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

Comparing AnyEvent/README (file contents):
Revision 1.19 by root, Mon Apr 28 08:02:14 2008 UTC vs.
Revision 1.47 by root, Mon Jul 20 22:39:57 2009 UTC

1NAME 1NAME
2 AnyEvent - provide framework for multiple event loops 2 AnyEvent - events independent of event loop implementation
3 3
4 EV, Event, Coro::EV, Coro::Event, Glib, Tk, Perl, Event::Lib, Qt, POE - 4 EV, Event, Glib, Tk, Perl, Event::Lib, Qt and POE are various supported
5 various supported event loops 5 event loops.
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 Respository, 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, AnyEvent::Impl::Perl, Tk, Event::Lib, Qt, POE. The first one found 116 AnyEvent::Impl::Perl, Tk, Event::Lib, Qt, POE. The first one found is
82 is used. If none are found, the module tries to load these modules 117 used. If none are found, the module tries to load these modules
83 (excluding Tk, Event::Lib, Qt and POE as the pure perl adaptor should 118 (excluding Tk, Event::Lib, Qt and POE as the pure perl adaptor should
84 always succeed) in the order given. The first one that can be 119 always succeed) in the order given. The first one that can be
85 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
86 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
87 very efficient, but should work everywhere. 122 very efficient, but should work everywhere.
99 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
100 to use AnyEvent so their modules work together with others seamlessly... 135 to use AnyEvent so their modules work together with others seamlessly...
101 136
102 The pure-perl implementation of AnyEvent is called 137 The pure-perl implementation of AnyEvent is called
103 "AnyEvent::Impl::Perl". Like other event modules you can load it 138 "AnyEvent::Impl::Perl". Like other event modules you can load it
104 explicitly. 139 explicitly and enjoy the high availability of that event loop :)
105 140
106WATCHERS 141WATCHERS
107 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
108 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
109 the callback to call, the filehandle to watch, etc. 144 the callback to call, the file handle to watch, etc.
110 145
111 These watchers are normal Perl objects with normal Perl lifetime. After 146 These watchers are normal Perl objects with normal Perl lifetime. After
112 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
113 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
114 in control). 149 in control).
115 150
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
116 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
117 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
118 to it). 159 to it).
119 160
120 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.
122 Many watchers either are used with "recursion" (repeating timers for 163 Many watchers either are used with "recursion" (repeating timers for
123 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.
124 165
125 An any way to achieve that is this pattern: 166 An any way to achieve that is this pattern:
126 167
127 my $w; $w = AnyEvent->type (arg => value ..., cb => sub { 168 my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
128 # you can use $w here, for example to undef it 169 # you can use $w here, for example to undef it
129 undef $w; 170 undef $w;
130 }); 171 });
131 172
132 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,
133 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
134 declared. 175 declared.
135 176
136 I/O WATCHERS 177 I/O WATCHERS
137 You can create an I/O watcher by calling the "AnyEvent->io" method with 178 You can create an I/O watcher by calling the "AnyEvent->io" method with
138 the following mandatory key-value pairs as arguments: 179 the following mandatory key-value pairs as arguments:
139 180
140 "fh" the Perl *file handle* (*not* file descriptor) to watch for events. 181 "fh" is the Perl *file handle* (or a naked file descriptor) to watch for
182 events (AnyEvent might or might not keep a reference to this file
183 handle). Note that only file handles pointing to things for which
184 non-blocking operation makes sense are allowed. This includes sockets,
185 most character devices, pipes, fifos and so on, but not for example
186 files or block devices.
187
141 "poll" must be a string that is either "r" or "w", which creates a 188 "poll" must be a string that is either "r" or "w", which creates a
142 watcher waiting for "r"eadable or "w"ritable events, respectively. "cb" 189 watcher waiting for "r"eadable or "w"ritable events, respectively.
190
143 is the callback to invoke each time the file handle becomes ready. 191 "cb" is the callback to invoke each time the file handle becomes ready.
144 192
145 Although the callback might get passed parameters, their value and 193 Although the callback might get passed parameters, their value and
146 presence is undefined and you cannot rely on them. Portable AnyEvent 194 presence is undefined and you cannot rely on them. Portable AnyEvent
147 callbacks cannot use arguments passed to I/O watcher callbacks. 195 callbacks cannot use arguments passed to I/O watcher callbacks.
148 196
152 200
153 Some event loops issue spurious readyness notifications, so you should 201 Some event loops issue spurious readyness notifications, so you should
154 always use non-blocking calls when reading/writing from/to your file 202 always use non-blocking calls when reading/writing from/to your file
155 handles. 203 handles.
156 204
157 Example:
158
159 # wait for readability of STDIN, then read a line and disable the watcher 205 Example: wait for readability of STDIN, then read a line and disable the
206 watcher.
207
160 my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub { 208 my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
161 chomp (my $input = <STDIN>); 209 chomp (my $input = <STDIN>);
162 warn "read: $input\n"; 210 warn "read: $input\n";
163 undef $w; 211 undef $w;
164 }); 212 });
173 221
174 Although the callback might get passed parameters, their value and 222 Although the callback might get passed parameters, their value and
175 presence is undefined and you cannot rely on them. Portable AnyEvent 223 presence is undefined and you cannot rely on them. Portable AnyEvent
176 callbacks cannot use arguments passed to time watcher callbacks. 224 callbacks cannot use arguments passed to time watcher callbacks.
177 225
178 The timer callback will be invoked at most once: if you want a repeating 226 The callback will normally be invoked once only. If you specify another
179 timer you have to create a new watcher (this is a limitation by both Tk 227 parameter, "interval", as a strictly positive number (> 0), then the
180 and Glib). 228 callback will be invoked regularly at that interval (in fractional
229 seconds) after the first invocation. If "interval" is specified with a
230 false value, then it is treated as if it were missing.
181 231
182 Example: 232 The callback will be rescheduled before invoking the callback, but no
233 attempt is done to avoid timer drift in most backends, so the interval
234 is only approximate.
183 235
184 # fire an event after 7.7 seconds 236 Example: fire an event after 7.7 seconds.
237
185 my $w = AnyEvent->timer (after => 7.7, cb => sub { 238 my $w = AnyEvent->timer (after => 7.7, cb => sub {
186 warn "timeout\n"; 239 warn "timeout\n";
187 }); 240 });
188 241
189 # to cancel the timer: 242 # to cancel the timer:
190 undef $w; 243 undef $w;
191 244
192 Example 2:
193
194 # fire an event after 0.5 seconds, then roughly every second 245 Example 2: fire an event after 0.5 seconds, then roughly every second.
195 my $w;
196 246
197 my $cb = sub {
198 # cancel the old timer while creating a new one
199 $w = AnyEvent->timer (after => 1, cb => $cb); 247 my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
248 warn "timeout\n";
200 }; 249 };
201
202 # start the "loop" by creating the first watcher
203 $w = AnyEvent->timer (after => 0.5, cb => $cb);
204 250
205 TIMING ISSUES 251 TIMING ISSUES
206 There are two ways to handle timers: based on real time (relative, "fire 252 There are two ways to handle timers: based on real time (relative, "fire
207 in 10 seconds") and based on wallclock time (absolute, "fire at 12 253 in 10 seconds") and based on wallclock time (absolute, "fire at 12
208 o'clock"). 254 o'clock").
220 on wallclock time) timers. 266 on wallclock time) timers.
221 267
222 AnyEvent always prefers relative timers, if available, matching the 268 AnyEvent always prefers relative timers, if available, matching the
223 AnyEvent API. 269 AnyEvent API.
224 270
271 AnyEvent has two additional methods that return the "current time":
272
273 AnyEvent->time
274 This returns the "current wallclock time" as a fractional number of
275 seconds since the Epoch (the same thing as "time" or
276 "Time::HiRes::time" return, and the result is guaranteed to be
277 compatible with those).
278
279 It progresses independently of any event loop processing, i.e. each
280 call will check the system clock, which usually gets updated
281 frequently.
282
283 AnyEvent->now
284 This also returns the "current wallclock time", but unlike "time",
285 above, this value might change only once per event loop iteration,
286 depending on the event loop (most return the same time as "time",
287 above). This is the time that AnyEvent's timers get scheduled
288 against.
289
290 *In almost all cases (in all cases if you don't care), this is the
291 function to call when you want to know the current time.*
292
293 This function is also often faster then "AnyEvent->time", and thus
294 the preferred method if you want some timestamp (for example,
295 AnyEvent::Handle uses this to update it's activity timeouts).
296
297 The rest of this section is only of relevance if you try to be very
298 exact with your timing, you can skip it without bad conscience.
299
300 For a practical example of when these times differ, consider
301 Event::Lib and EV and the following set-up:
302
303 The event loop is running and has just invoked one of your callback
304 at time=500 (assume no other callbacks delay processing). In your
305 callback, you wait a second by executing "sleep 1" (blocking the
306 process for a second) and then (at time=501) you create a relative
307 timer that fires after three seconds.
308
309 With Event::Lib, "AnyEvent->time" and "AnyEvent->now" will both
310 return 501, because that is the current time, and the timer will be
311 scheduled to fire at time=504 (501 + 3).
312
313 With EV, "AnyEvent->time" returns 501 (as that is the current time),
314 but "AnyEvent->now" returns 500, as that is the time the last event
315 processing phase started. With EV, your timer gets scheduled to run
316 at time=503 (500 + 3).
317
318 In one sense, Event::Lib is more exact, as it uses the current time
319 regardless of any delays introduced by event processing. However,
320 most callbacks do not expect large delays in processing, so this
321 causes a higher drift (and a lot more system calls to get the
322 current time).
323
324 In another sense, EV is more exact, as your timer will be scheduled
325 at the same time, regardless of how long event processing actually
326 took.
327
328 In either case, if you care (and in most cases, you don't), then you
329 can get whatever behaviour you want with any event loop, by taking
330 the difference between "AnyEvent->time" and "AnyEvent->now" into
331 account.
332
333 AnyEvent->now_update
334 Some event loops (such as EV or AnyEvent::Impl::Perl) cache the
335 current time for each loop iteration (see the discussion of
336 AnyEvent->now, above).
337
338 When a callback runs for a long time (or when the process sleeps),
339 then this "current" time will differ substantially from the real
340 time, which might affect timers and time-outs.
341
342 When this is the case, you can call this method, which will update
343 the event loop's idea of "current time".
344
345 Note that updating the time *might* cause some events to be handled.
346
225 SIGNAL WATCHERS 347 SIGNAL WATCHERS
226 You can watch for signals using a signal watcher, "signal" is the signal 348 You can watch for signals using a signal watcher, "signal" is the signal
227 *name* without any "SIG" prefix, "cb" is the Perl callback to be invoked 349 *name* in uppercase and without any "SIG" prefix, "cb" is the Perl
228 whenever a signal occurs. 350 callback to be invoked whenever a signal occurs.
229 351
230 Although the callback might get passed parameters, their value and 352 Although the callback might get passed parameters, their value and
231 presence is undefined and you cannot rely on them. Portable AnyEvent 353 presence is undefined and you cannot rely on them. Portable AnyEvent
232 callbacks cannot use arguments passed to signal watcher callbacks. 354 callbacks cannot use arguments passed to signal watcher callbacks.
233 355
234 Multiple signal occurances can be clumped together into one callback 356 Multiple signal occurrences can be clumped together into one callback
235 invocation, and callback invocation will be synchronous. synchronous 357 invocation, and callback invocation will be synchronous. Synchronous
236 means that it might take a while until the signal gets handled by the 358 means that it might take a while until the signal gets handled by the
237 process, but it is guarenteed not to interrupt any other callbacks. 359 process, but it is guaranteed not to interrupt any other callbacks.
238 360
239 The main advantage of using these watchers is that you can share a 361 The main advantage of using these watchers is that you can share a
240 signal between multiple watchers. 362 signal between multiple watchers, and AnyEvent will ensure that signals
363 will not interrupt your program at bad times.
241 364
242 This watcher might use %SIG, so programs overwriting those signals 365 This watcher might use %SIG (depending on the event loop used), so
243 directly will likely not work correctly. 366 programs overwriting those signals directly will likely not work
367 correctly.
244 368
245 Example: exit on SIGINT 369 Example: exit on SIGINT
246 370
247 my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 }); 371 my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
372
373 Signal Races, Delays and Workarounds
374 Many event loops (e.g. Glib, Tk, Qt, IO::Async) do not support attaching
375 callbacks to signals in a generic way, which is a pity, as you cannot do
376 race-free signal handling in perl. AnyEvent will try to do it's best,
377 but in some cases, signals will be delayed. The maximum time a signal
378 might be delayed is specified in $AnyEvent::MAX_SIGNAL_LATENCY (default:
379 10 seconds). This variable can be changed only before the first signal
380 watcher is created, and should be left alone otherwise. Higher values
381 will cause fewer spurious wake-ups, which is better for power and CPU
382 saving. All these problems can be avoided by installing the optional
383 Async::Interrupt module. This will not work with inherently broken event
384 loops such as Event or Event::Lib (and not with POE currently, as POE
385 does it's own workaround with one-second latency). With those, you just
386 have to suffer the delays.
248 387
249 CHILD PROCESS WATCHERS 388 CHILD PROCESS WATCHERS
250 You can also watch on a child process exit and catch its exit status. 389 You can also watch on a child process exit and catch its exit status.
251 390
252 The child process is specified by the "pid" argument (if set to 0, it 391 The child process is specified by the "pid" argument (if set to 0, it
253 watches for any child process exit). The watcher will trigger as often 392 watches for any child process exit). The watcher will triggered only
254 as status change for the child are received. This works by installing a 393 when the child process has finished and an exit status is available, not
255 signal handler for "SIGCHLD". The callback will be called with the pid 394 on any trace events (stopped/continued).
256 and exit status (as returned by waitpid), so unlike other watcher types, 395
257 you *can* rely on child watcher callback arguments. 396 The callback will be called with the pid and exit status (as returned by
397 waitpid), so unlike other watcher types, you *can* rely on child watcher
398 callback arguments.
399
400 This watcher type works by installing a signal handler for "SIGCHLD",
401 and since it cannot be shared, nothing else should use SIGCHLD or reap
402 random child processes (waiting for specific child processes, e.g.
403 inside "system", is just fine).
258 404
259 There is a slight catch to child watchers, however: you usually start 405 There is a slight catch to child watchers, however: you usually start
260 them *after* the child process was created, and this means the process 406 them *after* the child process was created, and this means the process
261 could have exited already (and no SIGCHLD will be sent anymore). 407 could have exited already (and no SIGCHLD will be sent anymore).
262 408
263 Not all event models handle this correctly (POE doesn't), but even for 409 Not all event models handle this correctly (neither POE nor IO::Async
410 do, see their AnyEvent::Impl manpages for details), but even for event
264 event models that *do* handle this correctly, they usually need to be 411 models that *do* handle this correctly, they usually need to be loaded
265 loaded before the process exits (i.e. before you fork in the first 412 before the process exits (i.e. before you fork in the first place).
266 place). 413 AnyEvent's pure perl event loop handles all cases correctly regardless
414 of when you start the watcher.
267 415
268 This means you cannot create a child watcher as the very first thing in 416 This means you cannot create a child watcher as the very first thing in
269 an AnyEvent program, you *have* to create at least one watcher before 417 an AnyEvent program, you *have* to create at least one watcher before
270 you "fork" the child (alternatively, you can call "AnyEvent::detect"). 418 you "fork" the child (alternatively, you can call "AnyEvent::detect").
271 419
420 As most event loops do not support waiting for child events, they will
421 be emulated by AnyEvent in most cases, in which the latency and race
422 problems mentioned in the description of signal watchers apply.
423
272 Example: fork a process and wait for it 424 Example: fork a process and wait for it
273 425
274 my $done = AnyEvent->condvar; 426 my $done = AnyEvent->condvar;
275 427
276 AnyEvent::detect; # force event module to be initialised
277
278 my $pid = fork or exit 5; 428 my $pid = fork or exit 5;
279 429
280 my $w = AnyEvent->child ( 430 my $w = AnyEvent->child (
281 pid => $pid, 431 pid => $pid,
282 cb => sub { 432 cb => sub {
283 my ($pid, $status) = @_; 433 my ($pid, $status) = @_;
284 warn "pid $pid exited with status $status"; 434 warn "pid $pid exited with status $status";
285 $done->broadcast; 435 $done->send;
286 }, 436 },
287 ); 437 );
288 438
289 # do something else, then wait for process exit 439 # do something else, then wait for process exit
290 $done->wait; 440 $done->recv;
441
442 IDLE WATCHERS
443 Sometimes there is a need to do something, but it is not so important to
444 do it instantly, but only when there is nothing better to do. This
445 "nothing better to do" is usually defined to be "no other events need
446 attention by the event loop".
447
448 Idle watchers ideally get invoked when the event loop has nothing better
449 to do, just before it would block the process to wait for new events.
450 Instead of blocking, the idle watcher is invoked.
451
452 Most event loops unfortunately do not really support idle watchers (only
453 EV, Event and Glib do it in a usable fashion) - for the rest, AnyEvent
454 will simply call the callback "from time to time".
455
456 Example: read lines from STDIN, but only process them when the program
457 is otherwise idle:
458
459 my @lines; # read data
460 my $idle_w;
461 my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
462 push @lines, scalar <STDIN>;
463
464 # start an idle watcher, if not already done
465 $idle_w ||= AnyEvent->idle (cb => sub {
466 # handle only one line, when there are lines left
467 if (my $line = shift @lines) {
468 print "handled when idle: $line";
469 } else {
470 # otherwise disable the idle watcher again
471 undef $idle_w;
472 }
473 });
474 });
291 475
292 CONDITION VARIABLES 476 CONDITION VARIABLES
477 If you are familiar with some event loops you will know that all of them
478 require you to run some blocking "loop", "run" or similar function that
479 will actively watch for new events and call your callbacks.
480
481 AnyEvent is slightly different: it expects somebody else to run the
482 event loop and will only block when necessary (usually when told by the
483 user).
484
485 The instrument to do that is called a "condition variable", so called
486 because they represent a condition that must become true.
487
488 Now is probably a good time to look at the examples further below.
489
293 Condition variables can be created by calling the "AnyEvent->condvar" 490 Condition variables can be created by calling the "AnyEvent->condvar"
294 method without any arguments. 491 method, usually without arguments. The only argument pair allowed is
492 "cb", which specifies a callback to be called when the condition
493 variable becomes true, with the condition variable as the first argument
494 (but not the results).
295 495
296 A condition variable waits for a condition - precisely that the 496 After creation, the condition variable is "false" until it becomes
297 "->broadcast" method has been called. 497 "true" by calling the "send" method (or calling the condition variable
498 as if it were a callback, read about the caveats in the description for
499 the "->send" method).
298 500
299 They are very useful to signal that a condition has been fulfilled, for 501 Condition variables are similar to callbacks, except that you can
502 optionally wait for them. They can also be called merge points - points
503 in time where multiple outstanding events have been processed. And yet
504 another way to call them is transactions - each condition variable can
505 be used to represent a transaction, which finishes at some point and
506 delivers a result. And yet some people know them as "futures" - a
507 promise to compute/deliver something that you can wait for.
508
509 Condition variables are very useful to signal that something has
300 example, if you write a module that does asynchronous http requests, 510 finished, for example, if you write a module that does asynchronous http
301 then a condition variable would be the ideal candidate to signal the 511 requests, then a condition variable would be the ideal candidate to
302 availability of results. 512 signal the availability of results. The user can either act when the
513 callback is called or can synchronously "->recv" for the results.
303 514
304 You can also use condition variables to block your main program until an 515 You can also use them to simulate traditional event loops - for example,
305 event occurs - for example, you could "->wait" in your main program 516 you can block your main program until an event occurs - for example, you
306 until the user clicks the Quit button in your app, which would 517 could "->recv" in your main program until the user clicks the Quit
307 "->broadcast" the "quit" event. 518 button of your app, which would "->send" the "quit" event.
308 519
309 Note that condition variables recurse into the event loop - if you have 520 Note that condition variables recurse into the event loop - if you have
310 two pirces of code that call "->wait" in a round-robbin fashion, you 521 two pieces of code that call "->recv" in a round-robin fashion, you
311 lose. Therefore, condition variables are good to export to your caller, 522 lose. Therefore, condition variables are good to export to your caller,
312 but you should avoid making a blocking wait yourself, at least in 523 but you should avoid making a blocking wait yourself, at least in
313 callbacks, as this asks for trouble. 524 callbacks, as this asks for trouble.
314 525
315 This object has two methods: 526 Condition variables are represented by hash refs in perl, and the keys
527 used by AnyEvent itself are all named "_ae_XXX" to make subclassing easy
528 (it is often useful to build your own transaction class on top of
529 AnyEvent). To subclass, use "AnyEvent::CondVar" as base class and call
530 it's "new" method in your own "new" method.
316 531
317 $cv->wait 532 There are two "sides" to a condition variable - the "producer side"
318 Wait (blocking if necessary) until the "->broadcast" method has been 533 which eventually calls "-> send", and the "consumer side", which waits
319 called on c<$cv>, while servicing other watchers normally. 534 for the send to occur.
320 535
321 You can only wait once on a condition - additional calls will return 536 Example: wait for a timer.
322 immediately.
323
324 Not all event models support a blocking wait - some die in that case
325 (programs might want to do that to stay interactive), so *if you are
326 using this from a module, never require a blocking wait*, but let
327 the caller decide whether the call will block or not (for example,
328 by coupling condition variables with some kind of request results
329 and supporting callbacks so the caller knows that getting the result
330 will not block, while still suppporting blocking waits if the caller
331 so desires).
332
333 Another reason *never* to "->wait" in a module is that you cannot
334 sensibly have two "->wait"'s in parallel, as that would require
335 multiple interpreters or coroutines/threads, none of which
336 "AnyEvent" can supply (the coroutine-aware backends
337 AnyEvent::Impl::CoroEV and AnyEvent::Impl::CoroEvent explicitly
338 support concurrent "->wait"'s from different coroutines, however).
339
340 $cv->broadcast
341 Flag the condition as ready - a running "->wait" and all further
342 calls to "wait" will (eventually) return after this method has been
343 called. If nobody is waiting the broadcast will be remembered..
344
345 Example:
346 537
347 # wait till the result is ready 538 # wait till the result is ready
348 my $result_ready = AnyEvent->condvar; 539 my $result_ready = AnyEvent->condvar;
349 540
350 # do something such as adding a timer 541 # do something such as adding a timer
351 # or socket watcher the calls $result_ready->broadcast 542 # or socket watcher the calls $result_ready->send
352 # when the "result" is ready. 543 # when the "result" is ready.
353 # in this case, we simply use a timer: 544 # in this case, we simply use a timer:
354 my $w = AnyEvent->timer ( 545 my $w = AnyEvent->timer (
355 after => 1, 546 after => 1,
356 cb => sub { $result_ready->broadcast }, 547 cb => sub { $result_ready->send },
357 ); 548 );
358 549
359 # this "blocks" (while handling events) till the watcher 550 # this "blocks" (while handling events) till the callback
360 # calls broadcast 551 # calls -<send
361 $result_ready->wait; 552 $result_ready->recv;
553
554 Example: wait for a timer, but take advantage of the fact that condition
555 variables are also callable directly.
556
557 my $done = AnyEvent->condvar;
558 my $delay = AnyEvent->timer (after => 5, cb => $done);
559 $done->recv;
560
561 Example: Imagine an API that returns a condvar and doesn't support
562 callbacks. This is how you make a synchronous call, for example from the
563 main program:
564
565 use AnyEvent::CouchDB;
566
567 ...
568
569 my @info = $couchdb->info->recv;
570
571 And this is how you would just set a callback to be called whenever the
572 results are available:
573
574 $couchdb->info->cb (sub {
575 my @info = $_[0]->recv;
576 });
577
578 METHODS FOR PRODUCERS
579 These methods should only be used by the producing side, i.e. the
580 code/module that eventually sends the signal. Note that it is also the
581 producer side which creates the condvar in most cases, but it isn't
582 uncommon for the consumer to create it as well.
583
584 $cv->send (...)
585 Flag the condition as ready - a running "->recv" and all further
586 calls to "recv" will (eventually) return after this method has been
587 called. If nobody is waiting the send will be remembered.
588
589 If a callback has been set on the condition variable, it is called
590 immediately from within send.
591
592 Any arguments passed to the "send" call will be returned by all
593 future "->recv" calls.
594
595 Condition variables are overloaded so one can call them directly (as
596 if they were a code reference). Calling them directly is the same as
597 calling "send".
598
599 $cv->croak ($error)
600 Similar to send, but causes all call's to "->recv" to invoke
601 "Carp::croak" with the given error message/object/scalar.
602
603 This can be used to signal any errors to the condition variable
604 user/consumer. Doing it this way instead of calling "croak" directly
605 delays the error detetcion, but has the overwhelmign advantage that
606 it diagnoses the error at the place where the result is expected,
607 and not deep in some event clalback without connection to the actual
608 code causing the problem.
609
610 $cv->begin ([group callback])
611 $cv->end
612 These two methods can be used to combine many transactions/events
613 into one. For example, a function that pings many hosts in parallel
614 might want to use a condition variable for the whole process.
615
616 Every call to "->begin" will increment a counter, and every call to
617 "->end" will decrement it. If the counter reaches 0 in "->end", the
618 (last) callback passed to "begin" will be executed. That callback is
619 *supposed* to call "->send", but that is not required. If no
620 callback was set, "send" will be called without any arguments.
621
622 You can think of "$cv->send" giving you an OR condition (one call
623 sends), while "$cv->begin" and "$cv->end" giving you an AND
624 condition (all "begin" calls must be "end"'ed before the condvar
625 sends).
626
627 Let's start with a simple example: you have two I/O watchers (for
628 example, STDOUT and STDERR for a program), and you want to wait for
629 both streams to close before activating a condvar:
630
631 my $cv = AnyEvent->condvar;
632
633 $cv->begin; # first watcher
634 my $w1 = AnyEvent->io (fh => $fh1, cb => sub {
635 defined sysread $fh1, my $buf, 4096
636 or $cv->end;
637 });
638
639 $cv->begin; # second watcher
640 my $w2 = AnyEvent->io (fh => $fh2, cb => sub {
641 defined sysread $fh2, my $buf, 4096
642 or $cv->end;
643 });
644
645 $cv->recv;
646
647 This works because for every event source (EOF on file handle),
648 there is one call to "begin", so the condvar waits for all calls to
649 "end" before sending.
650
651 The ping example mentioned above is slightly more complicated, as
652 the there are results to be passwd back, and the number of tasks
653 that are begung can potentially be zero:
654
655 my $cv = AnyEvent->condvar;
656
657 my %result;
658 $cv->begin (sub { $cv->send (\%result) });
659
660 for my $host (@list_of_hosts) {
661 $cv->begin;
662 ping_host_then_call_callback $host, sub {
663 $result{$host} = ...;
664 $cv->end;
665 };
666 }
667
668 $cv->end;
669
670 This code fragment supposedly pings a number of hosts and calls
671 "send" after results for all then have have been gathered - in any
672 order. To achieve this, the code issues a call to "begin" when it
673 starts each ping request and calls "end" when it has received some
674 result for it. Since "begin" and "end" only maintain a counter, the
675 order in which results arrive is not relevant.
676
677 There is an additional bracketing call to "begin" and "end" outside
678 the loop, which serves two important purposes: first, it sets the
679 callback to be called once the counter reaches 0, and second, it
680 ensures that "send" is called even when "no" hosts are being pinged
681 (the loop doesn't execute once).
682
683 This is the general pattern when you "fan out" into multiple (but
684 potentially none) subrequests: use an outer "begin"/"end" pair to
685 set the callback and ensure "end" is called at least once, and then,
686 for each subrequest you start, call "begin" and for each subrequest
687 you finish, call "end".
688
689 METHODS FOR CONSUMERS
690 These methods should only be used by the consuming side, i.e. the code
691 awaits the condition.
692
693 $cv->recv
694 Wait (blocking if necessary) until the "->send" or "->croak" methods
695 have been called on c<$cv>, while servicing other watchers normally.
696
697 You can only wait once on a condition - additional calls are valid
698 but will return immediately.
699
700 If an error condition has been set by calling "->croak", then this
701 function will call "croak".
702
703 In list context, all parameters passed to "send" will be returned,
704 in scalar context only the first one will be returned.
705
706 Note that doing a blocking wait in a callback is not supported by
707 any event loop, that is, recursive invocation of a blocking "->recv"
708 is not allowed, and the "recv" call will "croak" if such a condition
709 is detected. This condition can be slightly loosened by using
710 Coro::AnyEvent, which allows you to do a blocking "->recv" from any
711 thread that doesn't run the event loop itself.
712
713 Not all event models support a blocking wait - some die in that case
714 (programs might want to do that to stay interactive), so *if you are
715 using this from a module, never require a blocking wait*. Instead,
716 let the caller decide whether the call will block or not (for
717 example, by coupling condition variables with some kind of request
718 results and supporting callbacks so the caller knows that getting
719 the result will not block, while still supporting blocking waits if
720 the caller so desires).
721
722 You can ensure that "-recv" never blocks by setting a callback and
723 only calling "->recv" from within that callback (or at a later
724 time). This will work even when the event loop does not support
725 blocking waits otherwise.
726
727 $bool = $cv->ready
728 Returns true when the condition is "true", i.e. whether "send" or
729 "croak" have been called.
730
731 $cb = $cv->cb ($cb->($cv))
732 This is a mutator function that returns the callback set and
733 optionally replaces it before doing so.
734
735 The callback will be called when the condition becomes "true", i.e.
736 when "send" or "croak" are called, with the only argument being the
737 condition variable itself. Calling "recv" inside the callback or at
738 any later time is guaranteed not to block.
739
740SUPPORTED EVENT LOOPS/BACKENDS
741 The available backend classes are (every class has its own manpage):
742
743 Backends that are autoprobed when no other event loop can be found.
744 EV is the preferred backend when no other event loop seems to be in
745 use. If EV is not installed, then AnyEvent will try Event, and,
746 failing that, will fall back to its own pure-perl implementation,
747 which is available everywhere as it comes with AnyEvent itself.
748
749 AnyEvent::Impl::EV based on EV (interface to libev, best choice).
750 AnyEvent::Impl::Event based on Event, very stable, few glitches.
751 AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
752
753 Backends that are transparently being picked up when they are used.
754 These will be used when they are currently loaded when the first
755 watcher is created, in which case it is assumed that the application
756 is using them. This means that AnyEvent will automatically pick the
757 right backend when the main program loads an event module before
758 anything starts to create watchers. Nothing special needs to be done
759 by the main program.
760
761 AnyEvent::Impl::Glib based on Glib, slow but very stable.
762 AnyEvent::Impl::Tk based on Tk, very broken.
763 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
764 AnyEvent::Impl::POE based on POE, very slow, some limitations.
765
766 Backends with special needs.
767 Qt requires the Qt::Application to be instantiated first, but will
768 otherwise be picked up automatically. As long as the main program
769 instantiates the application before any AnyEvent watchers are
770 created, everything should just work.
771
772 AnyEvent::Impl::Qt based on Qt.
773
774 Support for IO::Async can only be partial, as it is too broken and
775 architecturally limited to even support the AnyEvent API. It also is
776 the only event loop that needs the loop to be set explicitly, so it
777 can only be used by a main program knowing about AnyEvent. See
778 AnyEvent::Impl::Async for the gory details.
779
780 AnyEvent::Impl::IOAsync based on IO::Async, cannot be autoprobed.
781
782 Event loops that are indirectly supported via other backends.
783 Some event loops can be supported via other modules:
784
785 There is no direct support for WxWidgets (Wx) or Prima.
786
787 WxWidgets has no support for watching file handles. However, you can
788 use WxWidgets through the POE adaptor, as POE has a Wx backend that
789 simply polls 20 times per second, which was considered to be too
790 horrible to even consider for AnyEvent.
791
792 Prima is not supported as nobody seems to be using it, but it has a
793 POE backend, so it can be supported through POE.
794
795 AnyEvent knows about both Prima and Wx, however, and will try to
796 load POE when detecting them, in the hope that POE will pick them
797 up, in which case everything will be automatic.
362 798
363GLOBAL VARIABLES AND FUNCTIONS 799GLOBAL VARIABLES AND FUNCTIONS
800 These are not normally required to use AnyEvent, but can be useful to
801 write AnyEvent extension modules.
802
364 $AnyEvent::MODEL 803 $AnyEvent::MODEL
365 Contains "undef" until the first watcher is being created. Then it 804 Contains "undef" until the first watcher is being created, before
805 the backend has been autodetected.
806
366 contains the event model that is being used, which is the name of 807 Afterwards it contains the event model that is being used, which is
367 the Perl class implementing the model. This class is usually one of 808 the name of the Perl class implementing the model. This class is
368 the "AnyEvent::Impl:xxx" modules, but can be any other class in the 809 usually one of the "AnyEvent::Impl:xxx" modules, but can be any
369 case AnyEvent has been extended at runtime (e.g. in *rxvt-unicode*). 810 other class in the case AnyEvent has been extended at runtime (e.g.
370 811 in *rxvt-unicode* it will be "urxvt::anyevent").
371 The known classes so far are:
372
373 AnyEvent::Impl::CoroEV based on Coro::EV, best choice.
374 AnyEvent::Impl::CoroEvent based on Coro::Event, second best choice.
375 AnyEvent::Impl::EV based on EV (an interface to libev, best choice).
376 AnyEvent::Impl::Event based on Event, second best choice.
377 AnyEvent::Impl::Glib based on Glib, third-best choice.
378 AnyEvent::Impl::Perl pure-perl implementation, inefficient but portable.
379 AnyEvent::Impl::Tk based on Tk, very bad choice.
380 AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).
381 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
382 AnyEvent::Impl::POE based on POE, not generic enough for full support.
383
384 There is no support for WxWidgets, as WxWidgets has no support for
385 watching file handles. However, you can use WxWidgets through the
386 POE Adaptor, as POE has a Wx backend that simply polls 20 times per
387 second, which was considered to be too horrible to even consider for
388 AnyEvent. Likewise, other POE backends can be used by AnyEvent by
389 using it's adaptor.
390
391 AnyEvent knows about Prima and Wx and will try to use POE when
392 autodetecting them.
393 812
394 AnyEvent::detect 813 AnyEvent::detect
395 Returns $AnyEvent::MODEL, forcing autodetection of the event model 814 Returns $AnyEvent::MODEL, forcing autodetection of the event model
396 if necessary. You should only call this function right before you 815 if necessary. You should only call this function right before you
397 would have created an AnyEvent watcher anyway, that is, as late as 816 would have created an AnyEvent watcher anyway, that is, as late as
398 possible at runtime. 817 possible at runtime, and not e.g. while initialising of your module.
818
819 If you need to do some initialisation before AnyEvent watchers are
820 created, use "post_detect".
821
822 $guard = AnyEvent::post_detect { BLOCK }
823 Arranges for the code block to be executed as soon as the event
824 model is autodetected (or immediately if this has already happened).
825
826 The block will be executed *after* the actual backend has been
827 detected ($AnyEvent::MODEL is set), but *before* any watchers have
828 been created, so it is possible to e.g. patch @AnyEvent::ISA or do
829 other initialisations - see the sources of AnyEvent::Strict or
830 AnyEvent::AIO to see how this is used.
831
832 The most common usage is to create some global watchers, without
833 forcing event module detection too early, for example, AnyEvent::AIO
834 creates and installs the global IO::AIO watcher in a "post_detect"
835 block to avoid autodetecting the event module at load time.
836
837 If called in scalar or list context, then it creates and returns an
838 object that automatically removes the callback again when it is
839 destroyed. See Coro::BDB for a case where this is useful.
840
841 @AnyEvent::post_detect
842 If there are any code references in this array (you can "push" to it
843 before or after loading AnyEvent), then they will called directly
844 after the event loop has been chosen.
845
846 You should check $AnyEvent::MODEL before adding to this array,
847 though: if it is defined then the event loop has already been
848 detected, and the array will be ignored.
849
850 Best use "AnyEvent::post_detect { BLOCK }" when your application
851 allows it,as it takes care of these details.
852
853 This variable is mainly useful for modules that can do something
854 useful when AnyEvent is used and thus want to know when it is
855 initialised, but do not need to even load it by default. This array
856 provides the means to hook into AnyEvent passively, without loading
857 it.
399 858
400WHAT TO DO IN A MODULE 859WHAT TO DO IN A MODULE
401 As a module author, you should "use AnyEvent" and call AnyEvent methods 860 As a module author, you should "use AnyEvent" and call AnyEvent methods
402 freely, but you should not load a specific event module or rely on it. 861 freely, but you should not load a specific event module or rely on it.
403 862
404 Be careful when you create watchers in the module body - AnyEvent will 863 Be careful when you create watchers in the module body - AnyEvent will
405 decide which event module to use as soon as the first method is called, 864 decide which event module to use as soon as the first method is called,
406 so by calling AnyEvent in your module body you force the user of your 865 so by calling AnyEvent in your module body you force the user of your
407 module to load the event module first. 866 module to load the event module first.
408 867
409 Never call "->wait" on a condition variable unless you *know* that the 868 Never call "->recv" on a condition variable unless you *know* that the
410 "->broadcast" method has been called on it already. This is because it 869 "->send" method has been called on it already. This is because it will
411 will stall the whole program, and the whole point of using events is to 870 stall the whole program, and the whole point of using events is to stay
412 stay interactive. 871 interactive.
413 872
414 It is fine, however, to call "->wait" when the user of your module 873 It is fine, however, to call "->recv" when the user of your module
415 requests it (i.e. if you create a http request object ad have a method 874 requests it (i.e. if you create a http request object ad have a method
416 called "results" that returns the results, it should call "->wait" 875 called "results" that returns the results, it should call "->recv"
417 freely, as the user of your module knows what she is doing. always). 876 freely, as the user of your module knows what she is doing. always).
418 877
419WHAT TO DO IN THE MAIN PROGRAM 878WHAT TO DO IN THE MAIN PROGRAM
420 There will always be a single main program - the only place that should 879 There will always be a single main program - the only place that should
421 dictate which event model to use. 880 dictate which event model to use.
423 If it doesn't care, it can just "use AnyEvent" and use it itself, or not 882 If it doesn't care, it can just "use AnyEvent" and use it itself, or not
424 do anything special (it does not need to be event-based) and let 883 do anything special (it does not need to be event-based) and let
425 AnyEvent decide which implementation to chose if some module relies on 884 AnyEvent decide which implementation to chose if some module relies on
426 it. 885 it.
427 886
428 If the main program relies on a specific event model. For example, in 887 If the main program relies on a specific event model - for example, in
429 Gtk2 programs you have to rely on the Glib module. You should load the 888 Gtk2 programs you have to rely on the Glib module - you should load the
430 event module before loading AnyEvent or any module that uses it: 889 event module before loading AnyEvent or any module that uses it:
431 generally speaking, you should load it as early as possible. The reason 890 generally speaking, you should load it as early as possible. The reason
432 is that modules might create watchers when they are loaded, and AnyEvent 891 is that modules might create watchers when they are loaded, and AnyEvent
433 will decide on the event model to use as soon as it creates watchers, 892 will decide on the event model to use as soon as it creates watchers,
434 and it might chose the wrong one unless you load the correct one 893 and it might chose the wrong one unless you load the correct one
435 yourself. 894 yourself.
436 895
437 You can chose to use a rather inefficient pure-perl implementation by 896 You can chose to use a pure-perl implementation by loading the
438 loading the "AnyEvent::Impl::Perl" module, which gives you similar 897 "AnyEvent::Impl::Perl" module, which gives you similar behaviour
439 behaviour everywhere, but letting AnyEvent chose is generally better. 898 everywhere, but letting AnyEvent chose the model is generally better.
899
900 MAINLOOP EMULATION
901 Sometimes (often for short test scripts, or even standalone programs who
902 only want to use AnyEvent), you do not want to run a specific event
903 loop.
904
905 In that case, you can use a condition variable like this:
906
907 AnyEvent->condvar->recv;
908
909 This has the effect of entering the event loop and looping forever.
910
911 Note that usually your program has some exit condition, in which case it
912 is better to use the "traditional" approach of storing a condition
913 variable somewhere, waiting for it, and sending it when the program
914 should exit cleanly.
440 915
441OTHER MODULES 916OTHER MODULES
442 The following is a non-exhaustive list of additional modules that use 917 The following is a non-exhaustive list of additional modules that use
443 AnyEvent and can therefore be mixed easily with other AnyEvent modules 918 AnyEvent as a client and can therefore be mixed easily with other
444 in the same program. Some of the modules come with AnyEvent, some are 919 AnyEvent modules and other event loops in the same program. Some of the
445 available via CPAN. 920 modules come with AnyEvent, most are available via CPAN.
446 921
447 AnyEvent::Util 922 AnyEvent::Util
448 Contains various utility functions that replace often-used but 923 Contains various utility functions that replace often-used but
449 blocking functions such as "inet_aton" by event-/callback-based 924 blocking functions such as "inet_aton" by event-/callback-based
450 versions. 925 versions.
451 926
927 AnyEvent::Socket
928 Provides various utility functions for (internet protocol) sockets,
929 addresses and name resolution. Also functions to create non-blocking
930 tcp connections or tcp servers, with IPv6 and SRV record support and
931 more.
932
452 AnyEvent::Handle 933 AnyEvent::Handle
453 Provide read and write buffers and manages watchers for reads and 934 Provide read and write buffers, manages watchers for reads and
454 writes. 935 writes, supports raw and formatted I/O, I/O queued and fully
936 transparent and non-blocking SSL/TLS (via AnyEvent::TLS.
455 937
456 AnyEvent::Socket 938 AnyEvent::DNS
457 Provides a means to do non-blocking connects, accepts etc. 939 Provides rich asynchronous DNS resolver capabilities.
940
941 AnyEvent::HTTP
942 A simple-to-use HTTP library that is capable of making a lot of
943 concurrent HTTP requests.
458 944
459 AnyEvent::HTTPD 945 AnyEvent::HTTPD
460 Provides a simple web application server framework. 946 Provides a simple web application server framework.
461 947
462 AnyEvent::DNS
463 Provides asynchronous DNS resolver capabilities, beyond what
464 AnyEvent::Util offers.
465
466 AnyEvent::FastPing 948 AnyEvent::FastPing
467 The fastest ping in the west. 949 The fastest ping in the west.
468 950
951 AnyEvent::DBI
952 Executes DBI requests asynchronously in a proxy process.
953
954 AnyEvent::AIO
955 Truly asynchronous I/O, should be in the toolbox of every event
956 programmer. AnyEvent::AIO transparently fuses IO::AIO and AnyEvent
957 together.
958
959 AnyEvent::BDB
960 Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently
961 fuses BDB and AnyEvent together.
962
963 AnyEvent::GPSD
964 A non-blocking interface to gpsd, a daemon delivering GPS
965 information.
966
967 AnyEvent::IRC
968 AnyEvent based IRC client module family (replacing the older
469 Net::IRC3 969 Net::IRC3).
470 AnyEvent based IRC client module family.
471 970
472 Net::XMPP2 971 AnyEvent::XMPP
473 AnyEvent based XMPP (Jabber protocol) module family. 972 AnyEvent based XMPP (Jabber protocol) module family (replacing the
973 older Net::XMPP2>.
974
975 AnyEvent::IGS
976 A non-blocking interface to the Internet Go Server protocol (used by
977 App::IGS).
474 978
475 Net::FCP 979 Net::FCP
476 AnyEvent-based implementation of the Freenet Client Protocol, 980 AnyEvent-based implementation of the Freenet Client Protocol,
477 birthplace of AnyEvent. 981 birthplace of AnyEvent.
478 982
479 Event::ExecFlow 983 Event::ExecFlow
480 High level API for event-based execution flow control. 984 High level API for event-based execution flow control.
481 985
482 Coro 986 Coro
483 Has special support for AnyEvent. 987 Has special support for AnyEvent via Coro::AnyEvent.
484 988
485 IO::Lambda 989ERROR AND EXCEPTION HANDLING
486 The lambda approach to I/O - don't ask, look there. Can use 990 In general, AnyEvent does not do any error handling - it relies on the
487 AnyEvent. 991 caller to do that if required. The AnyEvent::Strict module (see also the
992 "PERL_ANYEVENT_STRICT" environment variable, below) provides strict
993 checking of all AnyEvent methods, however, which is highly useful during
994 development.
488 995
489 IO::AIO 996 As for exception handling (i.e. runtime errors and exceptions thrown
490 Truly asynchronous I/O, should be in the toolbox of every event 997 while executing a callback), this is not only highly event-loop
491 programmer. Can be trivially made to use AnyEvent. 998 specific, but also not in any way wrapped by this module, as this is the
999 job of the main program.
492 1000
493 BDB Truly asynchronous Berkeley DB access. Can be trivially made to use 1001 The pure perl event loop simply re-throws the exception (usually within
494 AnyEvent. 1002 "condvar->recv"), the Event and EV modules call "$Event/EV::DIED->()",
1003 Glib uses "install_exception_handler" and so on.
1004
1005ENVIRONMENT VARIABLES
1006 The following environment variables are used by this module or its
1007 submodules.
1008
1009 Note that AnyEvent will remove *all* environment variables starting with
1010 "PERL_ANYEVENT_" from %ENV when it is loaded while taint mode is
1011 enabled.
1012
1013 "PERL_ANYEVENT_VERBOSE"
1014 By default, AnyEvent will be completely silent except in fatal
1015 conditions. You can set this environment variable to make AnyEvent
1016 more talkative.
1017
1018 When set to 1 or higher, causes AnyEvent to warn about unexpected
1019 conditions, such as not being able to load the event model specified
1020 by "PERL_ANYEVENT_MODEL".
1021
1022 When set to 2 or higher, cause AnyEvent to report to STDERR which
1023 event model it chooses.
1024
1025 When set to 8 or higher, then AnyEvent will report extra information
1026 on which optional modules it loads and how it implements certain
1027 features.
1028
1029 "PERL_ANYEVENT_STRICT"
1030 AnyEvent does not do much argument checking by default, as thorough
1031 argument checking is very costly. Setting this variable to a true
1032 value will cause AnyEvent to load "AnyEvent::Strict" and then to
1033 thoroughly check the arguments passed to most method calls. If it
1034 finds any problems, it will croak.
1035
1036 In other words, enables "strict" mode.
1037
1038 Unlike "use strict" (or it's modern cousin, "use common::sense", it
1039 is definitely recommended to keep it off in production. Keeping
1040 "PERL_ANYEVENT_STRICT=1" in your environment while developing
1041 programs can be very useful, however.
1042
1043 "PERL_ANYEVENT_MODEL"
1044 This can be used to specify the event model to be used by AnyEvent,
1045 before auto detection and -probing kicks in. It must be a string
1046 consisting entirely of ASCII letters. The string "AnyEvent::Impl::"
1047 gets prepended and the resulting module name is loaded and if the
1048 load was successful, used as event model. If it fails to load
1049 AnyEvent will proceed with auto detection and -probing.
1050
1051 This functionality might change in future versions.
1052
1053 For example, to force the pure perl model (AnyEvent::Impl::Perl) you
1054 could start your program like this:
1055
1056 PERL_ANYEVENT_MODEL=Perl perl ...
1057
1058 "PERL_ANYEVENT_PROTOCOLS"
1059 Used by both AnyEvent::DNS and AnyEvent::Socket to determine
1060 preferences for IPv4 or IPv6. The default is unspecified (and might
1061 change, or be the result of auto probing).
1062
1063 Must be set to a comma-separated list of protocols or address
1064 families, current supported: "ipv4" and "ipv6". Only protocols
1065 mentioned will be used, and preference will be given to protocols
1066 mentioned earlier in the list.
1067
1068 This variable can effectively be used for denial-of-service attacks
1069 against local programs (e.g. when setuid), although the impact is
1070 likely small, as the program has to handle conenction and other
1071 failures anyways.
1072
1073 Examples: "PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6" - prefer IPv4 over
1074 IPv6, but support both and try to use both.
1075 "PERL_ANYEVENT_PROTOCOLS=ipv4" - only support IPv4, never try to
1076 resolve or contact IPv6 addresses.
1077 "PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4" support either IPv4 or IPv6, but
1078 prefer IPv6 over IPv4.
1079
1080 "PERL_ANYEVENT_EDNS0"
1081 Used by AnyEvent::DNS to decide whether to use the EDNS0 extension
1082 for DNS. This extension is generally useful to reduce DNS traffic,
1083 but some (broken) firewalls drop such DNS packets, which is why it
1084 is off by default.
1085
1086 Setting this variable to 1 will cause AnyEvent::DNS to announce
1087 EDNS0 in its DNS requests.
1088
1089 "PERL_ANYEVENT_MAX_FORKS"
1090 The maximum number of child processes that
1091 "AnyEvent::Util::fork_call" will create in parallel.
1092
1093 "PERL_ANYEVENT_MAX_OUTSTANDING_DNS"
1094 The default value for the "max_outstanding" parameter for the
1095 default DNS resolver - this is the maximum number of parallel DNS
1096 requests that are sent to the DNS server.
1097
1098 "PERL_ANYEVENT_RESOLV_CONF"
1099 The file to use instead of /etc/resolv.conf (or OS-specific
1100 configuration) in the default resolver. When set to the empty
1101 string, no default config will be used.
1102
1103 "PERL_ANYEVENT_CA_FILE", "PERL_ANYEVENT_CA_PATH".
1104 When neither "ca_file" nor "ca_path" was specified during
1105 AnyEvent::TLS context creation, and either of these environment
1106 variables exist, they will be used to specify CA certificate
1107 locations instead of a system-dependent default.
1108
1109 "PERL_ANYEVENT_AVOID_GUARD" and "PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT"
1110 When these are set to 1, then the respective modules are not loaded.
1111 Mostly good for testing AnyEvent itself.
495 1112
496SUPPLYING YOUR OWN EVENT MODEL INTERFACE 1113SUPPLYING YOUR OWN EVENT MODEL INTERFACE
497 This is an advanced topic that you do not normally need to use AnyEvent 1114 This is an advanced topic that you do not normally need to use AnyEvent
498 in a module. This section is only of use to event loop authors who want 1115 in a module. This section is only of use to event loop authors who want
499 to provide AnyEvent compatibility. 1116 to provide AnyEvent compatibility.
533 1150
534 *rxvt-unicode* also cheats a bit by not providing blocking access to 1151 *rxvt-unicode* also cheats a bit by not providing blocking access to
535 condition variables: code blocking while waiting for a condition will 1152 condition variables: code blocking while waiting for a condition will
536 "die". This still works with most modules/usages, and blocking calls 1153 "die". This still works with most modules/usages, and blocking calls
537 must not be done in an interactive application, so it makes sense. 1154 must not be done in an interactive application, so it makes sense.
538
539ENVIRONMENT VARIABLES
540 The following environment variables are used by this module:
541
542 "PERL_ANYEVENT_VERBOSE"
543 By default, AnyEvent will be completely silent except in fatal
544 conditions. You can set this environment variable to make AnyEvent
545 more talkative.
546
547 When set to 1 or higher, causes AnyEvent to warn about unexpected
548 conditions, such as not being able to load the event model specified
549 by "PERL_ANYEVENT_MODEL".
550
551 When set to 2 or higher, cause AnyEvent to report to STDERR which
552 event model it chooses.
553
554 "PERL_ANYEVENT_MODEL"
555 This can be used to specify the event model to be used by AnyEvent,
556 before autodetection and -probing kicks in. It must be a string
557 consisting entirely of ASCII letters. The string "AnyEvent::Impl::"
558 gets prepended and the resulting module name is loaded and if the
559 load was successful, used as event model. If it fails to load
560 AnyEvent will proceed with autodetection and -probing.
561
562 This functionality might change in future versions.
563
564 For example, to force the pure perl model (AnyEvent::Impl::Perl) you
565 could start your program like this:
566
567 PERL_ANYEVENT_MODEL=Perl perl ...
568 1155
569EXAMPLE PROGRAM 1156EXAMPLE PROGRAM
570 The following program uses an I/O watcher to read data from STDIN, a 1157 The following program uses an I/O watcher to read data from STDIN, a
571 timer to display a message once per second, and a condition variable to 1158 timer to display a message once per second, and a condition variable to
572 quit the program when the user enters quit: 1159 quit the program when the user enters quit:
580 poll => 'r', 1167 poll => 'r',
581 cb => sub { 1168 cb => sub {
582 warn "io event <$_[0]>\n"; # will always output <r> 1169 warn "io event <$_[0]>\n"; # will always output <r>
583 chomp (my $input = <STDIN>); # read a line 1170 chomp (my $input = <STDIN>); # read a line
584 warn "read: $input\n"; # output what has been read 1171 warn "read: $input\n"; # output what has been read
585 $cv->broadcast if $input =~ /^q/i; # quit program if /^q/i 1172 $cv->send if $input =~ /^q/i; # quit program if /^q/i
586 }, 1173 },
587 ); 1174 );
588 1175
589 my $time_watcher; # can only be used once 1176 my $time_watcher; # can only be used once
590 1177
595 }); 1182 });
596 } 1183 }
597 1184
598 new_timer; # create first timer 1185 new_timer; # create first timer
599 1186
600 $cv->wait; # wait until user enters /^q/i 1187 $cv->recv; # wait until user enters /^q/i
601 1188
602REAL-WORLD EXAMPLE 1189REAL-WORLD EXAMPLE
603 Consider the Net::FCP module. It features (among others) the following 1190 Consider the Net::FCP module. It features (among others) the following
604 API calls, which are to freenet what HTTP GET requests are to http: 1191 API calls, which are to freenet what HTTP GET requests are to http:
605 1192
654 syswrite $txn->{fh}, $txn->{request} 1241 syswrite $txn->{fh}, $txn->{request}
655 or die "connection or write error"; 1242 or die "connection or write error";
656 $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r }); 1243 $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
657 1244
658 Again, "fh_ready_r" waits till all data has arrived, and then stores the 1245 Again, "fh_ready_r" waits till all data has arrived, and then stores the
659 result and signals any possible waiters that the request ahs finished: 1246 result and signals any possible waiters that the request has finished:
660 1247
661 sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf}; 1248 sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
662 1249
663 if (end-of-file or data complete) { 1250 if (end-of-file or data complete) {
664 $txn->{result} = $txn->{buf}; 1251 $txn->{result} = $txn->{buf};
665 $txn->{finished}->broadcast; 1252 $txn->{finished}->send;
666 $txb->{cb}->($txn) of $txn->{cb}; # also call callback 1253 $txb->{cb}->($txn) of $txn->{cb}; # also call callback
667 } 1254 }
668 1255
669 The "result" method, finally, just waits for the finished signal (if the 1256 The "result" method, finally, just waits for the finished signal (if the
670 request was already finished, it doesn't wait, of course, and returns 1257 request was already finished, it doesn't wait, of course, and returns
671 the data: 1258 the data:
672 1259
673 $txn->{finished}->wait; 1260 $txn->{finished}->recv;
674 return $txn->{result}; 1261 return $txn->{result};
675 1262
676 The actual code goes further and collects all errors ("die"s, 1263 The actual code goes further and collects all errors ("die"s,
677 exceptions) that occured during request processing. The "result" method 1264 exceptions) that occurred during request processing. The "result" method
678 detects whether an exception as thrown (it is stored inside the $txn 1265 detects whether an exception as thrown (it is stored inside the $txn
679 object) and just throws the exception, which means connection errors and 1266 object) and just throws the exception, which means connection errors and
680 other problems get reported tot he code that tries to use the result, 1267 other problems get reported tot he code that tries to use the result,
681 not in a random callback. 1268 not in a random callback.
682 1269
713 1300
714 my $quit = AnyEvent->condvar; 1301 my $quit = AnyEvent->condvar;
715 1302
716 $fcp->txn_client_get ($url)->cb (sub { 1303 $fcp->txn_client_get ($url)->cb (sub {
717 ... 1304 ...
718 $quit->broadcast; 1305 $quit->send;
719 }); 1306 });
720 1307
721 $quit->wait; 1308 $quit->recv;
722 1309
723BENCHMARKS 1310BENCHMARKS
724 To give you an idea of the performance and overheads that AnyEvent adds 1311 To give you an idea of the performance and overheads that AnyEvent adds
725 over the event loops themselves and to give you an impression of the 1312 over the event loops themselves and to give you an impression of the
726 speed of various event loops I prepared some benchmarks. 1313 speed of various event loops I prepared some benchmarks.
727 1314
728 BENCHMARKING ANYEVENT OVERHEAD 1315 BENCHMARKING ANYEVENT OVERHEAD
729 Here is a benchmark of various supported event models used natively and 1316 Here is a benchmark of various supported event models used natively and
730 through anyevent. The benchmark creates a lot of timers (with a zero 1317 through AnyEvent. The benchmark creates a lot of timers (with a zero
731 timeout) and I/O watchers (watching STDOUT, a pty, to become writable, 1318 timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
732 which it is), lets them fire exactly once and destroys them again. 1319 which it is), lets them fire exactly once and destroys them again.
733 1320
734 Source code for this benchmark is found as eg/bench in the AnyEvent 1321 Source code for this benchmark is found as eg/bench in the AnyEvent
735 distribution. 1322 distribution.
751 between all watchers, to avoid adding memory overhead. That means 1338 between all watchers, to avoid adding memory overhead. That means
752 closure creation and memory usage is not included in the figures. 1339 closure creation and memory usage is not included in the figures.
753 1340
754 *invoke* is the time, in microseconds, used to invoke a simple callback. 1341 *invoke* is the time, in microseconds, used to invoke a simple callback.
755 The callback simply counts down a Perl variable and after it was invoked 1342 The callback simply counts down a Perl variable and after it was invoked
756 "watcher" times, it would "->broadcast" a condvar once to signal the end 1343 "watcher" times, it would "->send" a condvar once to signal the end of
757 of this phase. 1344 this phase.
758 1345
759 *destroy* is the time, in microseconds, that it takes to destroy a 1346 *destroy* is the time, in microseconds, that it takes to destroy a
760 single watcher. 1347 single watcher.
761 1348
762 Results 1349 Results
763 name watchers bytes create invoke destroy comment 1350 name watchers bytes create invoke destroy comment
764 EV/EV 400000 244 0.56 0.46 0.31 EV native interface 1351 EV/EV 400000 224 0.47 0.35 0.27 EV native interface
765 EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers 1352 EV/Any 100000 224 2.88 0.34 0.27 EV + AnyEvent watchers
766 CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal 1353 CoroEV/Any 100000 224 2.85 0.35 0.28 coroutines + Coro::Signal
767 Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation 1354 Perl/Any 100000 452 4.13 0.73 0.95 pure perl implementation
768 Event/Event 16000 516 31.88 31.30 0.85 Event native interface 1355 Event/Event 16000 517 32.20 31.80 0.81 Event native interface
769 Event/Any 16000 590 35.75 31.42 1.08 Event + AnyEvent watchers 1356 Event/Any 16000 590 35.85 31.55 1.06 Event + AnyEvent watchers
1357 IOAsync/Any 16000 989 38.10 32.77 11.13 via IO::Async::Loop::IO_Poll
1358 IOAsync/Any 16000 990 37.59 29.50 10.61 via IO::Async::Loop::Epoll
770 Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour 1359 Glib/Any 16000 1357 102.33 12.31 51.00 quadratic behaviour
771 Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers 1360 Tk/Any 2000 1860 27.20 66.31 14.00 SEGV with >> 2000 watchers
772 POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event 1361 POE/Event 2000 6328 109.99 751.67 14.02 via POE::Loop::Event
773 POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select 1362 POE/Select 2000 6027 94.54 809.13 579.80 via POE::Loop::Select
774 1363
775 Discussion 1364 Discussion
776 The benchmark does *not* measure scalability of the event loop very 1365 The benchmark does *not* measure scalability of the event loop very
777 well. For example, a select-based event loop (such as the pure perl one) 1366 well. For example, a select-based event loop (such as the pure perl one)
778 can never compete with an event loop that uses epoll when the number of 1367 can never compete with an event loop that uses epoll when the number of
804 benchmark. 1393 benchmark.
805 1394
806 The "Event" module has a relatively high setup and callback invocation 1395 The "Event" module has a relatively high setup and callback invocation
807 cost, but overall scores in on the third place. 1396 cost, but overall scores in on the third place.
808 1397
1398 "IO::Async" performs admirably well, about on par with "Event", even
1399 when using its pure perl backend.
1400
809 "Glib"'s memory usage is quite a bit higher, but it features a faster 1401 "Glib"'s memory usage is quite a bit higher, but it features a faster
810 callback invocation and overall ends up in the same class as "Event". 1402 callback invocation and overall ends up in the same class as "Event".
811 However, Glib scales extremely badly, doubling the number of watchers 1403 However, Glib scales extremely badly, doubling the number of watchers
812 increases the processing time by more than a factor of four, making it 1404 increases the processing time by more than a factor of four, making it
813 completely unusable when using larger numbers of watchers (note that 1405 completely unusable when using larger numbers of watchers (note that
823 the figures above). 1415 the figures above).
824 1416
825 "POE", regardless of underlying event loop (whether using its pure perl 1417 "POE", regardless of underlying event loop (whether using its pure perl
826 select-based backend or the Event module, the POE-EV backend couldn't be 1418 select-based backend or the Event module, the POE-EV backend couldn't be
827 tested because it wasn't working) shows abysmal performance and memory 1419 tested because it wasn't working) shows abysmal performance and memory
828 usage: Watchers use almost 30 times as much memory as EV watchers, and 1420 usage with AnyEvent: Watchers use almost 30 times as much memory as EV
829 10 times as much memory as Event (the high memory requirements are 1421 watchers, and 10 times as much memory as Event (the high memory
830 caused by requiring a session for each watcher). Watcher invocation 1422 requirements are caused by requiring a session for each watcher).
831 speed is almost 900 times slower than with AnyEvent's pure perl 1423 Watcher invocation speed is almost 900 times slower than with AnyEvent's
1424 pure perl implementation.
1425
832 implementation. The design of the POE adaptor class in AnyEvent can not 1426 The design of the POE adaptor class in AnyEvent can not really account
833 really account for this, as session creation overhead is small compared 1427 for the performance issues, though, as session creation overhead is
834 to execution of the state machine, which is coded pretty optimally 1428 small compared to execution of the state machine, which is coded pretty
835 within AnyEvent::Impl::POE. POE simply seems to be abysmally slow. 1429 optimally within AnyEvent::Impl::POE (and while everybody agrees that
1430 using multiple sessions is not a good approach, especially regarding
1431 memory usage, even the author of POE could not come up with a faster
1432 design).
836 1433
837 Summary 1434 Summary
838 * Using EV through AnyEvent is faster than any other event loop (even 1435 * Using EV through AnyEvent is faster than any other event loop (even
839 when used without AnyEvent), but most event loops have acceptable 1436 when used without AnyEvent), but most event loops have acceptable
840 performance with or without AnyEvent. 1437 performance with or without AnyEvent.
845 1442
846 * You should avoid POE like the plague if you want performance or 1443 * You should avoid POE like the plague if you want performance or
847 reasonable memory usage. 1444 reasonable memory usage.
848 1445
849 BENCHMARKING THE LARGE SERVER CASE 1446 BENCHMARKING THE LARGE SERVER CASE
850 This benchmark atcually benchmarks the event loop itself. It works by 1447 This benchmark actually benchmarks the event loop itself. It works by
851 creating a number of "servers": each server consists of a socketpair, a 1448 creating a number of "servers": each server consists of a socket pair, a
852 timeout watcher that gets reset on activity (but never fires), and an 1449 timeout watcher that gets reset on activity (but never fires), and an
853 I/O watcher waiting for input on one side of the socket. Each time the 1450 I/O watcher waiting for input on one side of the socket. Each time the
854 socket watcher reads a byte it will write that byte to a random other 1451 socket watcher reads a byte it will write that byte to a random other
855 "server". 1452 "server".
856 1453
857 The effect is that there will be a lot of I/O watchers, only part of 1454 The effect is that there will be a lot of I/O watchers, only part of
858 which are active at any one point (so there is a constant number of 1455 which are active at any one point (so there is a constant number of
859 active fds for each loop iterstaion, but which fds these are is random). 1456 active fds for each loop iteration, but which fds these are is random).
860 The timeout is reset each time something is read because that reflects 1457 The timeout is reset each time something is read because that reflects
861 how most timeouts work (and puts extra pressure on the event loops). 1458 how most timeouts work (and puts extra pressure on the event loops).
862 1459
863 In this benchmark, we use 10000 socketpairs (20000 sockets), of which 1460 In this benchmark, we use 10000 socket pairs (20000 sockets), of which
864 100 (1%) are active. This mirrors the activity of large servers with 1461 100 (1%) are active. This mirrors the activity of large servers with
865 many connections, most of which are idle at any one point in time. 1462 many connections, most of which are idle at any one point in time.
866 1463
867 Source code for this benchmark is found as eg/bench2 in the AnyEvent 1464 Source code for this benchmark is found as eg/bench2 in the AnyEvent
868 distribution. 1465 distribution.
869 1466
870 Explanation of the columns 1467 Explanation of the columns
871 *sockets* is the number of sockets, and twice the number of "servers" 1468 *sockets* is the number of sockets, and twice the number of "servers"
872 (as each server has a read and write socket end). 1469 (as each server has a read and write socket end).
873 1470
874 *create* is the time it takes to create a socketpair (which is 1471 *create* is the time it takes to create a socket pair (which is
875 nontrivial) and two watchers: an I/O watcher and a timeout watcher. 1472 nontrivial) and two watchers: an I/O watcher and a timeout watcher.
876 1473
877 *request*, the most important value, is the time it takes to handle a 1474 *request*, the most important value, is the time it takes to handle a
878 single "request", that is, reading the token from the pipe and 1475 single "request", that is, reading the token from the pipe and
879 forwarding it to another server. This includes deleting the old timeout 1476 forwarding it to another server. This includes deleting the old timeout
880 and creating a new one that moves the timeout into the future. 1477 and creating a new one that moves the timeout into the future.
881 1478
882 Results 1479 Results
883 name sockets create request 1480 name sockets create request
884 EV 20000 69.01 11.16 1481 EV 20000 69.01 11.16
885 Perl 20000 73.32 35.87 1482 Perl 20000 73.32 35.87
1483 IOAsync 20000 157.00 98.14 epoll
1484 IOAsync 20000 159.31 616.06 poll
886 Event 20000 212.62 257.32 1485 Event 20000 212.62 257.32
887 Glib 20000 651.16 1896.30 1486 Glib 20000 651.16 1896.30
888 POE 20000 349.67 12317.24 uses POE::Loop::Event 1487 POE 20000 349.67 12317.24 uses POE::Loop::Event
889 1488
890 Discussion 1489 Discussion
891 This benchmark *does* measure scalability and overall performance of the 1490 This benchmark *does* measure scalability and overall performance of the
892 particular event loop. 1491 particular event loop.
893 1492
894 EV is again fastest. Since it is using epoll on my system, the setup 1493 EV is again fastest. Since it is using epoll on my system, the setup
895 time is relatively high, though. 1494 time is relatively high, though.
896 1495
897 Perl surprisingly comes second. It is much faster than the C-based event 1496 Perl surprisingly comes second. It is much faster than the C-based event
898 loops Event and Glib. 1497 loops Event and Glib.
1498
1499 IO::Async performs very well when using its epoll backend, and still
1500 quite good compared to Glib when using its pure perl backend.
899 1501
900 Event suffers from high setup time as well (look at its code and you 1502 Event suffers from high setup time as well (look at its code and you
901 will understand why). Callback invocation also has a high overhead 1503 will understand why). Callback invocation also has a high overhead
902 compared to the "$_->() for .."-style loop that the Perl event loop 1504 compared to the "$_->() for .."-style loop that the Perl event loop
903 uses. Event uses select or poll in basically all documented 1505 uses. Event uses select or poll in basically all documented
909 POE is still completely out of the picture, taking over 1000 times as 1511 POE is still completely out of the picture, taking over 1000 times as
910 long as EV, and over 100 times as long as the Perl implementation, even 1512 long as EV, and over 100 times as long as the Perl implementation, even
911 though it uses a C-based event loop in this case. 1513 though it uses a C-based event loop in this case.
912 1514
913 Summary 1515 Summary
914 * The pure perl implementation performs extremely well, considering 1516 * The pure perl implementation performs extremely well.
915 that it uses select.
916 1517
917 * Avoid Glib or POE in large projects where performance matters. 1518 * Avoid Glib or POE in large projects where performance matters.
918 1519
919 BENCHMARKING SMALL SERVERS 1520 BENCHMARKING SMALL SERVERS
920 While event loops should scale (and select-based ones do not...) even to 1521 While event loops should scale (and select-based ones do not...) even to
944 and speed most when you have lots of watchers, not when you only have a 1545 and speed most when you have lots of watchers, not when you only have a
945 few of them). 1546 few of them).
946 1547
947 EV is again fastest. 1548 EV is again fastest.
948 1549
949 Perl again comes second. It is noticably faster than the C-based event 1550 Perl again comes second. It is noticeably faster than the C-based event
950 loops Event and Glib, although the difference is too small to really 1551 loops Event and Glib, although the difference is too small to really
951 matter. 1552 matter.
952 1553
953 POE also performs much better in this case, but is is still far behind 1554 POE also performs much better in this case, but is is still far behind
954 the others. 1555 the others.
955 1556
956 Summary 1557 Summary
957 * C-based event loops perform very well with small number of watchers, 1558 * C-based event loops perform very well with small number of watchers,
958 as the management overhead dominates. 1559 as the management overhead dominates.
959 1560
1561 THE IO::Lambda BENCHMARK
1562 Recently I was told about the benchmark in the IO::Lambda manpage, which
1563 could be misinterpreted to make AnyEvent look bad. In fact, the
1564 benchmark simply compares IO::Lambda with POE, and IO::Lambda looks
1565 better (which shouldn't come as a surprise to anybody). As such, the
1566 benchmark is fine, and mostly shows that the AnyEvent backend from
1567 IO::Lambda isn't very optimal. But how would AnyEvent compare when used
1568 without the extra baggage? To explore this, I wrote the equivalent
1569 benchmark for AnyEvent.
1570
1571 The benchmark itself creates an echo-server, and then, for 500 times,
1572 connects to the echo server, sends a line, waits for the reply, and then
1573 creates the next connection. This is a rather bad benchmark, as it
1574 doesn't test the efficiency of the framework or much non-blocking I/O,
1575 but it is a benchmark nevertheless.
1576
1577 name runtime
1578 Lambda/select 0.330 sec
1579 + optimized 0.122 sec
1580 Lambda/AnyEvent 0.327 sec
1581 + optimized 0.138 sec
1582 Raw sockets/select 0.077 sec
1583 POE/select, components 0.662 sec
1584 POE/select, raw sockets 0.226 sec
1585 POE/select, optimized 0.404 sec
1586
1587 AnyEvent/select/nb 0.085 sec
1588 AnyEvent/EV/nb 0.068 sec
1589 +state machine 0.134 sec
1590
1591 The benchmark is also a bit unfair (my fault): the IO::Lambda/POE
1592 benchmarks actually make blocking connects and use 100% blocking I/O,
1593 defeating the purpose of an event-based solution. All of the newly
1594 written AnyEvent benchmarks use 100% non-blocking connects (using
1595 AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS
1596 resolver), so AnyEvent is at a disadvantage here, as non-blocking
1597 connects generally require a lot more bookkeeping and event handling
1598 than blocking connects (which involve a single syscall only).
1599
1600 The last AnyEvent benchmark additionally uses AnyEvent::Handle, which
1601 offers similar expressive power as POE and IO::Lambda, using
1602 conventional Perl syntax. This means that both the echo server and the
1603 client are 100% non-blocking, further placing it at a disadvantage.
1604
1605 As you can see, the AnyEvent + EV combination even beats the
1606 hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl
1607 backend easily beats IO::Lambda and POE.
1608
1609 And even the 100% non-blocking version written using the high-level (and
1610 slow :) AnyEvent::Handle abstraction beats both POE and IO::Lambda by a
1611 large margin, even though it does all of DNS, tcp-connect and socket I/O
1612 in a non-blocking way.
1613
1614 The two AnyEvent benchmarks programs can be found as eg/ae0.pl and
1615 eg/ae2.pl in the AnyEvent distribution, the remaining benchmarks are
1616 part of the IO::lambda distribution and were used without any changes.
1617
1618SIGNALS
1619 AnyEvent currently installs handlers for these signals:
1620
1621 SIGCHLD
1622 A handler for "SIGCHLD" is installed by AnyEvent's child watcher
1623 emulation for event loops that do not support them natively. Also,
1624 some event loops install a similar handler.
1625
1626 Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE,
1627 then AnyEvent will reset it to default, to avoid losing child exit
1628 statuses.
1629
1630 SIGPIPE
1631 A no-op handler is installed for "SIGPIPE" when $SIG{PIPE} is
1632 "undef" when AnyEvent gets loaded.
1633
1634 The rationale for this is that AnyEvent users usually do not really
1635 depend on SIGPIPE delivery (which is purely an optimisation for
1636 shell use, or badly-written programs), but "SIGPIPE" can cause
1637 spurious and rare program exits as a lot of people do not expect
1638 "SIGPIPE" when writing to some random socket.
1639
1640 The rationale for installing a no-op handler as opposed to ignoring
1641 it is that this way, the handler will be restored to defaults on
1642 exec.
1643
1644 Feel free to install your own handler, or reset it to defaults.
1645
1646RECOMMENDED/OPTIONAL MODULES
1647 One of AnyEvent's main goals is to be 100% Pure-Perl(tm): only perl (and
1648 it's built-in modules) are required to use it.
1649
1650 That does not mean that AnyEvent won't take advantage of some additional
1651 modules if they are installed.
1652
1653 This section epxlains which additional modules will be used, and how
1654 they affect AnyEvent's operetion.
1655
1656 Async::Interrupt
1657 This slightly arcane module is used to implement fast signal
1658 handling: To my knowledge, there is no way to do completely
1659 race-free and quick signal handling in pure perl. To ensure that
1660 signals still get delivered, AnyEvent will start an interval timer
1661 to wake up perl (and catch the signals) with some delay (default is
1662 10 seconds, look for $AnyEvent::MAX_SIGNAL_LATENCY).
1663
1664 If this module is available, then it will be used to implement
1665 signal catching, which means that signals will not be delayed, and
1666 the event loop will not be interrupted regularly, which is more
1667 efficient (And good for battery life on laptops).
1668
1669 This affects not just the pure-perl event loop, but also other event
1670 loops that have no signal handling on their own (e.g. Glib, Tk, Qt).
1671
1672 Some event loops (POE, Event, Event::Lib) offer signal watchers
1673 natively, and either employ their own workarounds (POE) or use
1674 AnyEvent's workaround (using $AnyEvent::MAX_SIGNAL_LATENCY).
1675 Installing Async::Interrupt does nothing for those backends.
1676
1677 EV This module isn't really "optional", as it is simply one of the
1678 backend event loops that AnyEvent can use. However, it is simply the
1679 best event loop available in terms of features, speed and stability:
1680 It supports the AnyEvent API optimally, implements all the watcher
1681 types in XS, does automatic timer adjustments even when no monotonic
1682 clock is available, can take avdantage of advanced kernel interfaces
1683 such as "epoll" and "kqueue", and is the fastest backend *by far*.
1684 You can even embed Glib/Gtk2 in it (or vice versa, see EV::Glib and
1685 Glib::EV).
1686
1687 Guard
1688 The guard module, when used, will be used to implement
1689 "AnyEvent::Util::guard". This speeds up guards considerably (and
1690 uses a lot less memory), but otherwise doesn't affect guard
1691 operation much. It is purely used for performance.
1692
1693 JSON and JSON::XS
1694 This module is required when you want to read or write JSON data via
1695 AnyEvent::Handle. It is also written in pure-perl, but can take
1696 advantage of the ultra-high-speed JSON::XS module when it is
1697 installed.
1698
1699 In fact, AnyEvent::Handle will use JSON::XS by default if it is
1700 installed.
1701
1702 Net::SSLeay
1703 Implementing TLS/SSL in Perl is certainly interesting, but not very
1704 worthwhile: If this module is installed, then AnyEvent::Handle (with
1705 the help of AnyEvent::TLS), gains the ability to do TLS/SSL.
1706
1707 Time::HiRes
1708 This module is part of perl since release 5.008. It will be used
1709 when the chosen event library does not come with a timing source on
1710 it's own. The pure-perl event loop (AnyEvent::Impl::Perl) will
1711 additionally use it to try to use a monotonic clock for timing
1712 stability.
1713
960FORK 1714FORK
961 Most event libraries are not fork-safe. The ones who are usually are 1715 Most event libraries are not fork-safe. The ones who are usually are
962 because they are so inefficient. Only EV is fully fork-aware. 1716 because they rely on inefficient but fork-safe "select" or "poll" calls.
1717 Only EV is fully fork-aware.
963 1718
964 If you have to fork, you must either do so *before* creating your first 1719 If you have to fork, you must either do so *before* creating your first
965 watcher OR you must not use AnyEvent at all in the child. 1720 watcher OR you must not use AnyEvent at all in the child OR you must do
1721 something completely out of the scope of AnyEvent.
966 1722
967SECURITY CONSIDERATIONS 1723SECURITY CONSIDERATIONS
968 AnyEvent can be forced to load any event model via 1724 AnyEvent can be forced to load any event model via
969 $ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used 1725 $ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used
970 to execute arbitrary code or directly gain access, it can easily be used 1726 to execute arbitrary code or directly gain access, it can easily be used
973 model than specified in the variable. 1729 model than specified in the variable.
974 1730
975 You can make AnyEvent completely ignore this variable by deleting it 1731 You can make AnyEvent completely ignore this variable by deleting it
976 before the first watcher gets created, e.g. with a "BEGIN" block: 1732 before the first watcher gets created, e.g. with a "BEGIN" block:
977 1733
978 BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} } 1734 BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
979 1735
980 use AnyEvent; 1736 use AnyEvent;
1737
1738 Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
1739 be used to probe what backend is used and gain other information (which
1740 is probably even less useful to an attacker than PERL_ANYEVENT_MODEL),
1741 and $ENV{PERL_ANYEVENT_STRICT}.
1742
1743 Note that AnyEvent will remove *all* environment variables starting with
1744 "PERL_ANYEVENT_" from %ENV when it is loaded while taint mode is
1745 enabled.
1746
1747BUGS
1748 Perl 5.8 has numerous memleaks that sometimes hit this module and are
1749 hard to work around. If you suffer from memleaks, first upgrade to Perl
1750 5.10 and check wether the leaks still show up. (Perl 5.10.0 has other
1751 annoying memleaks, such as leaking on "map" and "grep" but it is usually
1752 not as pronounced).
981 1753
982SEE ALSO 1754SEE ALSO
983 Event modules: Coro::EV, EV, EV::Glib, Glib::EV, Coro::Event, Event, 1755 Utility functions: AnyEvent::Util.
984 Glib::Event, Glib, Coro, Tk, Event::Lib, Qt, POE.
985 1756
986 Implementations: AnyEvent::Impl::CoroEV, AnyEvent::Impl::EV, 1757 Event modules: EV, EV::Glib, Glib::EV, Event, Glib::Event, Glib, Tk,
987 AnyEvent::Impl::CoroEvent, AnyEvent::Impl::Event, AnyEvent::Impl::Glib, 1758 Event::Lib, Qt, POE.
988 AnyEvent::Impl::Tk, AnyEvent::Impl::Perl, AnyEvent::Impl::EventLib, 1759
1760 Implementations: AnyEvent::Impl::EV, AnyEvent::Impl::Event,
1761 AnyEvent::Impl::Glib, AnyEvent::Impl::Tk, AnyEvent::Impl::Perl,
989 AnyEvent::Impl::Qt, AnyEvent::Impl::POE. 1762 AnyEvent::Impl::EventLib, AnyEvent::Impl::Qt, AnyEvent::Impl::POE,
1763 AnyEvent::Impl::IOAsync.
990 1764
1765 Non-blocking file handles, sockets, TCP clients and servers:
1766 AnyEvent::Handle, AnyEvent::Socket, AnyEvent::TLS.
1767
1768 Asynchronous DNS: AnyEvent::DNS.
1769
1770 Coroutine support: Coro, Coro::AnyEvent, Coro::EV, Coro::Event,
1771
991 Nontrivial usage examples: Net::FCP, Net::XMPP2. 1772 Nontrivial usage examples: AnyEvent::GPSD, AnyEvent::XMPP,
1773 AnyEvent::HTTP.
992 1774
993AUTHOR 1775AUTHOR
994 Marc Lehmann <schmorp@schmorp.de> 1776 Marc Lehmann <schmorp@schmorp.de>
995 http://home.schmorp.de/ 1777 http://home.schmorp.de/
996 1778

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines