 - higher performance APIs such as BSD's kqueue or the dreaded Linux
2047 epoll are usually badly thought-out hacks that are incompatible with
2048 fork in one way or another. Only EV is fully fork-aware and ensures that
2049 you continue event-processing in both parent and child (or both, if you
2050 know what you are doing).
2051
2052 This means that, in general, you cannot fork and do event processing in
2053 the child if the event library was initialised before the fork (which
2054 usually happens when the first AnyEvent watcher is created, or the
2055 library is loaded).
963 2056
964 If you have to fork, you must either do so *before* creating your first 2057 If you have to fork, you must either do so *before* creating your first
965 watcher OR you must not use AnyEvent at all in the child. 2058 watcher OR you must not use AnyEvent at all in the child OR you must do
2059 something completely out of the scope of AnyEvent (see below).
2060
2061 The problem of doing event processing in the parent *and* the child is
2062 much more complicated: even for backends that *are* fork-aware or
2063 fork-safe, their behaviour is not usually what you want: fork clones all
2064 watchers, that means all timers, I/O watchers etc. are active in both
2065 parent and child, which is almost never what you want. Using "exec" to
2066 start worker children from some kind of manage prrocess is usually
2067 preferred, because it is much easier and cleaner, at the expense of
2068 having to have another binary.
2069
2070 In addition to logical problems with fork, there are also implementation
2071 problems. For example, on POSIX systems, you cannot fork at all in Perl
2072 code if a thread (I am talking of pthreads here) was ever created in the
2073 process, and this is just the tip of the iceberg. In general, using fork
2074 from Perl is difficult, and attempting to use fork without an exec to
2075 implement some kind of parallel processing is almost certainly doomed.
2076
2077 To safely fork and exec, you should use a module such as Proc::FastSpawn
2078 that let's you safely fork and exec new processes.
2079
2080 If you want to do multiprocessing using processes, you can look at the
2081 AnyEvent::Fork module (and some related modules such as
2082 AnyEvent::Fork::RPC, AnyEvent::Fork::Pool and AnyEvent::Fork::Remote).
2083 This module allows you to safely create subprocesses without any
2084 limitations - you can use X11 toolkits or AnyEvent in the children
2085 created by AnyEvent::Fork safely and without any special precautions.
966 2086
967SECURITY CONSIDERATIONS 2087SECURITY CONSIDERATIONS
968 AnyEvent can be forced to load any event model via 2088 AnyEvent can be forced to load any event model via
969 $ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used 2089 $ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used
970 to execute arbitrary code or directly gain access, it can easily be used 2090 to execute arbitrary code or directly gain access, it can easily be used
… …
973 model than specified in the variable. 2093 model than specified in the variable.
974 2094
975 You can make AnyEvent completely ignore this variable by deleting it 2095 You can make AnyEvent completely ignore this variable by deleting it
976 before the first watcher gets created, e.g. with a "BEGIN" block: 2096 before the first watcher gets created, e.g. with a "BEGIN" block:
977 2097
978 BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} } 2098 BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
979 2099
980 use AnyEvent; 2100 use AnyEvent;
2101
2102 Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
2103 be used to probe what backend is used and gain other information (which
2104 is probably even less useful to an attacker than PERL_ANYEVENT_MODEL),
2105 and $ENV{PERL_ANYEVENT_STRICT}.
2106
2107 Note that AnyEvent will remove *all* environment variables starting with
2108 "PERL_ANYEVENT_" from %ENV when it is loaded while taint mode is
2109 enabled.
2110
2111BUGS
2112 Perl 5.8 has numerous memleaks that sometimes hit this module and are
2113 hard to work around. If you suffer from memleaks, first upgrade to Perl
2114 5.10 and check wether the leaks still show up. (Perl 5.10.0 has other
2115 annoying memleaks, such as leaking on "map" and "grep" but it is usually
2116 not as pronounced).
981 2117
982SEE ALSO 2118SEE ALSO
983 Event modules: Coro::EV, EV, EV::Glib, Glib::EV, Coro::Event, Event, 2119 Tutorial/Introduction: AnyEvent::Intro.
984 Glib::Event, Glib, Coro, Tk, Event::Lib, Qt, POE.
985 2120
986 Implementations: AnyEvent::Impl::CoroEV, AnyEvent::Impl::EV, 2121 FAQ: AnyEvent::FAQ.
987 AnyEvent::Impl::CoroEvent, AnyEvent::Impl::Event, AnyEvent::Impl::Glib, 2122
2123 Utility functions: AnyEvent::Util (misc. grab-bag), AnyEvent::Log
2124 (simply logging).
2125
2126 Development/Debugging: AnyEvent::Strict (stricter checking),
2127 AnyEvent::Debug (interactive shell, watcher tracing).
2128
2129 Supported event modules: AnyEvent::Loop, EV, EV::Glib, Glib::EV, Event,
2130 Glib::Event, Glib, Tk, Event::Lib, Qt, POE, FLTK, Cocoa::EventLoop, UV.
2131
2132 Implementations: AnyEvent::Impl::EV, AnyEvent::Impl::Event,
2133 AnyEvent::Impl::Glib, AnyEvent::Impl::Tk, AnyEvent::Impl::Perl,
2134 AnyEvent::Impl::EventLib, AnyEvent::Impl::Qt, AnyEvent::Impl::POE,
988 AnyEvent::Impl::Tk, AnyEvent::Impl::Perl, AnyEvent::Impl::EventLib, 2135 AnyEvent::Impl::IOAsync, AnyEvent::Impl::Irssi, AnyEvent::Impl::FLTK,
989 AnyEvent::Impl::Qt, AnyEvent::Impl::POE. 2136 AnyEvent::Impl::Cocoa, AnyEvent::Impl::UV.
990 2137
991 Nontrivial usage examples: Net::FCP, Net::XMPP2. 2138 Non-blocking handles, pipes, stream sockets, TCP clients and servers:
2139 AnyEvent::Handle, AnyEvent::Socket, AnyEvent::TLS.
2140
2141 Asynchronous File I/O: AnyEvent::IO.
2142
2143 Asynchronous DNS: AnyEvent::DNS.
2144
2145 Thread support: Coro, Coro::AnyEvent, Coro::EV, Coro::Event.
2146
2147 Nontrivial usage examples: AnyEvent::GPSD, AnyEvent::IRC,
2148 AnyEvent::HTTP.
992 2149
993AUTHOR 2150AUTHOR
994 Marc Lehmann <schmorp@schmorp.de> 2151 Marc Lehmann <schmorp@schmorp.de>
995 http://home.schmorp.de/ 2152 http://anyevent.schmorp.de
996 2153

Diff Legend

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