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

Comparing AnyEvent/README (file contents):
Revision 1.17 by root, Tue Apr 22 05:12:19 2008 UTC vs.
Revision 1.58 by root, Sun Dec 20 22:49:52 2009 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 - various supported 4 EV, Event, Glib, Tk, Perl, Event::Lib, Irssi, rxvt-unicode, IO::Async,
5 event loops 5 Qt and POE are various supported event loops/environments.
6 6
7SYNOPSIS 7SYNOPSIS
8 use AnyEvent; 8 use AnyEvent;
9 9
10 # file descriptor readable
10 my $w = AnyEvent->io (fh => $fh, poll => "r|w", cb => sub { 11 my $w = AnyEvent->io (fh => $fh, poll => "r", cb => sub { ... });
12
13 # one-shot or repeating timers
14 my $w = AnyEvent->timer (after => $seconds, cb => sub { ... });
15 my $w = AnyEvent->timer (after => $seconds, interval => $seconds, cb => ...
16
17 print AnyEvent->now; # prints current event loop time
18 print AnyEvent->time; # think Time::HiRes::time or simply CORE::time.
19
20 # POSIX signal
21 my $w = AnyEvent->signal (signal => "TERM", cb => sub { ... });
22
23 # child process exit
24 my $w = AnyEvent->child (pid => $pid, cb => sub {
25 my ($pid, $status) = @_;
11 ... 26 ...
12 }); 27 });
13 28
14 my $w = AnyEvent->timer (after => $seconds, cb => sub { 29 # called when event loop idle (if applicable)
15 ... 30 my $w = AnyEvent->idle (cb => sub { ... });
16 });
17 31
18 my $w = AnyEvent->condvar; # stores whether a condition was flagged 32 my $w = AnyEvent->condvar; # stores whether a condition was flagged
33 $w->send; # wake up current and all future recv's
19 $w->wait; # enters "main loop" till $condvar gets ->broadcast 34 $w->recv; # enters "main loop" till $condvar gets ->send
20 $w->broadcast; # wake up current and all future wait's 35 # use a condvar in callback mode:
36 $w->cb (sub { $_[0]->recv });
37
38INTRODUCTION/TUTORIAL
39 This manpage is mainly a reference manual. If you are interested in a
40 tutorial or some gentle introduction, have a look at the AnyEvent::Intro
41 manpage.
42
43SUPPORT
44 There is a mailinglist for discussing all things AnyEvent, and an IRC
45 channel, too.
46
47 See the AnyEvent project page at the Schmorpforge Ta-Sa Software
48 Repository, at <http://anyevent.schmorp.de>, for more info.
21 49
22WHY YOU SHOULD USE THIS MODULE (OR NOT) 50WHY YOU SHOULD USE THIS MODULE (OR NOT)
23 Glib, POE, IO::Async, Event... CPAN offers event models by the dozen 51 Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
24 nowadays. So what is different about AnyEvent? 52 nowadays. So what is different about AnyEvent?
25 53
26 Executive Summary: AnyEvent is *compatible*, AnyEvent is *free of 54 Executive Summary: AnyEvent is *compatible*, AnyEvent is *free of
27 policy* and AnyEvent is *small and efficient*. 55 policy* and AnyEvent is *small and efficient*.
28 56
29 First and foremost, *AnyEvent is not an event model* itself, it only 57 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 58 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, 59 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, 60 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. 61 only one event loop can be active at the same time in a process.
34 AnyEvent helps hiding the differences between those event loops. 62 AnyEvent cannot change this, but it can hide the differences between
63 those event loops.
35 64
36 The goal of AnyEvent is to offer module authors the ability to do event 65 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 66 programming (waiting for I/O or timer events) without subscribing to a
38 religion, a way of living, and most importantly: without forcing your 67 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 68 module users into the same thing by forcing them to use the same event
40 model you use. 69 model you use.
41 70
42 For modules like POE or IO::Async (which is a total misnomer as it is 71 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 72 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 73 like joining a cult: After you joined, you are dependent on them and you
45 cannot use anything else, as it is simply incompatible to everything 74 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 75 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. 76 are *also* forced to use the same event loop you use.
48 77
49 AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works 78 AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
50 fine. AnyEvent + Tk works fine etc. etc. but none of these work together 79 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 80 with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if your
52 module uses one of those, every user of your module has to use it, too. 81 module uses one of those, every user of your module has to use it, too.
53 But if your module uses AnyEvent, it works transparently with all event 82 But if your module uses AnyEvent, it works transparently with all event
54 models it supports (including stuff like POE and IO::Async, as long as 83 models it supports (including stuff like IO::Async, as long as those use
55 those use one of the supported event loops. It is trivial to add new 84 one of the supported event loops. It is trivial to add new event loops
56 event loops to AnyEvent, too, so it is future-proof). 85 to AnyEvent, too, so it is future-proof).
57 86
58 In addition to being free of having to use *the one and only true event 87 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 88 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 89 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 90 follow. AnyEvent, on the other hand, is lean and up to the point, by
62 only offering the functionality that is necessary, in as thin as a 91 only offering the functionality that is necessary, in as thin as a
63 wrapper as technically possible. 92 wrapper as technically possible.
64 93
94 Of course, AnyEvent comes with a big (and fully optional!) toolbox of
95 useful functionality, such as an asynchronous DNS resolver, 100%
96 non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
97 such as Windows) and lots of real-world knowledge and workarounds for
98 platform bugs and differences.
99
65 Of course, if you want lots of policy (this can arguably be somewhat 100 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 101 useful) and you want to force your users to use the one and only event
67 model, you should *not* use this module. 102 model, you should *not* use this module.
68 103
69DESCRIPTION 104DESCRIPTION
70 AnyEvent provides an identical interface to multiple event loops. This 105 AnyEvent provides an identical interface to multiple event loops. This
… …
75 The interface itself is vaguely similar, but not identical to the Event 110 The interface itself is vaguely similar, but not identical to the Event
76 module. 111 module.
77 112
78 During the first call of any watcher-creation method, the module tries 113 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 114 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, 115 following modules is already loaded: EV, Event, Glib,
81 Glib, Tk. The first one found is used. If none are found, the module 116 AnyEvent::Impl::Perl, Tk, Event::Lib, Qt, POE. The first one found is
82 tries to load these modules in the stated order. The first one that can 117 used. If none are found, the module tries to load these modules
118 (excluding Tk, Event::Lib, Qt and POE as the pure perl adaptor should
119 always succeed) in the order given. The first one that can be
83 be successfully loaded will be used. If, after this, still none could be 120 successfully loaded will be used. If, after this, still none could be
84 found, AnyEvent will fall back to a pure-perl event loop, which is not 121 found, AnyEvent will fall back to a pure-perl event loop, which is not
85 very efficient, but should work everywhere. 122 very efficient, but should work everywhere.
86 123
87 Because AnyEvent first checks for modules that are already loaded, 124 Because AnyEvent first checks for modules that are already loaded,
88 loading an event model explicitly before first using AnyEvent will 125 loading an event model explicitly before first using AnyEvent will
… …
97 starts using it, all bets are off. Maybe you should tell their authors 134 starts using it, all bets are off. Maybe you should tell their authors
98 to use AnyEvent so their modules work together with others seamlessly... 135 to use AnyEvent so their modules work together with others seamlessly...
99 136
100 The pure-perl implementation of AnyEvent is called 137 The pure-perl implementation of AnyEvent is called
101 "AnyEvent::Impl::Perl". Like other event modules you can load it 138 "AnyEvent::Impl::Perl". Like other event modules you can load it
102 explicitly. 139 explicitly and enjoy the high availability of that event loop :)
103 140
104WATCHERS 141WATCHERS
105 AnyEvent has the central concept of a *watcher*, which is an object that 142 AnyEvent has the central concept of a *watcher*, which is an object that
106 stores relevant data for each kind of event you are waiting for, such as 143 stores relevant data for each kind of event you are waiting for, such as
107 the callback to call, the filehandle to watch, etc. 144 the callback to call, the file handle to watch, etc.
108 145
109 These watchers are normal Perl objects with normal Perl lifetime. After 146 These watchers are normal Perl objects with normal Perl lifetime. After
110 creating a watcher it will immediately "watch" for events and invoke the 147 creating a watcher it will immediately "watch" for events and invoke the
111 callback when the event occurs (of course, only when the event model is 148 callback when the event occurs (of course, only when the event model is
112 in control). 149 in control).
113 150
151 Note that callbacks must not permanently change global variables
152 potentially in use by the event loop (such as $_ or $[) and that
153 callbacks must not "die". The former is good programming practise in
154 Perl and the latter stems from the fact that exception handling differs
155 widely between event loops.
156
114 To disable the watcher you have to destroy it (e.g. by setting the 157 To disable the watcher you have to destroy it (e.g. by setting the
115 variable you store it in to "undef" or otherwise deleting all references 158 variable you store it in to "undef" or otherwise deleting all references
116 to it). 159 to it).
117 160
118 All watchers are created by calling a method on the "AnyEvent" class. 161 All watchers are created by calling a method on the "AnyEvent" class.
… …
120 Many watchers either are used with "recursion" (repeating timers for 163 Many watchers either are used with "recursion" (repeating timers for
121 example), or need to refer to their watcher object in other ways. 164 example), or need to refer to their watcher object in other ways.
122 165
123 An any way to achieve that is this pattern: 166 An any way to achieve that is this pattern:
124 167
125 my $w; $w = AnyEvent->type (arg => value ..., cb => sub { 168 my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
126 # you can use $w here, for example to undef it 169 # you can use $w here, for example to undef it
127 undef $w; 170 undef $w;
128 }); 171 });
129 172
130 Note that "my $w; $w =" combination. This is necessary because in Perl, 173 Note that "my $w; $w =" combination. This is necessary because in Perl,
131 my variables are only visible after the statement in which they are 174 my variables are only visible after the statement in which they are
132 declared. 175 declared.
133 176
134 IO WATCHERS 177 I/O WATCHERS
178 $w = AnyEvent->io (
179 fh => <filehandle_or_fileno>,
180 poll => <"r" or "w">,
181 cb => <callback>,
182 );
183
135 You can create an I/O watcher by calling the "AnyEvent->io" method with 184 You can create an I/O watcher by calling the "AnyEvent->io" method with
136 the following mandatory key-value pairs as arguments: 185 the following mandatory key-value pairs as arguments:
137 186
138 "fh" the Perl *file handle* (*not* file descriptor) to watch for events. 187 "fh" is the Perl *file handle* (or a naked file descriptor) to watch for
188 events (AnyEvent might or might not keep a reference to this file
189 handle). Note that only file handles pointing to things for which
190 non-blocking operation makes sense are allowed. This includes sockets,
191 most character devices, pipes, fifos and so on, but not for example
192 files or block devices.
193
139 "poll" must be a string that is either "r" or "w", which creates a 194 "poll" must be a string that is either "r" or "w", which creates a
140 watcher waiting for "r"eadable or "w"ritable events, respectively. "cb" 195 watcher waiting for "r"eadable or "w"ritable events, respectively.
196
141 is the callback to invoke each time the file handle becomes ready. 197 "cb" is the callback to invoke each time the file handle becomes ready.
142 198
143 File handles will be kept alive, so as long as the watcher exists, the 199 Although the callback might get passed parameters, their value and
144 file handle exists, too. 200 presence is undefined and you cannot rely on them. Portable AnyEvent
201 callbacks cannot use arguments passed to I/O watcher callbacks.
145 202
203 The I/O watcher might use the underlying file descriptor or a copy of
146 It is not allowed to close a file handle as long as any watcher is 204 it. You must not close a file handle as long as any watcher is active on
147 active on the underlying file descriptor. 205 the underlying file descriptor.
148 206
149 Some event loops issue spurious readyness notifications, so you should 207 Some event loops issue spurious readyness notifications, so you should
150 always use non-blocking calls when reading/writing from/to your file 208 always use non-blocking calls when reading/writing from/to your file
151 handles. 209 handles.
152 210
153 Example:
154
155 # wait for readability of STDIN, then read a line and disable the watcher 211 Example: wait for readability of STDIN, then read a line and disable the
212 watcher.
213
156 my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub { 214 my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
157 chomp (my $input = <STDIN>); 215 chomp (my $input = <STDIN>);
158 warn "read: $input\n"; 216 warn "read: $input\n";
159 undef $w; 217 undef $w;
160 }); 218 });
161 219
162 TIME WATCHERS 220 TIME WATCHERS
221 $w = AnyEvent->timer (after => <seconds>, cb => <callback>);
222
223 $w = AnyEvent->timer (
224 after => <fractional_seconds>,
225 interval => <fractional_seconds>,
226 cb => <callback>,
227 );
228
163 You can create a time watcher by calling the "AnyEvent->timer" method 229 You can create a time watcher by calling the "AnyEvent->timer" method
164 with the following mandatory arguments: 230 with the following mandatory arguments:
165 231
166 "after" specifies after how many seconds (fractional values are 232 "after" specifies after how many seconds (fractional values are
167 supported) should the timer activate. "cb" the callback to invoke in 233 supported) the callback should be invoked. "cb" is the callback to
168 that case. 234 invoke in that case.
169 235
170 The timer callback will be invoked at most once: if you want a repeating 236 Although the callback might get passed parameters, their value and
171 timer you have to create a new watcher (this is a limitation by both Tk 237 presence is undefined and you cannot rely on them. Portable AnyEvent
172 and Glib). 238 callbacks cannot use arguments passed to time watcher callbacks.
173 239
174 Example: 240 The callback will normally be invoked once only. If you specify another
241 parameter, "interval", as a strictly positive number (> 0), then the
242 callback will be invoked regularly at that interval (in fractional
243 seconds) after the first invocation. If "interval" is specified with a
244 false value, then it is treated as if it were missing.
175 245
246 The callback will be rescheduled before invoking the callback, but no
247 attempt is done to avoid timer drift in most backends, so the interval
248 is only approximate.
249
176 # fire an event after 7.7 seconds 250 Example: fire an event after 7.7 seconds.
251
177 my $w = AnyEvent->timer (after => 7.7, cb => sub { 252 my $w = AnyEvent->timer (after => 7.7, cb => sub {
178 warn "timeout\n"; 253 warn "timeout\n";
179 }); 254 });
180 255
181 # to cancel the timer: 256 # to cancel the timer:
182 undef $w; 257 undef $w;
183 258
184 Example 2:
185
186 # fire an event after 0.5 seconds, then roughly every second 259 Example 2: fire an event after 0.5 seconds, then roughly every second.
187 my $w;
188 260
189 my $cb = sub {
190 # cancel the old timer while creating a new one
191 $w = AnyEvent->timer (after => 1, cb => $cb); 261 my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
262 warn "timeout\n";
192 }; 263 };
193
194 # start the "loop" by creating the first watcher
195 $w = AnyEvent->timer (after => 0.5, cb => $cb);
196 264
197 TIMING ISSUES 265 TIMING ISSUES
198 There are two ways to handle timers: based on real time (relative, "fire 266 There are two ways to handle timers: based on real time (relative, "fire
199 in 10 seconds") and based on wallclock time (absolute, "fire at 12 267 in 10 seconds") and based on wallclock time (absolute, "fire at 12
200 o'clock"). 268 o'clock").
201 269
202 While most event loops expect timers to specified in a relative way, 270 While most event loops expect timers to specified in a relative way,
203 they use absolute time internally. This makes a difference when your 271 they use absolute time internally. This makes a difference when your
204 clock "jumps", for example, when ntp decides to set your clock backwards 272 clock "jumps", for example, when ntp decides to set your clock backwards
205 from the wrong 2014-01-01 to 2008-01-01, a watcher that you created to 273 from the wrong date of 2014-01-01 to 2008-01-01, a watcher that is
206 fire "after" a second might actually take six years to finally fire. 274 supposed to fire "after" a second might actually take six years to
275 finally fire.
207 276
208 AnyEvent cannot compensate for this. The only event loop that is 277 AnyEvent cannot compensate for this. The only event loop that is
209 conscious about these issues is EV, which offers both relative 278 conscious about these issues is EV, which offers both relative
210 (ev_timer) and absolute (ev_periodic) timers. 279 (ev_timer, based on true relative time) and absolute (ev_periodic, based
280 on wallclock time) timers.
211 281
212 AnyEvent always prefers relative timers, if available, matching the 282 AnyEvent always prefers relative timers, if available, matching the
213 AnyEvent API. 283 AnyEvent API.
214 284
285 AnyEvent has two additional methods that return the "current time":
286
287 AnyEvent->time
288 This returns the "current wallclock time" as a fractional number of
289 seconds since the Epoch (the same thing as "time" or
290 "Time::HiRes::time" return, and the result is guaranteed to be
291 compatible with those).
292
293 It progresses independently of any event loop processing, i.e. each
294 call will check the system clock, which usually gets updated
295 frequently.
296
297 AnyEvent->now
298 This also returns the "current wallclock time", but unlike "time",
299 above, this value might change only once per event loop iteration,
300 depending on the event loop (most return the same time as "time",
301 above). This is the time that AnyEvent's timers get scheduled
302 against.
303
304 *In almost all cases (in all cases if you don't care), this is the
305 function to call when you want to know the current time.*
306
307 This function is also often faster then "AnyEvent->time", and thus
308 the preferred method if you want some timestamp (for example,
309 AnyEvent::Handle uses this to update it's activity timeouts).
310
311 The rest of this section is only of relevance if you try to be very
312 exact with your timing, you can skip it without bad conscience.
313
314 For a practical example of when these times differ, consider
315 Event::Lib and EV and the following set-up:
316
317 The event loop is running and has just invoked one of your callback
318 at time=500 (assume no other callbacks delay processing). In your
319 callback, you wait a second by executing "sleep 1" (blocking the
320 process for a second) and then (at time=501) you create a relative
321 timer that fires after three seconds.
322
323 With Event::Lib, "AnyEvent->time" and "AnyEvent->now" will both
324 return 501, because that is the current time, and the timer will be
325 scheduled to fire at time=504 (501 + 3).
326
327 With EV, "AnyEvent->time" returns 501 (as that is the current time),
328 but "AnyEvent->now" returns 500, as that is the time the last event
329 processing phase started. With EV, your timer gets scheduled to run
330 at time=503 (500 + 3).
331
332 In one sense, Event::Lib is more exact, as it uses the current time
333 regardless of any delays introduced by event processing. However,
334 most callbacks do not expect large delays in processing, so this
335 causes a higher drift (and a lot more system calls to get the
336 current time).
337
338 In another sense, EV is more exact, as your timer will be scheduled
339 at the same time, regardless of how long event processing actually
340 took.
341
342 In either case, if you care (and in most cases, you don't), then you
343 can get whatever behaviour you want with any event loop, by taking
344 the difference between "AnyEvent->time" and "AnyEvent->now" into
345 account.
346
347 AnyEvent->now_update
348 Some event loops (such as EV or AnyEvent::Impl::Perl) cache the
349 current time for each loop iteration (see the discussion of
350 AnyEvent->now, above).
351
352 When a callback runs for a long time (or when the process sleeps),
353 then this "current" time will differ substantially from the real
354 time, which might affect timers and time-outs.
355
356 When this is the case, you can call this method, which will update
357 the event loop's idea of "current time".
358
359 A typical example would be a script in a web server (e.g.
360 "mod_perl") - when mod_perl executes the script, then the event loop
361 will have the wrong idea about the "current time" (being potentially
362 far in the past, when the script ran the last time). In that case
363 you should arrange a call to "AnyEvent->now_update" each time the
364 web server process wakes up again (e.g. at the start of your script,
365 or in a handler).
366
367 Note that updating the time *might* cause some events to be handled.
368
215 SIGNAL WATCHERS 369 SIGNAL WATCHERS
370 $w = AnyEvent->signal (signal => <uppercase_signal_name>, cb => <callback>);
371
216 You can watch for signals using a signal watcher, "signal" is the signal 372 You can watch for signals using a signal watcher, "signal" is the signal
217 *name* without any "SIG" prefix, "cb" is the Perl callback to be invoked 373 *name* in uppercase and without any "SIG" prefix, "cb" is the Perl
218 whenever a signal occurs. 374 callback to be invoked whenever a signal occurs.
219 375
376 Although the callback might get passed parameters, their value and
377 presence is undefined and you cannot rely on them. Portable AnyEvent
378 callbacks cannot use arguments passed to signal watcher callbacks.
379
220 Multiple signals occurances can be clumped together into one callback 380 Multiple signal occurrences can be clumped together into one callback
221 invocation, and callback invocation will be synchronous. synchronous 381 invocation, and callback invocation will be synchronous. Synchronous
222 means that it might take a while until the signal gets handled by the 382 means that it might take a while until the signal gets handled by the
223 process, but it is guarenteed not to interrupt any other callbacks. 383 process, but it is guaranteed not to interrupt any other callbacks.
224 384
225 The main advantage of using these watchers is that you can share a 385 The main advantage of using these watchers is that you can share a
226 signal between multiple watchers. 386 signal between multiple watchers, and AnyEvent will ensure that signals
387 will not interrupt your program at bad times.
227 388
228 This watcher might use %SIG, so programs overwriting those signals 389 This watcher might use %SIG (depending on the event loop used), so
229 directly will likely not work correctly. 390 programs overwriting those signals directly will likely not work
391 correctly.
230 392
231 Example: exit on SIGINT 393 Example: exit on SIGINT
232 394
233 my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 }); 395 my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
234 396
397 Restart Behaviour
398 While restart behaviour is up to the event loop implementation, most
399 will not restart syscalls (that includes Async::Interrupt and AnyEvent's
400 pure perl implementation).
401
402 Safe/Unsafe Signals
403 Perl signals can be either "safe" (synchronous to opcode handling) or
404 "unsafe" (asynchronous) - the former might get delayed indefinitely, the
405 latter might corrupt your memory.
406
407 AnyEvent signal handlers are, in addition, synchronous to the event
408 loop, i.e. they will not interrupt your running perl program but will
409 only be called as part of the normal event handling (just like timer,
410 I/O etc. callbacks, too).
411
412 Signal Races, Delays and Workarounds
413 Many event loops (e.g. Glib, Tk, Qt, IO::Async) do not support attaching
414 callbacks to signals in a generic way, which is a pity, as you cannot do
415 race-free signal handling in perl, requiring C libraries for this.
416 AnyEvent will try to do it's best, which means in some cases, signals
417 will be delayed. The maximum time a signal might be delayed is specified
418 in $AnyEvent::MAX_SIGNAL_LATENCY (default: 10 seconds). This variable
419 can be changed only before the first signal watcher is created, and
420 should be left alone otherwise. This variable determines how often
421 AnyEvent polls for signals (in case a wake-up was missed). Higher values
422 will cause fewer spurious wake-ups, which is better for power and CPU
423 saving.
424
425 All these problems can be avoided by installing the optional
426 Async::Interrupt module, which works with most event loops. It will not
427 work with inherently broken event loops such as Event or Event::Lib (and
428 not with POE currently, as POE does it's own workaround with one-second
429 latency). For those, you just have to suffer the delays.
430
235 CHILD PROCESS WATCHERS 431 CHILD PROCESS WATCHERS
432 $w = AnyEvent->child (pid => <process id>, cb => <callback>);
433
236 You can also watch on a child process exit and catch its exit status. 434 You can also watch on a child process exit and catch its exit status.
237 435
238 The child process is specified by the "pid" argument (if set to 0, it 436 The child process is specified by the "pid" argument (one some backends,
239 watches for any child process exit). The watcher will trigger as often 437 using 0 watches for any child process exit, on others this will croak).
240 as status change for the child are received. This works by installing a 438 The watcher will be triggered only when the child process has finished
241 signal handler for "SIGCHLD". The callback will be called with the pid 439 and an exit status is available, not on any trace events
242 and exit status (as returned by waitpid). 440 (stopped/continued).
243 441
244 Example: wait for pid 1333 442 The callback will be called with the pid and exit status (as returned by
443 waitpid), so unlike other watcher types, you *can* rely on child watcher
444 callback arguments.
245 445
446 This watcher type works by installing a signal handler for "SIGCHLD",
447 and since it cannot be shared, nothing else should use SIGCHLD or reap
448 random child processes (waiting for specific child processes, e.g.
449 inside "system", is just fine).
450
451 There is a slight catch to child watchers, however: you usually start
452 them *after* the child process was created, and this means the process
453 could have exited already (and no SIGCHLD will be sent anymore).
454
455 Not all event models handle this correctly (neither POE nor IO::Async
456 do, see their AnyEvent::Impl manpages for details), but even for event
457 models that *do* handle this correctly, they usually need to be loaded
458 before the process exits (i.e. before you fork in the first place).
459 AnyEvent's pure perl event loop handles all cases correctly regardless
460 of when you start the watcher.
461
462 This means you cannot create a child watcher as the very first thing in
463 an AnyEvent program, you *have* to create at least one watcher before
464 you "fork" the child (alternatively, you can call "AnyEvent::detect").
465
466 As most event loops do not support waiting for child events, they will
467 be emulated by AnyEvent in most cases, in which the latency and race
468 problems mentioned in the description of signal watchers apply.
469
470 Example: fork a process and wait for it
471
472 my $done = AnyEvent->condvar;
473
474 my $pid = fork or exit 5;
475
246 my $w = AnyEvent->child ( 476 my $w = AnyEvent->child (
247 pid => 1333, 477 pid => $pid,
248 cb => sub { 478 cb => sub {
249 my ($pid, $status) = @_; 479 my ($pid, $status) = @_;
250 warn "pid $pid exited with status $status"; 480 warn "pid $pid exited with status $status";
481 $done->send;
251 }, 482 },
252 ); 483 );
484
485 # do something else, then wait for process exit
486 $done->recv;
487
488 IDLE WATCHERS
489 $w = AnyEvent->idle (cb => <callback>);
490
491 Sometimes there is a need to do something, but it is not so important to
492 do it instantly, but only when there is nothing better to do. This
493 "nothing better to do" is usually defined to be "no other events need
494 attention by the event loop".
495
496 Idle watchers ideally get invoked when the event loop has nothing better
497 to do, just before it would block the process to wait for new events.
498 Instead of blocking, the idle watcher is invoked.
499
500 Most event loops unfortunately do not really support idle watchers (only
501 EV, Event and Glib do it in a usable fashion) - for the rest, AnyEvent
502 will simply call the callback "from time to time".
503
504 Example: read lines from STDIN, but only process them when the program
505 is otherwise idle:
506
507 my @lines; # read data
508 my $idle_w;
509 my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
510 push @lines, scalar <STDIN>;
511
512 # start an idle watcher, if not already done
513 $idle_w ||= AnyEvent->idle (cb => sub {
514 # handle only one line, when there are lines left
515 if (my $line = shift @lines) {
516 print "handled when idle: $line";
517 } else {
518 # otherwise disable the idle watcher again
519 undef $idle_w;
520 }
521 });
522 });
253 523
254 CONDITION VARIABLES 524 CONDITION VARIABLES
525 $cv = AnyEvent->condvar;
526
527 $cv->send (<list>);
528 my @res = $cv->recv;
529
530 If you are familiar with some event loops you will know that all of them
531 require you to run some blocking "loop", "run" or similar function that
532 will actively watch for new events and call your callbacks.
533
534 AnyEvent is slightly different: it expects somebody else to run the
535 event loop and will only block when necessary (usually when told by the
536 user).
537
538 The instrument to do that is called a "condition variable", so called
539 because they represent a condition that must become true.
540
541 Now is probably a good time to look at the examples further below.
542
255 Condition variables can be created by calling the "AnyEvent->condvar" 543 Condition variables can be created by calling the "AnyEvent->condvar"
256 method without any arguments. 544 method, usually without arguments. The only argument pair allowed is
545 "cb", which specifies a callback to be called when the condition
546 variable becomes true, with the condition variable as the first argument
547 (but not the results).
257 548
258 A condition variable waits for a condition - precisely that the 549 After creation, the condition variable is "false" until it becomes
259 "->broadcast" method has been called. 550 "true" by calling the "send" method (or calling the condition variable
551 as if it were a callback, read about the caveats in the description for
552 the "->send" method).
260 553
261 They are very useful to signal that a condition has been fulfilled, for 554 Condition variables are similar to callbacks, except that you can
555 optionally wait for them. They can also be called merge points - points
556 in time where multiple outstanding events have been processed. And yet
557 another way to call them is transactions - each condition variable can
558 be used to represent a transaction, which finishes at some point and
559 delivers a result. And yet some people know them as "futures" - a
560 promise to compute/deliver something that you can wait for.
561
562 Condition variables are very useful to signal that something has
262 example, if you write a module that does asynchronous http requests, 563 finished, for example, if you write a module that does asynchronous http
263 then a condition variable would be the ideal candidate to signal the 564 requests, then a condition variable would be the ideal candidate to
264 availability of results. 565 signal the availability of results. The user can either act when the
566 callback is called or can synchronously "->recv" for the results.
265 567
266 You can also use condition variables to block your main program until an 568 You can also use them to simulate traditional event loops - for example,
267 event occurs - for example, you could "->wait" in your main program 569 you can block your main program until an event occurs - for example, you
268 until the user clicks the Quit button in your app, which would 570 could "->recv" in your main program until the user clicks the Quit
269 "->broadcast" the "quit" event. 571 button of your app, which would "->send" the "quit" event.
270 572
271 Note that condition variables recurse into the event loop - if you have 573 Note that condition variables recurse into the event loop - if you have
272 two pirces of code that call "->wait" in a round-robbin fashion, you 574 two pieces of code that call "->recv" in a round-robin fashion, you
273 lose. Therefore, condition variables are good to export to your caller, 575 lose. Therefore, condition variables are good to export to your caller,
274 but you should avoid making a blocking wait yourself, at least in 576 but you should avoid making a blocking wait yourself, at least in
275 callbacks, as this asks for trouble. 577 callbacks, as this asks for trouble.
276 578
277 This object has two methods: 579 Condition variables are represented by hash refs in perl, and the keys
580 used by AnyEvent itself are all named "_ae_XXX" to make subclassing easy
581 (it is often useful to build your own transaction class on top of
582 AnyEvent). To subclass, use "AnyEvent::CondVar" as base class and call
583 it's "new" method in your own "new" method.
278 584
279 $cv->wait 585 There are two "sides" to a condition variable - the "producer side"
280 Wait (blocking if necessary) until the "->broadcast" method has been 586 which eventually calls "-> send", and the "consumer side", which waits
281 called on c<$cv>, while servicing other watchers normally. 587 for the send to occur.
282 588
283 You can only wait once on a condition - additional calls will return 589 Example: wait for a timer.
284 immediately.
285
286 Not all event models support a blocking wait - some die in that case
287 (programs might want to do that to stay interactive), so *if you are
288 using this from a module, never require a blocking wait*, but let
289 the caller decide whether the call will block or not (for example,
290 by coupling condition variables with some kind of request results
291 and supporting callbacks so the caller knows that getting the result
292 will not block, while still suppporting blocking waits if the caller
293 so desires).
294
295 Another reason *never* to "->wait" in a module is that you cannot
296 sensibly have two "->wait"'s in parallel, as that would require
297 multiple interpreters or coroutines/threads, none of which
298 "AnyEvent" can supply (the coroutine-aware backends
299 AnyEvent::Impl::CoroEV and AnyEvent::Impl::CoroEvent explicitly
300 support concurrent "->wait"'s from different coroutines, however).
301
302 $cv->broadcast
303 Flag the condition as ready - a running "->wait" and all further
304 calls to "wait" will (eventually) return after this method has been
305 called. If nobody is waiting the broadcast will be remembered..
306
307 Example:
308 590
309 # wait till the result is ready 591 # wait till the result is ready
310 my $result_ready = AnyEvent->condvar; 592 my $result_ready = AnyEvent->condvar;
311 593
312 # do something such as adding a timer 594 # do something such as adding a timer
313 # or socket watcher the calls $result_ready->broadcast 595 # or socket watcher the calls $result_ready->send
314 # when the "result" is ready. 596 # when the "result" is ready.
315 # in this case, we simply use a timer: 597 # in this case, we simply use a timer:
316 my $w = AnyEvent->timer ( 598 my $w = AnyEvent->timer (
317 after => 1, 599 after => 1,
318 cb => sub { $result_ready->broadcast }, 600 cb => sub { $result_ready->send },
319 ); 601 );
320 602
321 # this "blocks" (while handling events) till the watcher 603 # this "blocks" (while handling events) till the callback
322 # calls broadcast 604 # calls ->send
323 $result_ready->wait; 605 $result_ready->recv;
606
607 Example: wait for a timer, but take advantage of the fact that condition
608 variables are also callable directly.
609
610 my $done = AnyEvent->condvar;
611 my $delay = AnyEvent->timer (after => 5, cb => $done);
612 $done->recv;
613
614 Example: Imagine an API that returns a condvar and doesn't support
615 callbacks. This is how you make a synchronous call, for example from the
616 main program:
617
618 use AnyEvent::CouchDB;
619
620 ...
621
622 my @info = $couchdb->info->recv;
623
624 And this is how you would just set a callback to be called whenever the
625 results are available:
626
627 $couchdb->info->cb (sub {
628 my @info = $_[0]->recv;
629 });
630
631 METHODS FOR PRODUCERS
632 These methods should only be used by the producing side, i.e. the
633 code/module that eventually sends the signal. Note that it is also the
634 producer side which creates the condvar in most cases, but it isn't
635 uncommon for the consumer to create it as well.
636
637 $cv->send (...)
638 Flag the condition as ready - a running "->recv" and all further
639 calls to "recv" will (eventually) return after this method has been
640 called. If nobody is waiting the send will be remembered.
641
642 If a callback has been set on the condition variable, it is called
643 immediately from within send.
644
645 Any arguments passed to the "send" call will be returned by all
646 future "->recv" calls.
647
648 Condition variables are overloaded so one can call them directly (as
649 if they were a code reference). Calling them directly is the same as
650 calling "send".
651
652 $cv->croak ($error)
653 Similar to send, but causes all call's to "->recv" to invoke
654 "Carp::croak" with the given error message/object/scalar.
655
656 This can be used to signal any errors to the condition variable
657 user/consumer. Doing it this way instead of calling "croak" directly
658 delays the error detetcion, but has the overwhelmign advantage that
659 it diagnoses the error at the place where the result is expected,
660 and not deep in some event clalback without connection to the actual
661 code causing the problem.
662
663 $cv->begin ([group callback])
664 $cv->end
665 These two methods can be used to combine many transactions/events
666 into one. For example, a function that pings many hosts in parallel
667 might want to use a condition variable for the whole process.
668
669 Every call to "->begin" will increment a counter, and every call to
670 "->end" will decrement it. If the counter reaches 0 in "->end", the
671 (last) callback passed to "begin" will be executed, passing the
672 condvar as first argument. That callback is *supposed* to call
673 "->send", but that is not required. If no group callback was set,
674 "send" will be called without any arguments.
675
676 You can think of "$cv->send" giving you an OR condition (one call
677 sends), while "$cv->begin" and "$cv->end" giving you an AND
678 condition (all "begin" calls must be "end"'ed before the condvar
679 sends).
680
681 Let's start with a simple example: you have two I/O watchers (for
682 example, STDOUT and STDERR for a program), and you want to wait for
683 both streams to close before activating a condvar:
684
685 my $cv = AnyEvent->condvar;
686
687 $cv->begin; # first watcher
688 my $w1 = AnyEvent->io (fh => $fh1, cb => sub {
689 defined sysread $fh1, my $buf, 4096
690 or $cv->end;
691 });
692
693 $cv->begin; # second watcher
694 my $w2 = AnyEvent->io (fh => $fh2, cb => sub {
695 defined sysread $fh2, my $buf, 4096
696 or $cv->end;
697 });
698
699 $cv->recv;
700
701 This works because for every event source (EOF on file handle),
702 there is one call to "begin", so the condvar waits for all calls to
703 "end" before sending.
704
705 The ping example mentioned above is slightly more complicated, as
706 the there are results to be passwd back, and the number of tasks
707 that are begung can potentially be zero:
708
709 my $cv = AnyEvent->condvar;
710
711 my %result;
712 $cv->begin (sub { shift->send (\%result) });
713
714 for my $host (@list_of_hosts) {
715 $cv->begin;
716 ping_host_then_call_callback $host, sub {
717 $result{$host} = ...;
718 $cv->end;
719 };
720 }
721
722 $cv->end;
723
724 This code fragment supposedly pings a number of hosts and calls
725 "send" after results for all then have have been gathered - in any
726 order. To achieve this, the code issues a call to "begin" when it
727 starts each ping request and calls "end" when it has received some
728 result for it. Since "begin" and "end" only maintain a counter, the
729 order in which results arrive is not relevant.
730
731 There is an additional bracketing call to "begin" and "end" outside
732 the loop, which serves two important purposes: first, it sets the
733 callback to be called once the counter reaches 0, and second, it
734 ensures that "send" is called even when "no" hosts are being pinged
735 (the loop doesn't execute once).
736
737 This is the general pattern when you "fan out" into multiple (but
738 potentially none) subrequests: use an outer "begin"/"end" pair to
739 set the callback and ensure "end" is called at least once, and then,
740 for each subrequest you start, call "begin" and for each subrequest
741 you finish, call "end".
742
743 METHODS FOR CONSUMERS
744 These methods should only be used by the consuming side, i.e. the code
745 awaits the condition.
746
747 $cv->recv
748 Wait (blocking if necessary) until the "->send" or "->croak" methods
749 have been called on c<$cv>, while servicing other watchers normally.
750
751 You can only wait once on a condition - additional calls are valid
752 but will return immediately.
753
754 If an error condition has been set by calling "->croak", then this
755 function will call "croak".
756
757 In list context, all parameters passed to "send" will be returned,
758 in scalar context only the first one will be returned.
759
760 Note that doing a blocking wait in a callback is not supported by
761 any event loop, that is, recursive invocation of a blocking "->recv"
762 is not allowed, and the "recv" call will "croak" if such a condition
763 is detected. This condition can be slightly loosened by using
764 Coro::AnyEvent, which allows you to do a blocking "->recv" from any
765 thread that doesn't run the event loop itself.
766
767 Not all event models support a blocking wait - some die in that case
768 (programs might want to do that to stay interactive), so *if you are
769 using this from a module, never require a blocking wait*. Instead,
770 let the caller decide whether the call will block or not (for
771 example, by coupling condition variables with some kind of request
772 results and supporting callbacks so the caller knows that getting
773 the result will not block, while still supporting blocking waits if
774 the caller so desires).
775
776 You can ensure that "-recv" never blocks by setting a callback and
777 only calling "->recv" from within that callback (or at a later
778 time). This will work even when the event loop does not support
779 blocking waits otherwise.
780
781 $bool = $cv->ready
782 Returns true when the condition is "true", i.e. whether "send" or
783 "croak" have been called.
784
785 $cb = $cv->cb ($cb->($cv))
786 This is a mutator function that returns the callback set and
787 optionally replaces it before doing so.
788
789 The callback will be called when the condition becomes (or already
790 was) "true", i.e. when "send" or "croak" are called (or were
791 called), with the only argument being the condition variable itself.
792 Calling "recv" inside the callback or at any later time is
793 guaranteed not to block.
794
795SUPPORTED EVENT LOOPS/BACKENDS
796 The available backend classes are (every class has its own manpage):
797
798 Backends that are autoprobed when no other event loop can be found.
799 EV is the preferred backend when no other event loop seems to be in
800 use. If EV is not installed, then AnyEvent will fall back to its own
801 pure-perl implementation, which is available everywhere as it comes
802 with AnyEvent itself.
803
804 AnyEvent::Impl::EV based on EV (interface to libev, best choice).
805 AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
806
807 Backends that are transparently being picked up when they are used.
808 These will be used when they are currently loaded when the first
809 watcher is created, in which case it is assumed that the application
810 is using them. This means that AnyEvent will automatically pick the
811 right backend when the main program loads an event module before
812 anything starts to create watchers. Nothing special needs to be done
813 by the main program.
814
815 AnyEvent::Impl::Event based on Event, very stable, few glitches.
816 AnyEvent::Impl::Glib based on Glib, slow but very stable.
817 AnyEvent::Impl::Tk based on Tk, very broken.
818 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
819 AnyEvent::Impl::POE based on POE, very slow, some limitations.
820 AnyEvent::Impl::Irssi used when running within irssi.
821
822 Backends with special needs.
823 Qt requires the Qt::Application to be instantiated first, but will
824 otherwise be picked up automatically. As long as the main program
825 instantiates the application before any AnyEvent watchers are
826 created, everything should just work.
827
828 AnyEvent::Impl::Qt based on Qt.
829
830 Support for IO::Async can only be partial, as it is too broken and
831 architecturally limited to even support the AnyEvent API. It also is
832 the only event loop that needs the loop to be set explicitly, so it
833 can only be used by a main program knowing about AnyEvent. See
834 AnyEvent::Impl::Async for the gory details.
835
836 AnyEvent::Impl::IOAsync based on IO::Async, cannot be autoprobed.
837
838 Event loops that are indirectly supported via other backends.
839 Some event loops can be supported via other modules:
840
841 There is no direct support for WxWidgets (Wx) or Prima.
842
843 WxWidgets has no support for watching file handles. However, you can
844 use WxWidgets through the POE adaptor, as POE has a Wx backend that
845 simply polls 20 times per second, which was considered to be too
846 horrible to even consider for AnyEvent.
847
848 Prima is not supported as nobody seems to be using it, but it has a
849 POE backend, so it can be supported through POE.
850
851 AnyEvent knows about both Prima and Wx, however, and will try to
852 load POE when detecting them, in the hope that POE will pick them
853 up, in which case everything will be automatic.
324 854
325GLOBAL VARIABLES AND FUNCTIONS 855GLOBAL VARIABLES AND FUNCTIONS
856 These are not normally required to use AnyEvent, but can be useful to
857 write AnyEvent extension modules.
858
326 $AnyEvent::MODEL 859 $AnyEvent::MODEL
327 Contains "undef" until the first watcher is being created. Then it 860 Contains "undef" until the first watcher is being created, before
861 the backend has been autodetected.
862
328 contains the event model that is being used, which is the name of 863 Afterwards it contains the event model that is being used, which is
329 the Perl class implementing the model. This class is usually one of 864 the name of the Perl class implementing the model. This class is
330 the "AnyEvent::Impl:xxx" modules, but can be any other class in the 865 usually one of the "AnyEvent::Impl:xxx" modules, but can be any
331 case AnyEvent has been extended at runtime (e.g. in *rxvt-unicode*). 866 other class in the case AnyEvent has been extended at runtime (e.g.
332 867 in *rxvt-unicode* it will be "urxvt::anyevent").
333 The known classes so far are:
334
335 AnyEvent::Impl::CoroEV based on Coro::EV, best choice.
336 AnyEvent::Impl::CoroEvent based on Coro::Event, second best choice.
337 AnyEvent::Impl::EV based on EV (an interface to libev, also best choice).
338 AnyEvent::Impl::Event based on Event, also second best choice :)
339 AnyEvent::Impl::Glib based on Glib, third-best choice.
340 AnyEvent::Impl::Tk based on Tk, very bad choice.
341 AnyEvent::Impl::Perl pure-perl implementation, inefficient but portable.
342 868
343 AnyEvent::detect 869 AnyEvent::detect
344 Returns $AnyEvent::MODEL, forcing autodetection of the event model 870 Returns $AnyEvent::MODEL, forcing autodetection of the event model
345 if necessary. You should only call this function right before you 871 if necessary. You should only call this function right before you
346 would have created an AnyEvent watcher anyway, that is, as late as 872 would have created an AnyEvent watcher anyway, that is, as late as
347 possible at runtime. 873 possible at runtime, and not e.g. while initialising of your module.
874
875 If you need to do some initialisation before AnyEvent watchers are
876 created, use "post_detect".
877
878 $guard = AnyEvent::post_detect { BLOCK }
879 Arranges for the code block to be executed as soon as the event
880 model is autodetected (or immediately if this has already happened).
881
882 The block will be executed *after* the actual backend has been
883 detected ($AnyEvent::MODEL is set), but *before* any watchers have
884 been created, so it is possible to e.g. patch @AnyEvent::ISA or do
885 other initialisations - see the sources of AnyEvent::Strict or
886 AnyEvent::AIO to see how this is used.
887
888 The most common usage is to create some global watchers, without
889 forcing event module detection too early, for example, AnyEvent::AIO
890 creates and installs the global IO::AIO watcher in a "post_detect"
891 block to avoid autodetecting the event module at load time.
892
893 If called in scalar or list context, then it creates and returns an
894 object that automatically removes the callback again when it is
895 destroyed (or "undef" when the hook was immediately executed). See
896 AnyEvent::AIO for a case where this is useful.
897
898 Example: Create a watcher for the IO::AIO module and store it in
899 $WATCHER. Only do so after the event loop is initialised, though.
900
901 our WATCHER;
902
903 my $guard = AnyEvent::post_detect {
904 $WATCHER = AnyEvent->io (fh => IO::AIO::poll_fileno, poll => 'r', cb => \&IO::AIO::poll_cb);
905 };
906
907 # the ||= is important in case post_detect immediately runs the block,
908 # as to not clobber the newly-created watcher. assigning both watcher and
909 # post_detect guard to the same variable has the advantage of users being
910 # able to just C<undef $WATCHER> if the watcher causes them grief.
911
912 $WATCHER ||= $guard;
913
914 @AnyEvent::post_detect
915 If there are any code references in this array (you can "push" to it
916 before or after loading AnyEvent), then they will called directly
917 after the event loop has been chosen.
918
919 You should check $AnyEvent::MODEL before adding to this array,
920 though: if it is defined then the event loop has already been
921 detected, and the array will be ignored.
922
923 Best use "AnyEvent::post_detect { BLOCK }" when your application
924 allows it, as it takes care of these details.
925
926 This variable is mainly useful for modules that can do something
927 useful when AnyEvent is used and thus want to know when it is
928 initialised, but do not need to even load it by default. This array
929 provides the means to hook into AnyEvent passively, without loading
930 it.
931
932 Example: To load Coro::AnyEvent whenever Coro and AnyEvent are used
933 together, you could put this into Coro (this is the actual code used
934 by Coro to accomplish this):
935
936 if (defined $AnyEvent::MODEL) {
937 # AnyEvent already initialised, so load Coro::AnyEvent
938 require Coro::AnyEvent;
939 } else {
940 # AnyEvent not yet initialised, so make sure to load Coro::AnyEvent
941 # as soon as it is
942 push @AnyEvent::post_detect, sub { require Coro::AnyEvent };
943 }
348 944
349WHAT TO DO IN A MODULE 945WHAT TO DO IN A MODULE
350 As a module author, you should "use AnyEvent" and call AnyEvent methods 946 As a module author, you should "use AnyEvent" and call AnyEvent methods
351 freely, but you should not load a specific event module or rely on it. 947 freely, but you should not load a specific event module or rely on it.
352 948
353 Be careful when you create watchers in the module body - AnyEvent will 949 Be careful when you create watchers in the module body - AnyEvent will
354 decide which event module to use as soon as the first method is called, 950 decide which event module to use as soon as the first method is called,
355 so by calling AnyEvent in your module body you force the user of your 951 so by calling AnyEvent in your module body you force the user of your
356 module to load the event module first. 952 module to load the event module first.
357 953
358 Never call "->wait" on a condition variable unless you *know* that the 954 Never call "->recv" on a condition variable unless you *know* that the
359 "->broadcast" method has been called on it already. This is because it 955 "->send" method has been called on it already. This is because it will
360 will stall the whole program, and the whole point of using events is to 956 stall the whole program, and the whole point of using events is to stay
361 stay interactive. 957 interactive.
362 958
363 It is fine, however, to call "->wait" when the user of your module 959 It is fine, however, to call "->recv" when the user of your module
364 requests it (i.e. if you create a http request object ad have a method 960 requests it (i.e. if you create a http request object ad have a method
365 called "results" that returns the results, it should call "->wait" 961 called "results" that returns the results, it should call "->recv"
366 freely, as the user of your module knows what she is doing. always). 962 freely, as the user of your module knows what she is doing. always).
367 963
368WHAT TO DO IN THE MAIN PROGRAM 964WHAT TO DO IN THE MAIN PROGRAM
369 There will always be a single main program - the only place that should 965 There will always be a single main program - the only place that should
370 dictate which event model to use. 966 dictate which event model to use.
… …
372 If it doesn't care, it can just "use AnyEvent" and use it itself, or not 968 If it doesn't care, it can just "use AnyEvent" and use it itself, or not
373 do anything special (it does not need to be event-based) and let 969 do anything special (it does not need to be event-based) and let
374 AnyEvent decide which implementation to chose if some module relies on 970 AnyEvent decide which implementation to chose if some module relies on
375 it. 971 it.
376 972
377 If the main program relies on a specific event model. For example, in 973 If the main program relies on a specific event model - for example, in
378 Gtk2 programs you have to rely on the Glib module. You should load the 974 Gtk2 programs you have to rely on the Glib module - you should load the
379 event module before loading AnyEvent or any module that uses it: 975 event module before loading AnyEvent or any module that uses it:
380 generally speaking, you should load it as early as possible. The reason 976 generally speaking, you should load it as early as possible. The reason
381 is that modules might create watchers when they are loaded, and AnyEvent 977 is that modules might create watchers when they are loaded, and AnyEvent
382 will decide on the event model to use as soon as it creates watchers, 978 will decide on the event model to use as soon as it creates watchers,
383 and it might chose the wrong one unless you load the correct one 979 and it might chose the wrong one unless you load the correct one
384 yourself. 980 yourself.
385 981
386 You can chose to use a rather inefficient pure-perl implementation by 982 You can chose to use a pure-perl implementation by loading the
387 loading the "AnyEvent::Impl::Perl" module, which gives you similar 983 "AnyEvent::Impl::Perl" module, which gives you similar behaviour
388 behaviour everywhere, but letting AnyEvent chose is generally better. 984 everywhere, but letting AnyEvent chose the model is generally better.
985
986 MAINLOOP EMULATION
987 Sometimes (often for short test scripts, or even standalone programs who
988 only want to use AnyEvent), you do not want to run a specific event
989 loop.
990
991 In that case, you can use a condition variable like this:
992
993 AnyEvent->condvar->recv;
994
995 This has the effect of entering the event loop and looping forever.
996
997 Note that usually your program has some exit condition, in which case it
998 is better to use the "traditional" approach of storing a condition
999 variable somewhere, waiting for it, and sending it when the program
1000 should exit cleanly.
1001
1002OTHER MODULES
1003 The following is a non-exhaustive list of additional modules that use
1004 AnyEvent as a client and can therefore be mixed easily with other
1005 AnyEvent modules and other event loops in the same program. Some of the
1006 modules come with AnyEvent, most are available via CPAN.
1007
1008 AnyEvent::Util
1009 Contains various utility functions that replace often-used but
1010 blocking functions such as "inet_aton" by event-/callback-based
1011 versions.
1012
1013 AnyEvent::Socket
1014 Provides various utility functions for (internet protocol) sockets,
1015 addresses and name resolution. Also functions to create non-blocking
1016 tcp connections or tcp servers, with IPv6 and SRV record support and
1017 more.
1018
1019 AnyEvent::Handle
1020 Provide read and write buffers, manages watchers for reads and
1021 writes, supports raw and formatted I/O, I/O queued and fully
1022 transparent and non-blocking SSL/TLS (via AnyEvent::TLS.
1023
1024 AnyEvent::DNS
1025 Provides rich asynchronous DNS resolver capabilities.
1026
1027 AnyEvent::HTTP
1028 A simple-to-use HTTP library that is capable of making a lot of
1029 concurrent HTTP requests.
1030
1031 AnyEvent::HTTPD
1032 Provides a simple web application server framework.
1033
1034 AnyEvent::FastPing
1035 The fastest ping in the west.
1036
1037 AnyEvent::DBI
1038 Executes DBI requests asynchronously in a proxy process.
1039
1040 AnyEvent::AIO
1041 Truly asynchronous I/O, should be in the toolbox of every event
1042 programmer. AnyEvent::AIO transparently fuses IO::AIO and AnyEvent
1043 together.
1044
1045 AnyEvent::BDB
1046 Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently
1047 fuses BDB and AnyEvent together.
1048
1049 AnyEvent::GPSD
1050 A non-blocking interface to gpsd, a daemon delivering GPS
1051 information.
1052
1053 AnyEvent::IRC
1054 AnyEvent based IRC client module family (replacing the older
1055 Net::IRC3).
1056
1057 AnyEvent::XMPP
1058 AnyEvent based XMPP (Jabber protocol) module family (replacing the
1059 older Net::XMPP2>.
1060
1061 AnyEvent::IGS
1062 A non-blocking interface to the Internet Go Server protocol (used by
1063 App::IGS).
1064
1065 Net::FCP
1066 AnyEvent-based implementation of the Freenet Client Protocol,
1067 birthplace of AnyEvent.
1068
1069 Event::ExecFlow
1070 High level API for event-based execution flow control.
1071
1072 Coro
1073 Has special support for AnyEvent via Coro::AnyEvent.
1074
1075SIMPLIFIED AE API
1076 Starting with version 5.0, AnyEvent officially supports a second, much
1077 simpler, API that is designed to reduce the calling, typing and memory
1078 overhead.
1079
1080 See the AE manpage for details.
1081
1082ERROR AND EXCEPTION HANDLING
1083 In general, AnyEvent does not do any error handling - it relies on the
1084 caller to do that if required. The AnyEvent::Strict module (see also the
1085 "PERL_ANYEVENT_STRICT" environment variable, below) provides strict
1086 checking of all AnyEvent methods, however, which is highly useful during
1087 development.
1088
1089 As for exception handling (i.e. runtime errors and exceptions thrown
1090 while executing a callback), this is not only highly event-loop
1091 specific, but also not in any way wrapped by this module, as this is the
1092 job of the main program.
1093
1094 The pure perl event loop simply re-throws the exception (usually within
1095 "condvar->recv"), the Event and EV modules call "$Event/EV::DIED->()",
1096 Glib uses "install_exception_handler" and so on.
1097
1098ENVIRONMENT VARIABLES
1099 The following environment variables are used by this module or its
1100 submodules.
1101
1102 Note that AnyEvent will remove *all* environment variables starting with
1103 "PERL_ANYEVENT_" from %ENV when it is loaded while taint mode is
1104 enabled.
1105
1106 "PERL_ANYEVENT_VERBOSE"
1107 By default, AnyEvent will be completely silent except in fatal
1108 conditions. You can set this environment variable to make AnyEvent
1109 more talkative.
1110
1111 When set to 1 or higher, causes AnyEvent to warn about unexpected
1112 conditions, such as not being able to load the event model specified
1113 by "PERL_ANYEVENT_MODEL".
1114
1115 When set to 2 or higher, cause AnyEvent to report to STDERR which
1116 event model it chooses.
1117
1118 When set to 8 or higher, then AnyEvent will report extra information
1119 on which optional modules it loads and how it implements certain
1120 features.
1121
1122 "PERL_ANYEVENT_STRICT"
1123 AnyEvent does not do much argument checking by default, as thorough
1124 argument checking is very costly. Setting this variable to a true
1125 value will cause AnyEvent to load "AnyEvent::Strict" and then to
1126 thoroughly check the arguments passed to most method calls. If it
1127 finds any problems, it will croak.
1128
1129 In other words, enables "strict" mode.
1130
1131 Unlike "use strict" (or it's modern cousin, "use common::sense", it
1132 is definitely recommended to keep it off in production. Keeping
1133 "PERL_ANYEVENT_STRICT=1" in your environment while developing
1134 programs can be very useful, however.
1135
1136 "PERL_ANYEVENT_MODEL"
1137 This can be used to specify the event model to be used by AnyEvent,
1138 before auto detection and -probing kicks in. It must be a string
1139 consisting entirely of ASCII letters. The string "AnyEvent::Impl::"
1140 gets prepended and the resulting module name is loaded and if the
1141 load was successful, used as event model. If it fails to load
1142 AnyEvent will proceed with auto detection and -probing.
1143
1144 This functionality might change in future versions.
1145
1146 For example, to force the pure perl model (AnyEvent::Impl::Perl) you
1147 could start your program like this:
1148
1149 PERL_ANYEVENT_MODEL=Perl perl ...
1150
1151 "PERL_ANYEVENT_PROTOCOLS"
1152 Used by both AnyEvent::DNS and AnyEvent::Socket to determine
1153 preferences for IPv4 or IPv6. The default is unspecified (and might
1154 change, or be the result of auto probing).
1155
1156 Must be set to a comma-separated list of protocols or address
1157 families, current supported: "ipv4" and "ipv6". Only protocols
1158 mentioned will be used, and preference will be given to protocols
1159 mentioned earlier in the list.
1160
1161 This variable can effectively be used for denial-of-service attacks
1162 against local programs (e.g. when setuid), although the impact is
1163 likely small, as the program has to handle conenction and other
1164 failures anyways.
1165
1166 Examples: "PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6" - prefer IPv4 over
1167 IPv6, but support both and try to use both.
1168 "PERL_ANYEVENT_PROTOCOLS=ipv4" - only support IPv4, never try to
1169 resolve or contact IPv6 addresses.
1170 "PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4" support either IPv4 or IPv6, but
1171 prefer IPv6 over IPv4.
1172
1173 "PERL_ANYEVENT_EDNS0"
1174 Used by AnyEvent::DNS to decide whether to use the EDNS0 extension
1175 for DNS. This extension is generally useful to reduce DNS traffic,
1176 but some (broken) firewalls drop such DNS packets, which is why it
1177 is off by default.
1178
1179 Setting this variable to 1 will cause AnyEvent::DNS to announce
1180 EDNS0 in its DNS requests.
1181
1182 "PERL_ANYEVENT_MAX_FORKS"
1183 The maximum number of child processes that
1184 "AnyEvent::Util::fork_call" will create in parallel.
1185
1186 "PERL_ANYEVENT_MAX_OUTSTANDING_DNS"
1187 The default value for the "max_outstanding" parameter for the
1188 default DNS resolver - this is the maximum number of parallel DNS
1189 requests that are sent to the DNS server.
1190
1191 "PERL_ANYEVENT_RESOLV_CONF"
1192 The file to use instead of /etc/resolv.conf (or OS-specific
1193 configuration) in the default resolver. When set to the empty
1194 string, no default config will be used.
1195
1196 "PERL_ANYEVENT_CA_FILE", "PERL_ANYEVENT_CA_PATH".
1197 When neither "ca_file" nor "ca_path" was specified during
1198 AnyEvent::TLS context creation, and either of these environment
1199 variables exist, they will be used to specify CA certificate
1200 locations instead of a system-dependent default.
1201
1202 "PERL_ANYEVENT_AVOID_GUARD" and "PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT"
1203 When these are set to 1, then the respective modules are not loaded.
1204 Mostly good for testing AnyEvent itself.
389 1205
390SUPPLYING YOUR OWN EVENT MODEL INTERFACE 1206SUPPLYING YOUR OWN EVENT MODEL INTERFACE
391 This is an advanced topic that you do not normally need to use AnyEvent 1207 This is an advanced topic that you do not normally need to use AnyEvent
392 in a module. This section is only of use to event loop authors who want 1208 in a module. This section is only of use to event loop authors who want
393 to provide AnyEvent compatibility. 1209 to provide AnyEvent compatibility.
… …
428 *rxvt-unicode* also cheats a bit by not providing blocking access to 1244 *rxvt-unicode* also cheats a bit by not providing blocking access to
429 condition variables: code blocking while waiting for a condition will 1245 condition variables: code blocking while waiting for a condition will
430 "die". This still works with most modules/usages, and blocking calls 1246 "die". This still works with most modules/usages, and blocking calls
431 must not be done in an interactive application, so it makes sense. 1247 must not be done in an interactive application, so it makes sense.
432 1248
433ENVIRONMENT VARIABLES
434 The following environment variables are used by this module:
435
436 "PERL_ANYEVENT_VERBOSE" when set to 2 or higher, cause AnyEvent to
437 report to STDERR which event model it chooses.
438
439EXAMPLE PROGRAM 1249EXAMPLE PROGRAM
440 The following program uses an IO watcher to read data from STDIN, a 1250 The following program uses an I/O watcher to read data from STDIN, a
441 timer to display a message once per second, and a condition variable to 1251 timer to display a message once per second, and a condition variable to
442 quit the program when the user enters quit: 1252 quit the program when the user enters quit:
443 1253
444 use AnyEvent; 1254 use AnyEvent;
445 1255
… …
450 poll => 'r', 1260 poll => 'r',
451 cb => sub { 1261 cb => sub {
452 warn "io event <$_[0]>\n"; # will always output <r> 1262 warn "io event <$_[0]>\n"; # will always output <r>
453 chomp (my $input = <STDIN>); # read a line 1263 chomp (my $input = <STDIN>); # read a line
454 warn "read: $input\n"; # output what has been read 1264 warn "read: $input\n"; # output what has been read
455 $cv->broadcast if $input =~ /^q/i; # quit program if /^q/i 1265 $cv->send if $input =~ /^q/i; # quit program if /^q/i
456 }, 1266 },
457 ); 1267 );
458 1268
459 my $time_watcher; # can only be used once
460
461 sub new_timer {
462 $timer = AnyEvent->timer (after => 1, cb => sub { 1269 my $time_watcher = AnyEvent->timer (after => 1, interval => 1, cb => sub {
463 warn "timeout\n"; # print 'timeout' about every second 1270 warn "timeout\n"; # print 'timeout' at most every second
464 &new_timer; # and restart the time
465 });
466 } 1271 });
467 1272
468 new_timer; # create first timer
469
470 $cv->wait; # wait until user enters /^q/i 1273 $cv->recv; # wait until user enters /^q/i
471 1274
472REAL-WORLD EXAMPLE 1275REAL-WORLD EXAMPLE
473 Consider the Net::FCP module. It features (among others) the following 1276 Consider the Net::FCP module. It features (among others) the following
474 API calls, which are to freenet what HTTP GET requests are to http: 1277 API calls, which are to freenet what HTTP GET requests are to http:
475 1278
… …
524 syswrite $txn->{fh}, $txn->{request} 1327 syswrite $txn->{fh}, $txn->{request}
525 or die "connection or write error"; 1328 or die "connection or write error";
526 $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r }); 1329 $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
527 1330
528 Again, "fh_ready_r" waits till all data has arrived, and then stores the 1331 Again, "fh_ready_r" waits till all data has arrived, and then stores the
529 result and signals any possible waiters that the request ahs finished: 1332 result and signals any possible waiters that the request has finished:
530 1333
531 sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf}; 1334 sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
532 1335
533 if (end-of-file or data complete) { 1336 if (end-of-file or data complete) {
534 $txn->{result} = $txn->{buf}; 1337 $txn->{result} = $txn->{buf};
535 $txn->{finished}->broadcast; 1338 $txn->{finished}->send;
536 $txb->{cb}->($txn) of $txn->{cb}; # also call callback 1339 $txb->{cb}->($txn) of $txn->{cb}; # also call callback
537 } 1340 }
538 1341
539 The "result" method, finally, just waits for the finished signal (if the 1342 The "result" method, finally, just waits for the finished signal (if the
540 request was already finished, it doesn't wait, of course, and returns 1343 request was already finished, it doesn't wait, of course, and returns
541 the data: 1344 the data:
542 1345
543 $txn->{finished}->wait; 1346 $txn->{finished}->recv;
544 return $txn->{result}; 1347 return $txn->{result};
545 1348
546 The actual code goes further and collects all errors ("die"s, 1349 The actual code goes further and collects all errors ("die"s,
547 exceptions) that occured during request processing. The "result" method 1350 exceptions) that occurred during request processing. The "result" method
548 detects whether an exception as thrown (it is stored inside the $txn 1351 detects whether an exception as thrown (it is stored inside the $txn
549 object) and just throws the exception, which means connection errors and 1352 object) and just throws the exception, which means connection errors and
550 other problems get reported tot he code that tries to use the result, 1353 other problems get reported tot he code that tries to use the result,
551 not in a random callback. 1354 not in a random callback.
552 1355
… …
583 1386
584 my $quit = AnyEvent->condvar; 1387 my $quit = AnyEvent->condvar;
585 1388
586 $fcp->txn_client_get ($url)->cb (sub { 1389 $fcp->txn_client_get ($url)->cb (sub {
587 ... 1390 ...
588 $quit->broadcast; 1391 $quit->send;
589 }); 1392 });
590 1393
591 $quit->wait; 1394 $quit->recv;
1395
1396BENCHMARKS
1397 To give you an idea of the performance and overheads that AnyEvent adds
1398 over the event loops themselves and to give you an impression of the
1399 speed of various event loops I prepared some benchmarks.
1400
1401 BENCHMARKING ANYEVENT OVERHEAD
1402 Here is a benchmark of various supported event models used natively and
1403 through AnyEvent. The benchmark creates a lot of timers (with a zero
1404 timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
1405 which it is), lets them fire exactly once and destroys them again.
1406
1407 Source code for this benchmark is found as eg/bench in the AnyEvent
1408 distribution. It uses the AE interface, which makes a real difference
1409 for the EV and Perl backends only.
1410
1411 Explanation of the columns
1412 *watcher* is the number of event watchers created/destroyed. Since
1413 different event models feature vastly different performances, each event
1414 loop was given a number of watchers so that overall runtime is
1415 acceptable and similar between tested event loop (and keep them from
1416 crashing): Glib would probably take thousands of years if asked to
1417 process the same number of watchers as EV in this benchmark.
1418
1419 *bytes* is the number of bytes (as measured by the resident set size,
1420 RSS) consumed by each watcher. This method of measuring captures both C
1421 and Perl-based overheads.
1422
1423 *create* is the time, in microseconds (millionths of seconds), that it
1424 takes to create a single watcher. The callback is a closure shared
1425 between all watchers, to avoid adding memory overhead. That means
1426 closure creation and memory usage is not included in the figures.
1427
1428 *invoke* is the time, in microseconds, used to invoke a simple callback.
1429 The callback simply counts down a Perl variable and after it was invoked
1430 "watcher" times, it would "->send" a condvar once to signal the end of
1431 this phase.
1432
1433 *destroy* is the time, in microseconds, that it takes to destroy a
1434 single watcher.
1435
1436 Results
1437 name watchers bytes create invoke destroy comment
1438 EV/EV 100000 223 0.47 0.43 0.27 EV native interface
1439 EV/Any 100000 223 0.48 0.42 0.26 EV + AnyEvent watchers
1440 Coro::EV/Any 100000 223 0.47 0.42 0.26 coroutines + Coro::Signal
1441 Perl/Any 100000 431 2.70 0.74 0.92 pure perl implementation
1442 Event/Event 16000 516 31.16 31.84 0.82 Event native interface
1443 Event/Any 16000 1203 42.61 34.79 1.80 Event + AnyEvent watchers
1444 IOAsync/Any 16000 1911 41.92 27.45 16.81 via IO::Async::Loop::IO_Poll
1445 IOAsync/Any 16000 1726 40.69 26.37 15.25 via IO::Async::Loop::Epoll
1446 Glib/Any 16000 1118 89.00 12.57 51.17 quadratic behaviour
1447 Tk/Any 2000 1346 20.96 10.75 8.00 SEGV with >> 2000 watchers
1448 POE/Any 2000 6951 108.97 795.32 14.24 via POE::Loop::Event
1449 POE/Any 2000 6648 94.79 774.40 575.51 via POE::Loop::Select
1450
1451 Discussion
1452 The benchmark does *not* measure scalability of the event loop very
1453 well. For example, a select-based event loop (such as the pure perl one)
1454 can never compete with an event loop that uses epoll when the number of
1455 file descriptors grows high. In this benchmark, all events become ready
1456 at the same time, so select/poll-based implementations get an unnatural
1457 speed boost.
1458
1459 Also, note that the number of watchers usually has a nonlinear effect on
1460 overall speed, that is, creating twice as many watchers doesn't take
1461 twice the time - usually it takes longer. This puts event loops tested
1462 with a higher number of watchers at a disadvantage.
1463
1464 To put the range of results into perspective, consider that on the
1465 benchmark machine, handling an event takes roughly 1600 CPU cycles with
1466 EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000
1467 CPU cycles with POE.
1468
1469 "EV" is the sole leader regarding speed and memory use, which are both
1470 maximal/minimal, respectively. When using the AE API there is zero
1471 overhead (when going through the AnyEvent API create is about 5-6 times
1472 slower, with other times being equal, so still uses far less memory than
1473 any other event loop and is still faster than Event natively).
1474
1475 The pure perl implementation is hit in a few sweet spots (both the
1476 constant timeout and the use of a single fd hit optimisations in the
1477 perl interpreter and the backend itself). Nevertheless this shows that
1478 it adds very little overhead in itself. Like any select-based backend
1479 its performance becomes really bad with lots of file descriptors (and
1480 few of them active), of course, but this was not subject of this
1481 benchmark.
1482
1483 The "Event" module has a relatively high setup and callback invocation
1484 cost, but overall scores in on the third place.
1485
1486 "IO::Async" performs admirably well, about on par with "Event", even
1487 when using its pure perl backend.
1488
1489 "Glib"'s memory usage is quite a bit higher, but it features a faster
1490 callback invocation and overall ends up in the same class as "Event".
1491 However, Glib scales extremely badly, doubling the number of watchers
1492 increases the processing time by more than a factor of four, making it
1493 completely unusable when using larger numbers of watchers (note that
1494 only a single file descriptor was used in the benchmark, so
1495 inefficiencies of "poll" do not account for this).
1496
1497 The "Tk" adaptor works relatively well. The fact that it crashes with
1498 more than 2000 watchers is a big setback, however, as correctness takes
1499 precedence over speed. Nevertheless, its performance is surprising, as
1500 the file descriptor is dup()ed for each watcher. This shows that the
1501 dup() employed by some adaptors is not a big performance issue (it does
1502 incur a hidden memory cost inside the kernel which is not reflected in
1503 the figures above).
1504
1505 "POE", regardless of underlying event loop (whether using its pure perl
1506 select-based backend or the Event module, the POE-EV backend couldn't be
1507 tested because it wasn't working) shows abysmal performance and memory
1508 usage with AnyEvent: Watchers use almost 30 times as much memory as EV
1509 watchers, and 10 times as much memory as Event (the high memory
1510 requirements are caused by requiring a session for each watcher).
1511 Watcher invocation speed is almost 900 times slower than with AnyEvent's
1512 pure perl implementation.
1513
1514 The design of the POE adaptor class in AnyEvent can not really account
1515 for the performance issues, though, as session creation overhead is
1516 small compared to execution of the state machine, which is coded pretty
1517 optimally within AnyEvent::Impl::POE (and while everybody agrees that
1518 using multiple sessions is not a good approach, especially regarding
1519 memory usage, even the author of POE could not come up with a faster
1520 design).
1521
1522 Summary
1523 * Using EV through AnyEvent is faster than any other event loop (even
1524 when used without AnyEvent), but most event loops have acceptable
1525 performance with or without AnyEvent.
1526
1527 * The overhead AnyEvent adds is usually much smaller than the overhead
1528 of the actual event loop, only with extremely fast event loops such
1529 as EV adds AnyEvent significant overhead.
1530
1531 * You should avoid POE like the plague if you want performance or
1532 reasonable memory usage.
1533
1534 BENCHMARKING THE LARGE SERVER CASE
1535 This benchmark actually benchmarks the event loop itself. It works by
1536 creating a number of "servers": each server consists of a socket pair, a
1537 timeout watcher that gets reset on activity (but never fires), and an
1538 I/O watcher waiting for input on one side of the socket. Each time the
1539 socket watcher reads a byte it will write that byte to a random other
1540 "server".
1541
1542 The effect is that there will be a lot of I/O watchers, only part of
1543 which are active at any one point (so there is a constant number of
1544 active fds for each loop iteration, but which fds these are is random).
1545 The timeout is reset each time something is read because that reflects
1546 how most timeouts work (and puts extra pressure on the event loops).
1547
1548 In this benchmark, we use 10000 socket pairs (20000 sockets), of which
1549 100 (1%) are active. This mirrors the activity of large servers with
1550 many connections, most of which are idle at any one point in time.
1551
1552 Source code for this benchmark is found as eg/bench2 in the AnyEvent
1553 distribution. It uses the AE interface, which makes a real difference
1554 for the EV and Perl backends only.
1555
1556 Explanation of the columns
1557 *sockets* is the number of sockets, and twice the number of "servers"
1558 (as each server has a read and write socket end).
1559
1560 *create* is the time it takes to create a socket pair (which is
1561 nontrivial) and two watchers: an I/O watcher and a timeout watcher.
1562
1563 *request*, the most important value, is the time it takes to handle a
1564 single "request", that is, reading the token from the pipe and
1565 forwarding it to another server. This includes deleting the old timeout
1566 and creating a new one that moves the timeout into the future.
1567
1568 Results
1569 name sockets create request
1570 EV 20000 62.66 7.99
1571 Perl 20000 68.32 32.64
1572 IOAsync 20000 174.06 101.15 epoll
1573 IOAsync 20000 174.67 610.84 poll
1574 Event 20000 202.69 242.91
1575 Glib 20000 557.01 1689.52
1576 POE 20000 341.54 12086.32 uses POE::Loop::Event
1577
1578 Discussion
1579 This benchmark *does* measure scalability and overall performance of the
1580 particular event loop.
1581
1582 EV is again fastest. Since it is using epoll on my system, the setup
1583 time is relatively high, though.
1584
1585 Perl surprisingly comes second. It is much faster than the C-based event
1586 loops Event and Glib.
1587
1588 IO::Async performs very well when using its epoll backend, and still
1589 quite good compared to Glib when using its pure perl backend.
1590
1591 Event suffers from high setup time as well (look at its code and you
1592 will understand why). Callback invocation also has a high overhead
1593 compared to the "$_->() for .."-style loop that the Perl event loop
1594 uses. Event uses select or poll in basically all documented
1595 configurations.
1596
1597 Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It
1598 clearly fails to perform with many filehandles or in busy servers.
1599
1600 POE is still completely out of the picture, taking over 1000 times as
1601 long as EV, and over 100 times as long as the Perl implementation, even
1602 though it uses a C-based event loop in this case.
1603
1604 Summary
1605 * The pure perl implementation performs extremely well.
1606
1607 * Avoid Glib or POE in large projects where performance matters.
1608
1609 BENCHMARKING SMALL SERVERS
1610 While event loops should scale (and select-based ones do not...) even to
1611 large servers, most programs we (or I :) actually write have only a few
1612 I/O watchers.
1613
1614 In this benchmark, I use the same benchmark program as in the large
1615 server case, but it uses only eight "servers", of which three are active
1616 at any one time. This should reflect performance for a small server
1617 relatively well.
1618
1619 The columns are identical to the previous table.
1620
1621 Results
1622 name sockets create request
1623 EV 16 20.00 6.54
1624 Perl 16 25.75 12.62
1625 Event 16 81.27 35.86
1626 Glib 16 32.63 15.48
1627 POE 16 261.87 276.28 uses POE::Loop::Event
1628
1629 Discussion
1630 The benchmark tries to test the performance of a typical small server.
1631 While knowing how various event loops perform is interesting, keep in
1632 mind that their overhead in this case is usually not as important, due
1633 to the small absolute number of watchers (that is, you need efficiency
1634 and speed most when you have lots of watchers, not when you only have a
1635 few of them).
1636
1637 EV is again fastest.
1638
1639 Perl again comes second. It is noticeably faster than the C-based event
1640 loops Event and Glib, although the difference is too small to really
1641 matter.
1642
1643 POE also performs much better in this case, but is is still far behind
1644 the others.
1645
1646 Summary
1647 * C-based event loops perform very well with small number of watchers,
1648 as the management overhead dominates.
1649
1650 THE IO::Lambda BENCHMARK
1651 Recently I was told about the benchmark in the IO::Lambda manpage, which
1652 could be misinterpreted to make AnyEvent look bad. In fact, the
1653 benchmark simply compares IO::Lambda with POE, and IO::Lambda looks
1654 better (which shouldn't come as a surprise to anybody). As such, the
1655 benchmark is fine, and mostly shows that the AnyEvent backend from
1656 IO::Lambda isn't very optimal. But how would AnyEvent compare when used
1657 without the extra baggage? To explore this, I wrote the equivalent
1658 benchmark for AnyEvent.
1659
1660 The benchmark itself creates an echo-server, and then, for 500 times,
1661 connects to the echo server, sends a line, waits for the reply, and then
1662 creates the next connection. This is a rather bad benchmark, as it
1663 doesn't test the efficiency of the framework or much non-blocking I/O,
1664 but it is a benchmark nevertheless.
1665
1666 name runtime
1667 Lambda/select 0.330 sec
1668 + optimized 0.122 sec
1669 Lambda/AnyEvent 0.327 sec
1670 + optimized 0.138 sec
1671 Raw sockets/select 0.077 sec
1672 POE/select, components 0.662 sec
1673 POE/select, raw sockets 0.226 sec
1674 POE/select, optimized 0.404 sec
1675
1676 AnyEvent/select/nb 0.085 sec
1677 AnyEvent/EV/nb 0.068 sec
1678 +state machine 0.134 sec
1679
1680 The benchmark is also a bit unfair (my fault): the IO::Lambda/POE
1681 benchmarks actually make blocking connects and use 100% blocking I/O,
1682 defeating the purpose of an event-based solution. All of the newly
1683 written AnyEvent benchmarks use 100% non-blocking connects (using
1684 AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS
1685 resolver), so AnyEvent is at a disadvantage here, as non-blocking
1686 connects generally require a lot more bookkeeping and event handling
1687 than blocking connects (which involve a single syscall only).
1688
1689 The last AnyEvent benchmark additionally uses AnyEvent::Handle, which
1690 offers similar expressive power as POE and IO::Lambda, using
1691 conventional Perl syntax. This means that both the echo server and the
1692 client are 100% non-blocking, further placing it at a disadvantage.
1693
1694 As you can see, the AnyEvent + EV combination even beats the
1695 hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl
1696 backend easily beats IO::Lambda and POE.
1697
1698 And even the 100% non-blocking version written using the high-level (and
1699 slow :) AnyEvent::Handle abstraction beats both POE and IO::Lambda
1700 higher level ("unoptimised") abstractions by a large margin, even though
1701 it does all of DNS, tcp-connect and socket I/O in a non-blocking way.
1702
1703 The two AnyEvent benchmarks programs can be found as eg/ae0.pl and
1704 eg/ae2.pl in the AnyEvent distribution, the remaining benchmarks are
1705 part of the IO::Lambda distribution and were used without any changes.
1706
1707SIGNALS
1708 AnyEvent currently installs handlers for these signals:
1709
1710 SIGCHLD
1711 A handler for "SIGCHLD" is installed by AnyEvent's child watcher
1712 emulation for event loops that do not support them natively. Also,
1713 some event loops install a similar handler.
1714
1715 Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE,
1716 then AnyEvent will reset it to default, to avoid losing child exit
1717 statuses.
1718
1719 SIGPIPE
1720 A no-op handler is installed for "SIGPIPE" when $SIG{PIPE} is
1721 "undef" when AnyEvent gets loaded.
1722
1723 The rationale for this is that AnyEvent users usually do not really
1724 depend on SIGPIPE delivery (which is purely an optimisation for
1725 shell use, or badly-written programs), but "SIGPIPE" can cause
1726 spurious and rare program exits as a lot of people do not expect
1727 "SIGPIPE" when writing to some random socket.
1728
1729 The rationale for installing a no-op handler as opposed to ignoring
1730 it is that this way, the handler will be restored to defaults on
1731 exec.
1732
1733 Feel free to install your own handler, or reset it to defaults.
1734
1735RECOMMENDED/OPTIONAL MODULES
1736 One of AnyEvent's main goals is to be 100% Pure-Perl(tm): only perl (and
1737 it's built-in modules) are required to use it.
1738
1739 That does not mean that AnyEvent won't take advantage of some additional
1740 modules if they are installed.
1741
1742 This section explains which additional modules will be used, and how
1743 they affect AnyEvent's operation.
1744
1745 Async::Interrupt
1746 This slightly arcane module is used to implement fast signal
1747 handling: To my knowledge, there is no way to do completely
1748 race-free and quick signal handling in pure perl. To ensure that
1749 signals still get delivered, AnyEvent will start an interval timer
1750 to wake up perl (and catch the signals) with some delay (default is
1751 10 seconds, look for $AnyEvent::MAX_SIGNAL_LATENCY).
1752
1753 If this module is available, then it will be used to implement
1754 signal catching, which means that signals will not be delayed, and
1755 the event loop will not be interrupted regularly, which is more
1756 efficient (and good for battery life on laptops).
1757
1758 This affects not just the pure-perl event loop, but also other event
1759 loops that have no signal handling on their own (e.g. Glib, Tk, Qt).
1760
1761 Some event loops (POE, Event, Event::Lib) offer signal watchers
1762 natively, and either employ their own workarounds (POE) or use
1763 AnyEvent's workaround (using $AnyEvent::MAX_SIGNAL_LATENCY).
1764 Installing Async::Interrupt does nothing for those backends.
1765
1766 EV This module isn't really "optional", as it is simply one of the
1767 backend event loops that AnyEvent can use. However, it is simply the
1768 best event loop available in terms of features, speed and stability:
1769 It supports the AnyEvent API optimally, implements all the watcher
1770 types in XS, does automatic timer adjustments even when no monotonic
1771 clock is available, can take avdantage of advanced kernel interfaces
1772 such as "epoll" and "kqueue", and is the fastest backend *by far*.
1773 You can even embed Glib/Gtk2 in it (or vice versa, see EV::Glib and
1774 Glib::EV).
1775
1776 Guard
1777 The guard module, when used, will be used to implement
1778 "AnyEvent::Util::guard". This speeds up guards considerably (and
1779 uses a lot less memory), but otherwise doesn't affect guard
1780 operation much. It is purely used for performance.
1781
1782 JSON and JSON::XS
1783 One of these modules is required when you want to read or write JSON
1784 data via AnyEvent::Handle. It is also written in pure-perl, but can
1785 take advantage of the ultra-high-speed JSON::XS module when it is
1786 installed.
1787
1788 In fact, AnyEvent::Handle will use JSON::XS by default if it is
1789 installed.
1790
1791 Net::SSLeay
1792 Implementing TLS/SSL in Perl is certainly interesting, but not very
1793 worthwhile: If this module is installed, then AnyEvent::Handle (with
1794 the help of AnyEvent::TLS), gains the ability to do TLS/SSL.
1795
1796 Time::HiRes
1797 This module is part of perl since release 5.008. It will be used
1798 when the chosen event library does not come with a timing source on
1799 it's own. The pure-perl event loop (AnyEvent::Impl::Perl) will
1800 additionally use it to try to use a monotonic clock for timing
1801 stability.
1802
1803FORK
1804 Most event libraries are not fork-safe. The ones who are usually are
1805 because they rely on inefficient but fork-safe "select" or "poll" calls.
1806 Only EV is fully fork-aware.
1807
1808 This means that, in general, you cannot fork and do event processing in
1809 the child if a watcher was created before the fork (which in turn
1810 initialises the event library).
1811
1812 If you have to fork, you must either do so *before* creating your first
1813 watcher OR you must not use AnyEvent at all in the child OR you must do
1814 something completely out of the scope of AnyEvent.
1815
1816 The problem of doing event processing in the parent *and* the child is
1817 much more complicated: even for backends that *are* fork-aware or
1818 fork-safe, their behaviour is not usually what you want: fork clones all
1819 watchers, that means all timers, I/O watchers etc. are active in both
1820 parent and child, which is almost never what you want.
1821
1822SECURITY CONSIDERATIONS
1823 AnyEvent can be forced to load any event model via
1824 $ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used
1825 to execute arbitrary code or directly gain access, it can easily be used
1826 to make the program hang or malfunction in subtle ways, as AnyEvent
1827 watchers will not be active when the program uses a different event
1828 model than specified in the variable.
1829
1830 You can make AnyEvent completely ignore this variable by deleting it
1831 before the first watcher gets created, e.g. with a "BEGIN" block:
1832
1833 BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
1834
1835 use AnyEvent;
1836
1837 Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
1838 be used to probe what backend is used and gain other information (which
1839 is probably even less useful to an attacker than PERL_ANYEVENT_MODEL),
1840 and $ENV{PERL_ANYEVENT_STRICT}.
1841
1842 Note that AnyEvent will remove *all* environment variables starting with
1843 "PERL_ANYEVENT_" from %ENV when it is loaded while taint mode is
1844 enabled.
1845
1846BUGS
1847 Perl 5.8 has numerous memleaks that sometimes hit this module and are
1848 hard to work around. If you suffer from memleaks, first upgrade to Perl
1849 5.10 and check wether the leaks still show up. (Perl 5.10.0 has other
1850 annoying memleaks, such as leaking on "map" and "grep" but it is usually
1851 not as pronounced).
592 1852
593SEE ALSO 1853SEE ALSO
594 Event modules: Coro::EV, EV, EV::Glib, Glib::EV, Coro::Event, Event, 1854 Utility functions: AnyEvent::Util.
595 Glib::Event, Glib, Coro, Tk.
596 1855
597 Implementations: AnyEvent::Impl::CoroEV, AnyEvent::Impl::EV, 1856 Event modules: EV, EV::Glib, Glib::EV, Event, Glib::Event, Glib, Tk,
598 AnyEvent::Impl::CoroEvent, AnyEvent::Impl::Event, AnyEvent::Impl::Glib, 1857 Event::Lib, Qt, POE.
1858
1859 Implementations: AnyEvent::Impl::EV, AnyEvent::Impl::Event,
1860 AnyEvent::Impl::Glib, AnyEvent::Impl::Tk, AnyEvent::Impl::Perl,
1861 AnyEvent::Impl::EventLib, AnyEvent::Impl::Qt, AnyEvent::Impl::POE,
599 AnyEvent::Impl::Tk, AnyEvent::Impl::Perl. 1862 AnyEvent::Impl::IOAsync, Anyevent::Impl::Irssi.
600 1863
1864 Non-blocking file handles, sockets, TCP clients and servers:
1865 AnyEvent::Handle, AnyEvent::Socket, AnyEvent::TLS.
1866
1867 Asynchronous DNS: AnyEvent::DNS.
1868
1869 Coroutine support: Coro, Coro::AnyEvent, Coro::EV, Coro::Event,
1870
601 Nontrivial usage examples: Net::FCP, Net::XMPP2. 1871 Nontrivial usage examples: AnyEvent::GPSD, AnyEvent::XMPP,
1872 AnyEvent::HTTP.
602 1873
603AUTHOR 1874AUTHOR
604 Marc Lehmann <schmorp@schmorp.de> 1875 Marc Lehmann <schmorp@schmorp.de>
605 http://home.schmorp.de/ 1876 http://home.schmorp.de/
606 1877

Diff Legend

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