|
|
1 | NAME |
|
|
2 | AnyEvent - the DBI of event loop programming |
|
|
3 | |
|
|
4 | EV, Event, Glib, Tk, Perl, Event::Lib, Irssi, rxvt-unicode, IO::Async, |
|
|
5 | Qt and POE are various supported event loops/environments. |
|
|
6 | |
|
|
7 | SYNOPSIS |
|
|
8 | use AnyEvent; |
|
|
9 | |
|
|
10 | # file descriptor readable |
|
|
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) = @_; |
|
|
26 | ... |
|
|
27 | }); |
|
|
28 | |
|
|
29 | # called when event loop idle (if applicable) |
|
|
30 | my $w = AnyEvent->idle (cb => sub { ... }); |
|
|
31 | |
|
|
32 | my $w = AnyEvent->condvar; # stores whether a condition was flagged |
|
|
33 | $w->send; # wake up current and all future recv's |
|
|
34 | $w->recv; # enters "main loop" till $condvar gets ->send |
|
|
35 | # use a condvar in callback mode: |
|
|
36 | $w->cb (sub { $_[0]->recv }); |
|
|
37 | |
|
|
38 | INTRODUCTION/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 | |
|
|
43 | SUPPORT |
|
|
44 | There is a mailinglist for discussing all things AnyEvent, and an IRC |
|
|
45 | channel, too. |
|
|
46 | |
|
|
47 | See the AnyEvent project page at the Schmorpforge Ta-Sa Software |
|
|
48 | Repository, at <http://anyevent.schmorp.de>, for more info. |
|
|
49 | |
|
|
50 | WHY YOU SHOULD USE THIS MODULE (OR NOT) |
|
|
51 | Glib, POE, IO::Async, Event... CPAN offers event models by the dozen |
|
|
52 | nowadays. So what is different about AnyEvent? |
|
|
53 | |
|
|
54 | Executive Summary: AnyEvent is *compatible*, AnyEvent is *free of |
|
|
55 | policy* and AnyEvent is *small and efficient*. |
|
|
56 | |
|
|
57 | First and foremost, *AnyEvent is not an event model* itself, it only |
|
|
58 | interfaces to whatever event model the main program happens to use, in a |
|
|
59 | pragmatic way. For event models and certain classes of immortals alike, |
|
|
60 | the statement "there can only be one" is a bitter reality: In general, |
|
|
61 | only one event loop can be active at the same time in a process. |
|
|
62 | AnyEvent cannot change this, but it can hide the differences between |
|
|
63 | those event loops. |
|
|
64 | |
|
|
65 | The goal of AnyEvent is to offer module authors the ability to do event |
|
|
66 | programming (waiting for I/O or timer events) without subscribing to a |
|
|
67 | religion, a way of living, and most importantly: without forcing your |
|
|
68 | module users into the same thing by forcing them to use the same event |
|
|
69 | model you use. |
|
|
70 | |
|
|
71 | For modules like POE or IO::Async (which is a total misnomer as it is |
|
|
72 | actually doing all I/O *synchronously*...), using them in your module is |
|
|
73 | like joining a cult: After you joined, you are dependent on them and you |
|
|
74 | cannot use anything else, as they are simply incompatible to everything |
|
|
75 | that isn't them. What's worse, all the potential users of your module |
|
|
76 | are *also* forced to use the same event loop you use. |
|
|
77 | |
|
|
78 | AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works |
|
|
79 | fine. AnyEvent + Tk works fine etc. etc. but none of these work together |
|
|
80 | with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if your |
|
|
81 | module uses one of those, every user of your module has to use it, too. |
|
|
82 | But if your module uses AnyEvent, it works transparently with all event |
|
|
83 | models it supports (including stuff like IO::Async, as long as those use |
|
|
84 | one of the supported event loops. It is trivial to add new event loops |
|
|
85 | to AnyEvent, too, so it is future-proof). |
|
|
86 | |
|
|
87 | In addition to being free of having to use *the one and only true event |
|
|
88 | model*, AnyEvent also is free of bloat and policy: with POE or similar |
|
|
89 | modules, you get an enormous amount of code and strict rules you have to |
|
|
90 | follow. AnyEvent, on the other hand, is lean and up to the point, by |
|
|
91 | only offering the functionality that is necessary, in as thin as a |
|
|
92 | wrapper as technically possible. |
|
|
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 | |
|
|
100 | Now, if you *do want* lots of policy (this can arguably be somewhat |
|
|
101 | useful) and you want to force your users to use the one and only event |
|
|
102 | model, you should *not* use this module. |
|
|
103 | |
|
|
104 | DESCRIPTION |
|
|
105 | AnyEvent provides an identical interface to multiple event loops. This |
|
|
106 | allows module authors to utilise an event loop without forcing module |
|
|
107 | users to use the same event loop (as only a single event loop can |
|
|
108 | coexist peacefully at any one time). |
|
|
109 | |
|
|
110 | The interface itself is vaguely similar, but not identical to the Event |
|
|
111 | module. |
|
|
112 | |
|
|
113 | During the first call of any watcher-creation method, the module tries |
|
|
114 | to detect the currently loaded event loop by probing whether one of the |
|
|
115 | following modules is already loaded: EV, Event, Glib, |
|
|
116 | AnyEvent::Impl::Perl, Tk, Event::Lib, Qt, POE. The first one found is |
|
|
117 | used. If none are found, the module tries to load these modules |
|
|
118 | (excluding Tk, Event::Lib, Qt and POE as the pure perl adaptor should |
|
|
119 | always succeed) in the order given. The first one that can be |
|
|
120 | successfully loaded will be used. If, after this, still none could be |
|
|
121 | found, AnyEvent will fall back to a pure-perl event loop, which is not |
|
|
122 | very efficient, but should work everywhere. |
|
|
123 | |
|
|
124 | Because AnyEvent first checks for modules that are already loaded, |
|
|
125 | loading an event model explicitly before first using AnyEvent will |
|
|
126 | likely make that model the default. For example: |
|
|
127 | |
|
|
128 | use Tk; |
|
|
129 | use AnyEvent; |
|
|
130 | |
|
|
131 | # .. AnyEvent will likely default to Tk |
|
|
132 | |
|
|
133 | The *likely* means that, if any module loads another event model and |
|
|
134 | starts using it, all bets are off. Maybe you should tell their authors |
|
|
135 | to use AnyEvent so their modules work together with others seamlessly... |
|
|
136 | |
|
|
137 | The pure-perl implementation of AnyEvent is called |
|
|
138 | "AnyEvent::Impl::Perl". Like other event modules you can load it |
|
|
139 | explicitly and enjoy the high availability of that event loop :) |
|
|
140 | |
|
|
141 | WATCHERS |
|
|
142 | AnyEvent has the central concept of a *watcher*, which is an object that |
|
|
143 | stores relevant data for each kind of event you are waiting for, such as |
|
|
144 | the callback to call, the file handle to watch, etc. |
|
|
145 | |
|
|
146 | These watchers are normal Perl objects with normal Perl lifetime. After |
|
|
147 | creating a watcher it will immediately "watch" for events and invoke the |
|
|
148 | callback when the event occurs (of course, only when the event model is |
|
|
149 | in control). |
|
|
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 | |
|
|
157 | To disable the watcher you have to destroy it (e.g. by setting the |
|
|
158 | variable you store it in to "undef" or otherwise deleting all references |
|
|
159 | to it). |
|
|
160 | |
|
|
161 | All watchers are created by calling a method on the "AnyEvent" class. |
|
|
162 | |
|
|
163 | Many watchers either are used with "recursion" (repeating timers for |
|
|
164 | example), or need to refer to their watcher object in other ways. |
|
|
165 | |
|
|
166 | An any way to achieve that is this pattern: |
|
|
167 | |
|
|
168 | my $w; $w = AnyEvent->type (arg => value ..., cb => sub { |
|
|
169 | # you can use $w here, for example to undef it |
|
|
170 | undef $w; |
|
|
171 | }); |
|
|
172 | |
|
|
173 | Note that "my $w; $w =" combination. This is necessary because in Perl, |
|
|
174 | my variables are only visible after the statement in which they are |
|
|
175 | declared. |
|
|
176 | |
|
|
177 | I/O WATCHERS |
|
|
178 | $w = AnyEvent->io ( |
|
|
179 | fh => <filehandle_or_fileno>, |
|
|
180 | poll => <"r" or "w">, |
|
|
181 | cb => <callback>, |
|
|
182 | ); |
|
|
183 | |
|
|
184 | You can create an I/O watcher by calling the "AnyEvent->io" method with |
|
|
185 | the following mandatory key-value pairs as arguments: |
|
|
186 | |
|
|
187 | "fh" is the Perl *file handle* (or a naked file descriptor) to watch for |
|
|
188 | events (AnyEvent might or might not keep a reference to this file |
|
|
189 | handle). Note that only file handles pointing to things for which |
|
|
190 | non-blocking operation makes sense are allowed. This includes sockets, |
|
|
191 | most character devices, pipes, fifos and so on, but not for example |
|
|
192 | files or block devices. |
|
|
193 | |
|
|
194 | "poll" must be a string that is either "r" or "w", which creates a |
|
|
195 | watcher waiting for "r"eadable or "w"ritable events, respectively. |
|
|
196 | |
|
|
197 | "cb" is the callback to invoke each time the file handle becomes ready. |
|
|
198 | |
|
|
199 | Although the callback might get passed parameters, their value and |
|
|
200 | presence is undefined and you cannot rely on them. Portable AnyEvent |
|
|
201 | callbacks cannot use arguments passed to I/O watcher callbacks. |
|
|
202 | |
|
|
203 | The I/O watcher might use the underlying file descriptor or a copy of |
|
|
204 | it. You must not close a file handle as long as any watcher is active on |
|
|
205 | the underlying file descriptor. |
|
|
206 | |
|
|
207 | Some event loops issue spurious readyness notifications, so you should |
|
|
208 | always use non-blocking calls when reading/writing from/to your file |
|
|
209 | handles. |
|
|
210 | |
|
|
211 | Example: wait for readability of STDIN, then read a line and disable the |
|
|
212 | watcher. |
|
|
213 | |
|
|
214 | my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub { |
|
|
215 | chomp (my $input = <STDIN>); |
|
|
216 | warn "read: $input\n"; |
|
|
217 | undef $w; |
|
|
218 | }); |
|
|
219 | |
|
|
220 | TIME WATCHERS |
|
|
221 | $w = AnyEvent->timer (after => <seconds>, cb => <callback>); |
|
|
222 | |
|
|
223 | $w = AnyEvent->timer ( |
|
|
224 | after => <fractional_seconds>, |
|
|
225 | interval => <fractional_seconds>, |
|
|
226 | cb => <callback>, |
|
|
227 | ); |
|
|
228 | |
|
|
229 | You can create a time watcher by calling the "AnyEvent->timer" method |
|
|
230 | with the following mandatory arguments: |
|
|
231 | |
|
|
232 | "after" specifies after how many seconds (fractional values are |
|
|
233 | supported) the callback should be invoked. "cb" is the callback to |
|
|
234 | invoke in that case. |
|
|
235 | |
|
|
236 | Although the callback might get passed parameters, their value and |
|
|
237 | presence is undefined and you cannot rely on them. Portable AnyEvent |
|
|
238 | callbacks cannot use arguments passed to time watcher callbacks. |
|
|
239 | |
|
|
240 | The callback will normally be invoked once only. If you specify another |
|
|
241 | parameter, "interval", as a strictly positive number (> 0), then the |
|
|
242 | callback will be invoked regularly at that interval (in fractional |
|
|
243 | seconds) after the first invocation. If "interval" is specified with a |
|
|
244 | false value, then it is treated as if it were missing. |
|
|
245 | |
|
|
246 | The callback will be rescheduled before invoking the callback, but no |
|
|
247 | attempt is done to avoid timer drift in most backends, so the interval |
|
|
248 | is only approximate. |
|
|
249 | |
|
|
250 | Example: fire an event after 7.7 seconds. |
|
|
251 | |
|
|
252 | my $w = AnyEvent->timer (after => 7.7, cb => sub { |
|
|
253 | warn "timeout\n"; |
|
|
254 | }); |
|
|
255 | |
|
|
256 | # to cancel the timer: |
|
|
257 | undef $w; |
|
|
258 | |
|
|
259 | Example 2: fire an event after 0.5 seconds, then roughly every second. |
|
|
260 | |
|
|
261 | my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub { |
|
|
262 | warn "timeout\n"; |
|
|
263 | }; |
|
|
264 | |
|
|
265 | TIMING ISSUES |
|
|
266 | There are two ways to handle timers: based on real time (relative, "fire |
|
|
267 | in 10 seconds") and based on wallclock time (absolute, "fire at 12 |
|
|
268 | o'clock"). |
|
|
269 | |
|
|
270 | While most event loops expect timers to specified in a relative way, |
|
|
271 | they use absolute time internally. This makes a difference when your |
|
|
272 | clock "jumps", for example, when ntp decides to set your clock backwards |
|
|
273 | from the wrong date of 2014-01-01 to 2008-01-01, a watcher that is |
|
|
274 | supposed to fire "after" a second might actually take six years to |
|
|
275 | finally fire. |
|
|
276 | |
|
|
277 | AnyEvent cannot compensate for this. The only event loop that is |
|
|
278 | conscious about these issues is EV, which offers both relative |
|
|
279 | (ev_timer, based on true relative time) and absolute (ev_periodic, based |
|
|
280 | on wallclock time) timers. |
|
|
281 | |
|
|
282 | AnyEvent always prefers relative timers, if available, matching the |
|
|
283 | AnyEvent API. |
|
|
284 | |
|
|
285 | AnyEvent has two additional methods that return the "current time": |
|
|
286 | |
|
|
287 | AnyEvent->time |
|
|
288 | This returns the "current wallclock time" as a fractional number of |
|
|
289 | seconds since the Epoch (the same thing as "time" or |
|
|
290 | "Time::HiRes::time" return, and the result is guaranteed to be |
|
|
291 | compatible with those). |
|
|
292 | |
|
|
293 | It progresses independently of any event loop processing, i.e. each |
|
|
294 | call will check the system clock, which usually gets updated |
|
|
295 | frequently. |
|
|
296 | |
|
|
297 | AnyEvent->now |
|
|
298 | This also returns the "current wallclock time", but unlike "time", |
|
|
299 | above, this value might change only once per event loop iteration, |
|
|
300 | depending on the event loop (most return the same time as "time", |
|
|
301 | above). This is the time that AnyEvent's timers get scheduled |
|
|
302 | against. |
|
|
303 | |
|
|
304 | *In almost all cases (in all cases if you don't care), this is the |
|
|
305 | function to call when you want to know the current time.* |
|
|
306 | |
|
|
307 | This function is also often faster then "AnyEvent->time", and thus |
|
|
308 | the preferred method if you want some timestamp (for example, |
|
|
309 | AnyEvent::Handle uses this to update it's activity timeouts). |
|
|
310 | |
|
|
311 | The rest of this section is only of relevance if you try to be very |
|
|
312 | exact with your timing, you can skip it without bad conscience. |
|
|
313 | |
|
|
314 | For a practical example of when these times differ, consider |
|
|
315 | Event::Lib and EV and the following set-up: |
|
|
316 | |
|
|
317 | The event loop is running and has just invoked one of your callback |
|
|
318 | at time=500 (assume no other callbacks delay processing). In your |
|
|
319 | callback, you wait a second by executing "sleep 1" (blocking the |
|
|
320 | process for a second) and then (at time=501) you create a relative |
|
|
321 | timer that fires after three seconds. |
|
|
322 | |
|
|
323 | With Event::Lib, "AnyEvent->time" and "AnyEvent->now" will both |
|
|
324 | return 501, because that is the current time, and the timer will be |
|
|
325 | scheduled to fire at time=504 (501 + 3). |
|
|
326 | |
|
|
327 | With EV, "AnyEvent->time" returns 501 (as that is the current time), |
|
|
328 | but "AnyEvent->now" returns 500, as that is the time the last event |
|
|
329 | processing phase started. With EV, your timer gets scheduled to run |
|
|
330 | at time=503 (500 + 3). |
|
|
331 | |
|
|
332 | In one sense, Event::Lib is more exact, as it uses the current time |
|
|
333 | regardless of any delays introduced by event processing. However, |
|
|
334 | most callbacks do not expect large delays in processing, so this |
|
|
335 | causes a higher drift (and a lot more system calls to get the |
|
|
336 | current time). |
|
|
337 | |
|
|
338 | In another sense, EV is more exact, as your timer will be scheduled |
|
|
339 | at the same time, regardless of how long event processing actually |
|
|
340 | took. |
|
|
341 | |
|
|
342 | In either case, if you care (and in most cases, you don't), then you |
|
|
343 | can get whatever behaviour you want with any event loop, by taking |
|
|
344 | the difference between "AnyEvent->time" and "AnyEvent->now" into |
|
|
345 | account. |
|
|
346 | |
|
|
347 | AnyEvent->now_update |
|
|
348 | Some event loops (such as EV or AnyEvent::Impl::Perl) cache the |
|
|
349 | current time for each loop iteration (see the discussion of |
|
|
350 | AnyEvent->now, above). |
|
|
351 | |
|
|
352 | When a callback runs for a long time (or when the process sleeps), |
|
|
353 | then this "current" time will differ substantially from the real |
|
|
354 | time, which might affect timers and time-outs. |
|
|
355 | |
|
|
356 | When this is the case, you can call this method, which will update |
|
|
357 | the event loop's idea of "current time". |
|
|
358 | |
|
|
359 | A typical example would be a script in a web server (e.g. |
|
|
360 | "mod_perl") - when mod_perl executes the script, then the event loop |
|
|
361 | will have the wrong idea about the "current time" (being potentially |
|
|
362 | far in the past, when the script ran the last time). In that case |
|
|
363 | you should arrange a call to "AnyEvent->now_update" each time the |
|
|
364 | web server process wakes up again (e.g. at the start of your script, |
|
|
365 | or in a handler). |
|
|
366 | |
|
|
367 | Note that updating the time *might* cause some events to be handled. |
|
|
368 | |
|
|
369 | SIGNAL WATCHERS |
|
|
370 | $w = AnyEvent->signal (signal => <uppercase_signal_name>, cb => <callback>); |
|
|
371 | |
|
|
372 | You can watch for signals using a signal watcher, "signal" is the signal |
|
|
373 | *name* in uppercase and without any "SIG" prefix, "cb" is the Perl |
|
|
374 | callback to be invoked whenever a signal occurs. |
|
|
375 | |
|
|
376 | Although the callback might get passed parameters, their value and |
|
|
377 | presence is undefined and you cannot rely on them. Portable AnyEvent |
|
|
378 | callbacks cannot use arguments passed to signal watcher callbacks. |
|
|
379 | |
|
|
380 | Multiple signal occurrences can be clumped together into one callback |
|
|
381 | invocation, and callback invocation will be synchronous. Synchronous |
|
|
382 | means that it might take a while until the signal gets handled by the |
|
|
383 | process, but it is guaranteed not to interrupt any other callbacks. |
|
|
384 | |
|
|
385 | The main advantage of using these watchers is that you can share a |
|
|
386 | signal between multiple watchers, and AnyEvent will ensure that signals |
|
|
387 | will not interrupt your program at bad times. |
|
|
388 | |
|
|
389 | This watcher might use %SIG (depending on the event loop used), so |
|
|
390 | programs overwriting those signals directly will likely not work |
|
|
391 | correctly. |
|
|
392 | |
|
|
393 | Example: exit on SIGINT |
|
|
394 | |
|
|
395 | my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 }); |
|
|
396 | |
|
|
397 | Signal Races, Delays and Workarounds |
|
|
398 | Many event loops (e.g. Glib, Tk, Qt, IO::Async) do not support attaching |
|
|
399 | callbacks to signals in a generic way, which is a pity, as you cannot do |
|
|
400 | race-free signal handling in perl, requiring C libraries for this. |
|
|
401 | AnyEvent will try to do it's best, which means in some cases, signals |
|
|
402 | will be delayed. The maximum time a signal might be delayed is specified |
|
|
403 | in $AnyEvent::MAX_SIGNAL_LATENCY (default: 10 seconds). This variable |
|
|
404 | can be changed only before the first signal watcher is created, and |
|
|
405 | should be left alone otherwise. This variable determines how often |
|
|
406 | AnyEvent polls for signals (in case a wake-up was missed). Higher values |
|
|
407 | will cause fewer spurious wake-ups, which is better for power and CPU |
|
|
408 | saving. |
|
|
409 | |
|
|
410 | All these problems can be avoided by installing the optional |
|
|
411 | Async::Interrupt module, which works with most event loops. It will not |
|
|
412 | work with inherently broken event loops such as Event or Event::Lib (and |
|
|
413 | not with POE currently, as POE does it's own workaround with one-second |
|
|
414 | latency). For those, you just have to suffer the delays. |
|
|
415 | |
|
|
416 | CHILD PROCESS WATCHERS |
|
|
417 | $w = AnyEvent->child (pid => <process id>, cb => <callback>); |
|
|
418 | |
|
|
419 | You can also watch on a child process exit and catch its exit status. |
|
|
420 | |
|
|
421 | The child process is specified by the "pid" argument (one some backends, |
|
|
422 | using 0 watches for any child process exit, on others this will croak). |
|
|
423 | The watcher will be triggered only when the child process has finished |
|
|
424 | and an exit status is available, not on any trace events |
|
|
425 | (stopped/continued). |
|
|
426 | |
|
|
427 | The callback will be called with the pid and exit status (as returned by |
|
|
428 | waitpid), so unlike other watcher types, you *can* rely on child watcher |
|
|
429 | callback arguments. |
|
|
430 | |
|
|
431 | This watcher type works by installing a signal handler for "SIGCHLD", |
|
|
432 | and since it cannot be shared, nothing else should use SIGCHLD or reap |
|
|
433 | random child processes (waiting for specific child processes, e.g. |
|
|
434 | inside "system", is just fine). |
|
|
435 | |
|
|
436 | There is a slight catch to child watchers, however: you usually start |
|
|
437 | them *after* the child process was created, and this means the process |
|
|
438 | could have exited already (and no SIGCHLD will be sent anymore). |
|
|
439 | |
|
|
440 | Not all event models handle this correctly (neither POE nor IO::Async |
|
|
441 | do, see their AnyEvent::Impl manpages for details), but even for event |
|
|
442 | models that *do* handle this correctly, they usually need to be loaded |
|
|
443 | before the process exits (i.e. before you fork in the first place). |
|
|
444 | AnyEvent's pure perl event loop handles all cases correctly regardless |
|
|
445 | of when you start the watcher. |
|
|
446 | |
|
|
447 | This means you cannot create a child watcher as the very first thing in |
|
|
448 | an AnyEvent program, you *have* to create at least one watcher before |
|
|
449 | you "fork" the child (alternatively, you can call "AnyEvent::detect"). |
|
|
450 | |
|
|
451 | As most event loops do not support waiting for child events, they will |
|
|
452 | be emulated by AnyEvent in most cases, in which the latency and race |
|
|
453 | problems mentioned in the description of signal watchers apply. |
|
|
454 | |
|
|
455 | Example: fork a process and wait for it |
|
|
456 | |
|
|
457 | my $done = AnyEvent->condvar; |
|
|
458 | |
|
|
459 | my $pid = fork or exit 5; |
|
|
460 | |
|
|
461 | my $w = AnyEvent->child ( |
|
|
462 | pid => $pid, |
|
|
463 | cb => sub { |
|
|
464 | my ($pid, $status) = @_; |
|
|
465 | warn "pid $pid exited with status $status"; |
|
|
466 | $done->send; |
|
|
467 | }, |
|
|
468 | ); |
|
|
469 | |
|
|
470 | # do something else, then wait for process exit |
|
|
471 | $done->recv; |
|
|
472 | |
|
|
473 | IDLE WATCHERS |
|
|
474 | $w = AnyEvent->idle (cb => <callback>); |
|
|
475 | |
|
|
476 | Sometimes there is a need to do something, but it is not so important to |
|
|
477 | do it instantly, but only when there is nothing better to do. This |
|
|
478 | "nothing better to do" is usually defined to be "no other events need |
|
|
479 | attention by the event loop". |
|
|
480 | |
|
|
481 | Idle watchers ideally get invoked when the event loop has nothing better |
|
|
482 | to do, just before it would block the process to wait for new events. |
|
|
483 | Instead of blocking, the idle watcher is invoked. |
|
|
484 | |
|
|
485 | Most event loops unfortunately do not really support idle watchers (only |
|
|
486 | EV, Event and Glib do it in a usable fashion) - for the rest, AnyEvent |
|
|
487 | will simply call the callback "from time to time". |
|
|
488 | |
|
|
489 | Example: read lines from STDIN, but only process them when the program |
|
|
490 | is otherwise idle: |
|
|
491 | |
|
|
492 | my @lines; # read data |
|
|
493 | my $idle_w; |
|
|
494 | my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub { |
|
|
495 | push @lines, scalar <STDIN>; |
|
|
496 | |
|
|
497 | # start an idle watcher, if not already done |
|
|
498 | $idle_w ||= AnyEvent->idle (cb => sub { |
|
|
499 | # handle only one line, when there are lines left |
|
|
500 | if (my $line = shift @lines) { |
|
|
501 | print "handled when idle: $line"; |
|
|
502 | } else { |
|
|
503 | # otherwise disable the idle watcher again |
|
|
504 | undef $idle_w; |
|
|
505 | } |
|
|
506 | }); |
|
|
507 | }); |
|
|
508 | |
|
|
509 | CONDITION VARIABLES |
|
|
510 | $cv = AnyEvent->condvar; |
|
|
511 | |
|
|
512 | $cv->send (<list>); |
|
|
513 | my @res = $cv->recv; |
|
|
514 | |
|
|
515 | If you are familiar with some event loops you will know that all of them |
|
|
516 | require you to run some blocking "loop", "run" or similar function that |
|
|
517 | will actively watch for new events and call your callbacks. |
|
|
518 | |
|
|
519 | AnyEvent is slightly different: it expects somebody else to run the |
|
|
520 | event loop and will only block when necessary (usually when told by the |
|
|
521 | user). |
|
|
522 | |
|
|
523 | The instrument to do that is called a "condition variable", so called |
|
|
524 | because they represent a condition that must become true. |
|
|
525 | |
|
|
526 | Now is probably a good time to look at the examples further below. |
|
|
527 | |
|
|
528 | Condition variables can be created by calling the "AnyEvent->condvar" |
|
|
529 | method, usually without arguments. The only argument pair allowed is |
|
|
530 | "cb", which specifies a callback to be called when the condition |
|
|
531 | variable becomes true, with the condition variable as the first argument |
|
|
532 | (but not the results). |
|
|
533 | |
|
|
534 | After creation, the condition variable is "false" until it becomes |
|
|
535 | "true" by calling the "send" method (or calling the condition variable |
|
|
536 | as if it were a callback, read about the caveats in the description for |
|
|
537 | the "->send" method). |
|
|
538 | |
|
|
539 | Condition variables are similar to callbacks, except that you can |
|
|
540 | optionally wait for them. They can also be called merge points - points |
|
|
541 | in time where multiple outstanding events have been processed. And yet |
|
|
542 | another way to call them is transactions - each condition variable can |
|
|
543 | be used to represent a transaction, which finishes at some point and |
|
|
544 | delivers a result. And yet some people know them as "futures" - a |
|
|
545 | promise to compute/deliver something that you can wait for. |
|
|
546 | |
|
|
547 | Condition variables are very useful to signal that something has |
|
|
548 | finished, for example, if you write a module that does asynchronous http |
|
|
549 | requests, then a condition variable would be the ideal candidate to |
|
|
550 | signal the availability of results. The user can either act when the |
|
|
551 | callback is called or can synchronously "->recv" for the results. |
|
|
552 | |
|
|
553 | You can also use them to simulate traditional event loops - for example, |
|
|
554 | you can block your main program until an event occurs - for example, you |
|
|
555 | could "->recv" in your main program until the user clicks the Quit |
|
|
556 | button of your app, which would "->send" the "quit" event. |
|
|
557 | |
|
|
558 | Note that condition variables recurse into the event loop - if you have |
|
|
559 | two pieces of code that call "->recv" in a round-robin fashion, you |
|
|
560 | lose. Therefore, condition variables are good to export to your caller, |
|
|
561 | but you should avoid making a blocking wait yourself, at least in |
|
|
562 | callbacks, as this asks for trouble. |
|
|
563 | |
|
|
564 | Condition variables are represented by hash refs in perl, and the keys |
|
|
565 | used by AnyEvent itself are all named "_ae_XXX" to make subclassing easy |
|
|
566 | (it is often useful to build your own transaction class on top of |
|
|
567 | AnyEvent). To subclass, use "AnyEvent::CondVar" as base class and call |
|
|
568 | it's "new" method in your own "new" method. |
|
|
569 | |
|
|
570 | There are two "sides" to a condition variable - the "producer side" |
|
|
571 | which eventually calls "-> send", and the "consumer side", which waits |
|
|
572 | for the send to occur. |
|
|
573 | |
|
|
574 | Example: wait for a timer. |
|
|
575 | |
|
|
576 | # wait till the result is ready |
|
|
577 | my $result_ready = AnyEvent->condvar; |
|
|
578 | |
|
|
579 | # do something such as adding a timer |
|
|
580 | # or socket watcher the calls $result_ready->send |
|
|
581 | # when the "result" is ready. |
|
|
582 | # in this case, we simply use a timer: |
|
|
583 | my $w = AnyEvent->timer ( |
|
|
584 | after => 1, |
|
|
585 | cb => sub { $result_ready->send }, |
|
|
586 | ); |
|
|
587 | |
|
|
588 | # this "blocks" (while handling events) till the callback |
|
|
589 | # calls ->send |
|
|
590 | $result_ready->recv; |
|
|
591 | |
|
|
592 | Example: wait for a timer, but take advantage of the fact that condition |
|
|
593 | variables are also callable directly. |
|
|
594 | |
|
|
595 | my $done = AnyEvent->condvar; |
|
|
596 | my $delay = AnyEvent->timer (after => 5, cb => $done); |
|
|
597 | $done->recv; |
|
|
598 | |
|
|
599 | Example: Imagine an API that returns a condvar and doesn't support |
|
|
600 | callbacks. This is how you make a synchronous call, for example from the |
|
|
601 | main program: |
|
|
602 | |
|
|
603 | use AnyEvent::CouchDB; |
|
|
604 | |
|
|
605 | ... |
|
|
606 | |
|
|
607 | my @info = $couchdb->info->recv; |
|
|
608 | |
|
|
609 | And this is how you would just set a callback to be called whenever the |
|
|
610 | results are available: |
|
|
611 | |
|
|
612 | $couchdb->info->cb (sub { |
|
|
613 | my @info = $_[0]->recv; |
|
|
614 | }); |
|
|
615 | |
|
|
616 | METHODS FOR PRODUCERS |
|
|
617 | These methods should only be used by the producing side, i.e. the |
|
|
618 | code/module that eventually sends the signal. Note that it is also the |
|
|
619 | producer side which creates the condvar in most cases, but it isn't |
|
|
620 | uncommon for the consumer to create it as well. |
|
|
621 | |
|
|
622 | $cv->send (...) |
|
|
623 | Flag the condition as ready - a running "->recv" and all further |
|
|
624 | calls to "recv" will (eventually) return after this method has been |
|
|
625 | called. If nobody is waiting the send will be remembered. |
|
|
626 | |
|
|
627 | If a callback has been set on the condition variable, it is called |
|
|
628 | immediately from within send. |
|
|
629 | |
|
|
630 | Any arguments passed to the "send" call will be returned by all |
|
|
631 | future "->recv" calls. |
|
|
632 | |
|
|
633 | Condition variables are overloaded so one can call them directly (as |
|
|
634 | if they were a code reference). Calling them directly is the same as |
|
|
635 | calling "send". |
|
|
636 | |
|
|
637 | $cv->croak ($error) |
|
|
638 | Similar to send, but causes all call's to "->recv" to invoke |
|
|
639 | "Carp::croak" with the given error message/object/scalar. |
|
|
640 | |
|
|
641 | This can be used to signal any errors to the condition variable |
|
|
642 | user/consumer. Doing it this way instead of calling "croak" directly |
|
|
643 | delays the error detetcion, but has the overwhelmign advantage that |
|
|
644 | it diagnoses the error at the place where the result is expected, |
|
|
645 | and not deep in some event clalback without connection to the actual |
|
|
646 | code causing the problem. |
|
|
647 | |
|
|
648 | $cv->begin ([group callback]) |
|
|
649 | $cv->end |
|
|
650 | These two methods can be used to combine many transactions/events |
|
|
651 | into one. For example, a function that pings many hosts in parallel |
|
|
652 | might want to use a condition variable for the whole process. |
|
|
653 | |
|
|
654 | Every call to "->begin" will increment a counter, and every call to |
|
|
655 | "->end" will decrement it. If the counter reaches 0 in "->end", the |
|
|
656 | (last) callback passed to "begin" will be executed, passing the |
|
|
657 | condvar as first argument. That callback is *supposed* to call |
|
|
658 | "->send", but that is not required. If no group callback was set, |
|
|
659 | "send" will be called without any arguments. |
|
|
660 | |
|
|
661 | You can think of "$cv->send" giving you an OR condition (one call |
|
|
662 | sends), while "$cv->begin" and "$cv->end" giving you an AND |
|
|
663 | condition (all "begin" calls must be "end"'ed before the condvar |
|
|
664 | sends). |
|
|
665 | |
|
|
666 | Let's start with a simple example: you have two I/O watchers (for |
|
|
667 | example, STDOUT and STDERR for a program), and you want to wait for |
|
|
668 | both streams to close before activating a condvar: |
|
|
669 | |
|
|
670 | my $cv = AnyEvent->condvar; |
|
|
671 | |
|
|
672 | $cv->begin; # first watcher |
|
|
673 | my $w1 = AnyEvent->io (fh => $fh1, cb => sub { |
|
|
674 | defined sysread $fh1, my $buf, 4096 |
|
|
675 | or $cv->end; |
|
|
676 | }); |
|
|
677 | |
|
|
678 | $cv->begin; # second watcher |
|
|
679 | my $w2 = AnyEvent->io (fh => $fh2, cb => sub { |
|
|
680 | defined sysread $fh2, my $buf, 4096 |
|
|
681 | or $cv->end; |
|
|
682 | }); |
|
|
683 | |
|
|
684 | $cv->recv; |
|
|
685 | |
|
|
686 | This works because for every event source (EOF on file handle), |
|
|
687 | there is one call to "begin", so the condvar waits for all calls to |
|
|
688 | "end" before sending. |
|
|
689 | |
|
|
690 | The ping example mentioned above is slightly more complicated, as |
|
|
691 | the there are results to be passwd back, and the number of tasks |
|
|
692 | that are begung can potentially be zero: |
|
|
693 | |
|
|
694 | my $cv = AnyEvent->condvar; |
|
|
695 | |
|
|
696 | my %result; |
|
|
697 | $cv->begin (sub { shift->send (\%result) }); |
|
|
698 | |
|
|
699 | for my $host (@list_of_hosts) { |
|
|
700 | $cv->begin; |
|
|
701 | ping_host_then_call_callback $host, sub { |
|
|
702 | $result{$host} = ...; |
|
|
703 | $cv->end; |
|
|
704 | }; |
|
|
705 | } |
|
|
706 | |
|
|
707 | $cv->end; |
|
|
708 | |
|
|
709 | This code fragment supposedly pings a number of hosts and calls |
|
|
710 | "send" after results for all then have have been gathered - in any |
|
|
711 | order. To achieve this, the code issues a call to "begin" when it |
|
|
712 | starts each ping request and calls "end" when it has received some |
|
|
713 | result for it. Since "begin" and "end" only maintain a counter, the |
|
|
714 | order in which results arrive is not relevant. |
|
|
715 | |
|
|
716 | There is an additional bracketing call to "begin" and "end" outside |
|
|
717 | the loop, which serves two important purposes: first, it sets the |
|
|
718 | callback to be called once the counter reaches 0, and second, it |
|
|
719 | ensures that "send" is called even when "no" hosts are being pinged |
|
|
720 | (the loop doesn't execute once). |
|
|
721 | |
|
|
722 | This is the general pattern when you "fan out" into multiple (but |
|
|
723 | potentially none) subrequests: use an outer "begin"/"end" pair to |
|
|
724 | set the callback and ensure "end" is called at least once, and then, |
|
|
725 | for each subrequest you start, call "begin" and for each subrequest |
|
|
726 | you finish, call "end". |
|
|
727 | |
|
|
728 | METHODS FOR CONSUMERS |
|
|
729 | These methods should only be used by the consuming side, i.e. the code |
|
|
730 | awaits the condition. |
|
|
731 | |
|
|
732 | $cv->recv |
|
|
733 | Wait (blocking if necessary) until the "->send" or "->croak" methods |
|
|
734 | have been called on c<$cv>, while servicing other watchers normally. |
|
|
735 | |
|
|
736 | You can only wait once on a condition - additional calls are valid |
|
|
737 | but will return immediately. |
|
|
738 | |
|
|
739 | If an error condition has been set by calling "->croak", then this |
|
|
740 | function will call "croak". |
|
|
741 | |
|
|
742 | In list context, all parameters passed to "send" will be returned, |
|
|
743 | in scalar context only the first one will be returned. |
|
|
744 | |
|
|
745 | Note that doing a blocking wait in a callback is not supported by |
|
|
746 | any event loop, that is, recursive invocation of a blocking "->recv" |
|
|
747 | is not allowed, and the "recv" call will "croak" if such a condition |
|
|
748 | is detected. This condition can be slightly loosened by using |
|
|
749 | Coro::AnyEvent, which allows you to do a blocking "->recv" from any |
|
|
750 | thread that doesn't run the event loop itself. |
|
|
751 | |
|
|
752 | Not all event models support a blocking wait - some die in that case |
|
|
753 | (programs might want to do that to stay interactive), so *if you are |
|
|
754 | using this from a module, never require a blocking wait*. Instead, |
|
|
755 | let the caller decide whether the call will block or not (for |
|
|
756 | example, by coupling condition variables with some kind of request |
|
|
757 | results and supporting callbacks so the caller knows that getting |
|
|
758 | the result will not block, while still supporting blocking waits if |
|
|
759 | the caller so desires). |
|
|
760 | |
|
|
761 | You can ensure that "-recv" never blocks by setting a callback and |
|
|
762 | only calling "->recv" from within that callback (or at a later |
|
|
763 | time). This will work even when the event loop does not support |
|
|
764 | blocking waits otherwise. |
|
|
765 | |
|
|
766 | $bool = $cv->ready |
|
|
767 | Returns true when the condition is "true", i.e. whether "send" or |
|
|
768 | "croak" have been called. |
|
|
769 | |
|
|
770 | $cb = $cv->cb ($cb->($cv)) |
|
|
771 | This is a mutator function that returns the callback set and |
|
|
772 | optionally replaces it before doing so. |
|
|
773 | |
|
|
774 | The callback will be called when the condition becomes (or already |
|
|
775 | was) "true", i.e. when "send" or "croak" are called (or were |
|
|
776 | called), with the only argument being the condition variable itself. |
|
|
777 | Calling "recv" inside the callback or at any later time is |
|
|
778 | guaranteed not to block. |
|
|
779 | |
|
|
780 | SUPPORTED EVENT LOOPS/BACKENDS |
|
|
781 | The available backend classes are (every class has its own manpage): |
|
|
782 | |
|
|
783 | Backends that are autoprobed when no other event loop can be found. |
|
|
784 | EV is the preferred backend when no other event loop seems to be in |
|
|
785 | use. If EV is not installed, then AnyEvent will fall back to its own |
|
|
786 | pure-perl implementation, which is available everywhere as it comes |
|
|
787 | with AnyEvent itself. |
|
|
788 | |
|
|
789 | AnyEvent::Impl::EV based on EV (interface to libev, best choice). |
|
|
790 | AnyEvent::Impl::Perl pure-perl implementation, fast and portable. |
|
|
791 | |
|
|
792 | Backends that are transparently being picked up when they are used. |
|
|
793 | These will be used when they are currently loaded when the first |
|
|
794 | watcher is created, in which case it is assumed that the application |
|
|
795 | is using them. This means that AnyEvent will automatically pick the |
|
|
796 | right backend when the main program loads an event module before |
|
|
797 | anything starts to create watchers. Nothing special needs to be done |
|
|
798 | by the main program. |
|
|
799 | |
|
|
800 | AnyEvent::Impl::Event based on Event, very stable, few glitches. |
|
|
801 | AnyEvent::Impl::Glib based on Glib, slow but very stable. |
|
|
802 | AnyEvent::Impl::Tk based on Tk, very broken. |
|
|
803 | AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse. |
|
|
804 | AnyEvent::Impl::POE based on POE, very slow, some limitations. |
|
|
805 | AnyEvent::Impl::Irssi used when running within irssi. |
|
|
806 | |
|
|
807 | Backends with special needs. |
|
|
808 | Qt requires the Qt::Application to be instantiated first, but will |
|
|
809 | otherwise be picked up automatically. As long as the main program |
|
|
810 | instantiates the application before any AnyEvent watchers are |
|
|
811 | created, everything should just work. |
|
|
812 | |
|
|
813 | AnyEvent::Impl::Qt based on Qt. |
|
|
814 | |
|
|
815 | Support for IO::Async can only be partial, as it is too broken and |
|
|
816 | architecturally limited to even support the AnyEvent API. It also is |
|
|
817 | the only event loop that needs the loop to be set explicitly, so it |
|
|
818 | can only be used by a main program knowing about AnyEvent. See |
|
|
819 | AnyEvent::Impl::Async for the gory details. |
|
|
820 | |
|
|
821 | AnyEvent::Impl::IOAsync based on IO::Async, cannot be autoprobed. |
|
|
822 | |
|
|
823 | Event loops that are indirectly supported via other backends. |
|
|
824 | Some event loops can be supported via other modules: |
|
|
825 | |
|
|
826 | There is no direct support for WxWidgets (Wx) or Prima. |
|
|
827 | |
|
|
828 | WxWidgets has no support for watching file handles. However, you can |
|
|
829 | use WxWidgets through the POE adaptor, as POE has a Wx backend that |
|
|
830 | simply polls 20 times per second, which was considered to be too |
|
|
831 | horrible to even consider for AnyEvent. |
|
|
832 | |
|
|
833 | Prima is not supported as nobody seems to be using it, but it has a |
|
|
834 | POE backend, so it can be supported through POE. |
|
|
835 | |
|
|
836 | AnyEvent knows about both Prima and Wx, however, and will try to |
|
|
837 | load POE when detecting them, in the hope that POE will pick them |
|
|
838 | up, in which case everything will be automatic. |
|
|
839 | |
|
|
840 | GLOBAL VARIABLES AND FUNCTIONS |
|
|
841 | These are not normally required to use AnyEvent, but can be useful to |
|
|
842 | write AnyEvent extension modules. |
|
|
843 | |
|
|
844 | $AnyEvent::MODEL |
|
|
845 | Contains "undef" until the first watcher is being created, before |
|
|
846 | the backend has been autodetected. |
|
|
847 | |
|
|
848 | Afterwards it contains the event model that is being used, which is |
|
|
849 | the name of the Perl class implementing the model. This class is |
|
|
850 | usually one of the "AnyEvent::Impl:xxx" modules, but can be any |
|
|
851 | other class in the case AnyEvent has been extended at runtime (e.g. |
|
|
852 | in *rxvt-unicode* it will be "urxvt::anyevent"). |
|
|
853 | |
|
|
854 | AnyEvent::detect |
|
|
855 | Returns $AnyEvent::MODEL, forcing autodetection of the event model |
|
|
856 | if necessary. You should only call this function right before you |
|
|
857 | would have created an AnyEvent watcher anyway, that is, as late as |
|
|
858 | possible at runtime, and not e.g. while initialising of your module. |
|
|
859 | |
|
|
860 | If you need to do some initialisation before AnyEvent watchers are |
|
|
861 | created, use "post_detect". |
|
|
862 | |
|
|
863 | $guard = AnyEvent::post_detect { BLOCK } |
|
|
864 | Arranges for the code block to be executed as soon as the event |
|
|
865 | model is autodetected (or immediately if this has already happened). |
|
|
866 | |
|
|
867 | The block will be executed *after* the actual backend has been |
|
|
868 | detected ($AnyEvent::MODEL is set), but *before* any watchers have |
|
|
869 | been created, so it is possible to e.g. patch @AnyEvent::ISA or do |
|
|
870 | other initialisations - see the sources of AnyEvent::Strict or |
|
|
871 | AnyEvent::AIO to see how this is used. |
|
|
872 | |
|
|
873 | The most common usage is to create some global watchers, without |
|
|
874 | forcing event module detection too early, for example, AnyEvent::AIO |
|
|
875 | creates and installs the global IO::AIO watcher in a "post_detect" |
|
|
876 | block to avoid autodetecting the event module at load time. |
|
|
877 | |
|
|
878 | If called in scalar or list context, then it creates and returns an |
|
|
879 | object that automatically removes the callback again when it is |
|
|
880 | destroyed (or "undef" when the hook was immediately executed). See |
|
|
881 | AnyEvent::AIO for a case where this is useful. |
|
|
882 | |
|
|
883 | Example: Create a watcher for the IO::AIO module and store it in |
|
|
884 | $WATCHER. Only do so after the event loop is initialised, though. |
|
|
885 | |
|
|
886 | our WATCHER; |
|
|
887 | |
|
|
888 | my $guard = AnyEvent::post_detect { |
|
|
889 | $WATCHER = AnyEvent->io (fh => IO::AIO::poll_fileno, poll => 'r', cb => \&IO::AIO::poll_cb); |
|
|
890 | }; |
|
|
891 | |
|
|
892 | # the ||= is important in case post_detect immediately runs the block, |
|
|
893 | # as to not clobber the newly-created watcher. assigning both watcher and |
|
|
894 | # post_detect guard to the same variable has the advantage of users being |
|
|
895 | # able to just C<undef $WATCHER> if the watcher causes them grief. |
|
|
896 | |
|
|
897 | $WATCHER ||= $guard; |
|
|
898 | |
|
|
899 | @AnyEvent::post_detect |
|
|
900 | If there are any code references in this array (you can "push" to it |
|
|
901 | before or after loading AnyEvent), then they will called directly |
|
|
902 | after the event loop has been chosen. |
|
|
903 | |
|
|
904 | You should check $AnyEvent::MODEL before adding to this array, |
|
|
905 | though: if it is defined then the event loop has already been |
|
|
906 | detected, and the array will be ignored. |
|
|
907 | |
|
|
908 | Best use "AnyEvent::post_detect { BLOCK }" when your application |
|
|
909 | allows it,as it takes care of these details. |
|
|
910 | |
|
|
911 | This variable is mainly useful for modules that can do something |
|
|
912 | useful when AnyEvent is used and thus want to know when it is |
|
|
913 | initialised, but do not need to even load it by default. This array |
|
|
914 | provides the means to hook into AnyEvent passively, without loading |
|
|
915 | it. |
|
|
916 | |
|
|
917 | WHAT TO DO IN A MODULE |
|
|
918 | As a module author, you should "use AnyEvent" and call AnyEvent methods |
|
|
919 | freely, but you should not load a specific event module or rely on it. |
|
|
920 | |
|
|
921 | Be careful when you create watchers in the module body - AnyEvent will |
|
|
922 | decide which event module to use as soon as the first method is called, |
|
|
923 | so by calling AnyEvent in your module body you force the user of your |
|
|
924 | module to load the event module first. |
|
|
925 | |
|
|
926 | Never call "->recv" on a condition variable unless you *know* that the |
|
|
927 | "->send" method has been called on it already. This is because it will |
|
|
928 | stall the whole program, and the whole point of using events is to stay |
|
|
929 | interactive. |
|
|
930 | |
|
|
931 | It is fine, however, to call "->recv" when the user of your module |
|
|
932 | requests it (i.e. if you create a http request object ad have a method |
|
|
933 | called "results" that returns the results, it should call "->recv" |
|
|
934 | freely, as the user of your module knows what she is doing. always). |
|
|
935 | |
|
|
936 | WHAT TO DO IN THE MAIN PROGRAM |
|
|
937 | There will always be a single main program - the only place that should |
|
|
938 | dictate which event model to use. |
|
|
939 | |
|
|
940 | If it doesn't care, it can just "use AnyEvent" and use it itself, or not |
|
|
941 | do anything special (it does not need to be event-based) and let |
|
|
942 | AnyEvent decide which implementation to chose if some module relies on |
|
|
943 | it. |
|
|
944 | |
|
|
945 | If the main program relies on a specific event model - for example, in |
|
|
946 | Gtk2 programs you have to rely on the Glib module - you should load the |
|
|
947 | event module before loading AnyEvent or any module that uses it: |
|
|
948 | generally speaking, you should load it as early as possible. The reason |
|
|
949 | is that modules might create watchers when they are loaded, and AnyEvent |
|
|
950 | will decide on the event model to use as soon as it creates watchers, |
|
|
951 | and it might chose the wrong one unless you load the correct one |
|
|
952 | yourself. |
|
|
953 | |
|
|
954 | You can chose to use a pure-perl implementation by loading the |
|
|
955 | "AnyEvent::Impl::Perl" module, which gives you similar behaviour |
|
|
956 | everywhere, but letting AnyEvent chose the model is generally better. |
|
|
957 | |
|
|
958 | MAINLOOP EMULATION |
|
|
959 | Sometimes (often for short test scripts, or even standalone programs who |
|
|
960 | only want to use AnyEvent), you do not want to run a specific event |
|
|
961 | loop. |
|
|
962 | |
|
|
963 | In that case, you can use a condition variable like this: |
|
|
964 | |
|
|
965 | AnyEvent->condvar->recv; |
|
|
966 | |
|
|
967 | This has the effect of entering the event loop and looping forever. |
|
|
968 | |
|
|
969 | Note that usually your program has some exit condition, in which case it |
|
|
970 | is better to use the "traditional" approach of storing a condition |
|
|
971 | variable somewhere, waiting for it, and sending it when the program |
|
|
972 | should exit cleanly. |
|
|
973 | |
|
|
974 | OTHER MODULES |
|
|
975 | The following is a non-exhaustive list of additional modules that use |
|
|
976 | AnyEvent as a client and can therefore be mixed easily with other |
|
|
977 | AnyEvent modules and other event loops in the same program. Some of the |
|
|
978 | modules come with AnyEvent, most are available via CPAN. |
|
|
979 | |
|
|
980 | AnyEvent::Util |
|
|
981 | Contains various utility functions that replace often-used but |
|
|
982 | blocking functions such as "inet_aton" by event-/callback-based |
|
|
983 | versions. |
|
|
984 | |
|
|
985 | AnyEvent::Socket |
|
|
986 | Provides various utility functions for (internet protocol) sockets, |
|
|
987 | addresses and name resolution. Also functions to create non-blocking |
|
|
988 | tcp connections or tcp servers, with IPv6 and SRV record support and |
|
|
989 | more. |
|
|
990 | |
|
|
991 | AnyEvent::Handle |
|
|
992 | Provide read and write buffers, manages watchers for reads and |
|
|
993 | writes, supports raw and formatted I/O, I/O queued and fully |
|
|
994 | transparent and non-blocking SSL/TLS (via AnyEvent::TLS. |
|
|
995 | |
|
|
996 | AnyEvent::DNS |
|
|
997 | Provides rich asynchronous DNS resolver capabilities. |
|
|
998 | |
|
|
999 | AnyEvent::HTTP |
|
|
1000 | A simple-to-use HTTP library that is capable of making a lot of |
|
|
1001 | concurrent HTTP requests. |
|
|
1002 | |
|
|
1003 | AnyEvent::HTTPD |
|
|
1004 | Provides a simple web application server framework. |
|
|
1005 | |
|
|
1006 | AnyEvent::FastPing |
|
|
1007 | The fastest ping in the west. |
|
|
1008 | |
|
|
1009 | AnyEvent::DBI |
|
|
1010 | Executes DBI requests asynchronously in a proxy process. |
|
|
1011 | |
|
|
1012 | AnyEvent::AIO |
|
|
1013 | Truly asynchronous I/O, should be in the toolbox of every event |
|
|
1014 | programmer. AnyEvent::AIO transparently fuses IO::AIO and AnyEvent |
|
|
1015 | together. |
|
|
1016 | |
|
|
1017 | AnyEvent::BDB |
|
|
1018 | Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently |
|
|
1019 | fuses BDB and AnyEvent together. |
|
|
1020 | |
|
|
1021 | AnyEvent::GPSD |
|
|
1022 | A non-blocking interface to gpsd, a daemon delivering GPS |
|
|
1023 | information. |
|
|
1024 | |
|
|
1025 | AnyEvent::IRC |
|
|
1026 | AnyEvent based IRC client module family (replacing the older |
|
|
1027 | Net::IRC3). |
|
|
1028 | |
|
|
1029 | AnyEvent::XMPP |
|
|
1030 | AnyEvent based XMPP (Jabber protocol) module family (replacing the |
|
|
1031 | older Net::XMPP2>. |
|
|
1032 | |
|
|
1033 | AnyEvent::IGS |
|
|
1034 | A non-blocking interface to the Internet Go Server protocol (used by |
|
|
1035 | App::IGS). |
|
|
1036 | |
|
|
1037 | Net::FCP |
|
|
1038 | AnyEvent-based implementation of the Freenet Client Protocol, |
|
|
1039 | birthplace of AnyEvent. |
|
|
1040 | |
|
|
1041 | Event::ExecFlow |
|
|
1042 | High level API for event-based execution flow control. |
|
|
1043 | |
|
|
1044 | Coro |
|
|
1045 | Has special support for AnyEvent via Coro::AnyEvent. |
|
|
1046 | |
|
|
1047 | SIMPLIFIED AE API |
|
|
1048 | Starting with version 5.0, AnyEvent officially supports a second, much |
|
|
1049 | simpler, API that is designed to reduce the calling, typing and memory |
|
|
1050 | overhead. |
|
|
1051 | |
|
|
1052 | See the AE manpage for details. |
|
|
1053 | |
|
|
1054 | ERROR AND EXCEPTION HANDLING |
|
|
1055 | In general, AnyEvent does not do any error handling - it relies on the |
|
|
1056 | caller to do that if required. The AnyEvent::Strict module (see also the |
|
|
1057 | "PERL_ANYEVENT_STRICT" environment variable, below) provides strict |
|
|
1058 | checking of all AnyEvent methods, however, which is highly useful during |
|
|
1059 | development. |
|
|
1060 | |
|
|
1061 | As for exception handling (i.e. runtime errors and exceptions thrown |
|
|
1062 | while executing a callback), this is not only highly event-loop |
|
|
1063 | specific, but also not in any way wrapped by this module, as this is the |
|
|
1064 | job of the main program. |
|
|
1065 | |
|
|
1066 | The pure perl event loop simply re-throws the exception (usually within |
|
|
1067 | "condvar->recv"), the Event and EV modules call "$Event/EV::DIED->()", |
|
|
1068 | Glib uses "install_exception_handler" and so on. |
|
|
1069 | |
|
|
1070 | ENVIRONMENT VARIABLES |
|
|
1071 | The following environment variables are used by this module or its |
|
|
1072 | submodules. |
|
|
1073 | |
|
|
1074 | Note that AnyEvent will remove *all* environment variables starting with |
|
|
1075 | "PERL_ANYEVENT_" from %ENV when it is loaded while taint mode is |
|
|
1076 | enabled. |
|
|
1077 | |
|
|
1078 | "PERL_ANYEVENT_VERBOSE" |
|
|
1079 | By default, AnyEvent will be completely silent except in fatal |
|
|
1080 | conditions. You can set this environment variable to make AnyEvent |
|
|
1081 | more talkative. |
|
|
1082 | |
|
|
1083 | When set to 1 or higher, causes AnyEvent to warn about unexpected |
|
|
1084 | conditions, such as not being able to load the event model specified |
|
|
1085 | by "PERL_ANYEVENT_MODEL". |
|
|
1086 | |
|
|
1087 | When set to 2 or higher, cause AnyEvent to report to STDERR which |
|
|
1088 | event model it chooses. |
|
|
1089 | |
|
|
1090 | When set to 8 or higher, then AnyEvent will report extra information |
|
|
1091 | on which optional modules it loads and how it implements certain |
|
|
1092 | features. |
|
|
1093 | |
|
|
1094 | "PERL_ANYEVENT_STRICT" |
|
|
1095 | AnyEvent does not do much argument checking by default, as thorough |
|
|
1096 | argument checking is very costly. Setting this variable to a true |
|
|
1097 | value will cause AnyEvent to load "AnyEvent::Strict" and then to |
|
|
1098 | thoroughly check the arguments passed to most method calls. If it |
|
|
1099 | finds any problems, it will croak. |
|
|
1100 | |
|
|
1101 | In other words, enables "strict" mode. |
|
|
1102 | |
|
|
1103 | Unlike "use strict" (or it's modern cousin, "use common::sense", it |
|
|
1104 | is definitely recommended to keep it off in production. Keeping |
|
|
1105 | "PERL_ANYEVENT_STRICT=1" in your environment while developing |
|
|
1106 | programs can be very useful, however. |
|
|
1107 | |
|
|
1108 | "PERL_ANYEVENT_MODEL" |
|
|
1109 | This can be used to specify the event model to be used by AnyEvent, |
|
|
1110 | before auto detection and -probing kicks in. It must be a string |
|
|
1111 | consisting entirely of ASCII letters. The string "AnyEvent::Impl::" |
|
|
1112 | gets prepended and the resulting module name is loaded and if the |
|
|
1113 | load was successful, used as event model. If it fails to load |
|
|
1114 | AnyEvent will proceed with auto detection and -probing. |
|
|
1115 | |
|
|
1116 | This functionality might change in future versions. |
|
|
1117 | |
|
|
1118 | For example, to force the pure perl model (AnyEvent::Impl::Perl) you |
|
|
1119 | could start your program like this: |
|
|
1120 | |
|
|
1121 | PERL_ANYEVENT_MODEL=Perl perl ... |
|
|
1122 | |
|
|
1123 | "PERL_ANYEVENT_PROTOCOLS" |
|
|
1124 | Used by both AnyEvent::DNS and AnyEvent::Socket to determine |
|
|
1125 | preferences for IPv4 or IPv6. The default is unspecified (and might |
|
|
1126 | change, or be the result of auto probing). |
|
|
1127 | |
|
|
1128 | Must be set to a comma-separated list of protocols or address |
|
|
1129 | families, current supported: "ipv4" and "ipv6". Only protocols |
|
|
1130 | mentioned will be used, and preference will be given to protocols |
|
|
1131 | mentioned earlier in the list. |
|
|
1132 | |
|
|
1133 | This variable can effectively be used for denial-of-service attacks |
|
|
1134 | against local programs (e.g. when setuid), although the impact is |
|
|
1135 | likely small, as the program has to handle conenction and other |
|
|
1136 | failures anyways. |
|
|
1137 | |
|
|
1138 | Examples: "PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6" - prefer IPv4 over |
|
|
1139 | IPv6, but support both and try to use both. |
|
|
1140 | "PERL_ANYEVENT_PROTOCOLS=ipv4" - only support IPv4, never try to |
|
|
1141 | resolve or contact IPv6 addresses. |
|
|
1142 | "PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4" support either IPv4 or IPv6, but |
|
|
1143 | prefer IPv6 over IPv4. |
|
|
1144 | |
|
|
1145 | "PERL_ANYEVENT_EDNS0" |
|
|
1146 | Used by AnyEvent::DNS to decide whether to use the EDNS0 extension |
|
|
1147 | for DNS. This extension is generally useful to reduce DNS traffic, |
|
|
1148 | but some (broken) firewalls drop such DNS packets, which is why it |
|
|
1149 | is off by default. |
|
|
1150 | |
|
|
1151 | Setting this variable to 1 will cause AnyEvent::DNS to announce |
|
|
1152 | EDNS0 in its DNS requests. |
|
|
1153 | |
|
|
1154 | "PERL_ANYEVENT_MAX_FORKS" |
|
|
1155 | The maximum number of child processes that |
|
|
1156 | "AnyEvent::Util::fork_call" will create in parallel. |
|
|
1157 | |
|
|
1158 | "PERL_ANYEVENT_MAX_OUTSTANDING_DNS" |
|
|
1159 | The default value for the "max_outstanding" parameter for the |
|
|
1160 | default DNS resolver - this is the maximum number of parallel DNS |
|
|
1161 | requests that are sent to the DNS server. |
|
|
1162 | |
|
|
1163 | "PERL_ANYEVENT_RESOLV_CONF" |
|
|
1164 | The file to use instead of /etc/resolv.conf (or OS-specific |
|
|
1165 | configuration) in the default resolver. When set to the empty |
|
|
1166 | string, no default config will be used. |
|
|
1167 | |
|
|
1168 | "PERL_ANYEVENT_CA_FILE", "PERL_ANYEVENT_CA_PATH". |
|
|
1169 | When neither "ca_file" nor "ca_path" was specified during |
|
|
1170 | AnyEvent::TLS context creation, and either of these environment |
|
|
1171 | variables exist, they will be used to specify CA certificate |
|
|
1172 | locations instead of a system-dependent default. |
|
|
1173 | |
|
|
1174 | "PERL_ANYEVENT_AVOID_GUARD" and "PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT" |
|
|
1175 | When these are set to 1, then the respective modules are not loaded. |
|
|
1176 | Mostly good for testing AnyEvent itself. |
|
|
1177 | |
|
|
1178 | SUPPLYING YOUR OWN EVENT MODEL INTERFACE |
|
|
1179 | This is an advanced topic that you do not normally need to use AnyEvent |
|
|
1180 | in a module. This section is only of use to event loop authors who want |
|
|
1181 | to provide AnyEvent compatibility. |
|
|
1182 | |
|
|
1183 | If you need to support another event library which isn't directly |
|
|
1184 | supported by AnyEvent, you can supply your own interface to it by |
|
|
1185 | pushing, before the first watcher gets created, the package name of the |
|
|
1186 | event module and the package name of the interface to use onto |
|
|
1187 | @AnyEvent::REGISTRY. You can do that before and even without loading |
|
|
1188 | AnyEvent, so it is reasonably cheap. |
|
|
1189 | |
|
|
1190 | Example: |
|
|
1191 | |
|
|
1192 | push @AnyEvent::REGISTRY, [urxvt => urxvt::anyevent::]; |
|
|
1193 | |
|
|
1194 | This tells AnyEvent to (literally) use the "urxvt::anyevent::" |
|
|
1195 | package/class when it finds the "urxvt" package/module is already |
|
|
1196 | loaded. |
|
|
1197 | |
|
|
1198 | When AnyEvent is loaded and asked to find a suitable event model, it |
|
|
1199 | will first check for the presence of urxvt by trying to "use" the |
|
|
1200 | "urxvt::anyevent" module. |
|
|
1201 | |
|
|
1202 | The class should provide implementations for all watcher types. See |
|
|
1203 | AnyEvent::Impl::EV (source code), AnyEvent::Impl::Glib (Source code) and |
|
|
1204 | so on for actual examples. Use "perldoc -m AnyEvent::Impl::Glib" to see |
|
|
1205 | the sources. |
|
|
1206 | |
|
|
1207 | If you don't provide "signal" and "child" watchers than AnyEvent will |
|
|
1208 | provide suitable (hopefully) replacements. |
|
|
1209 | |
|
|
1210 | The above example isn't fictitious, the *rxvt-unicode* (a.k.a. urxvt) |
|
|
1211 | terminal emulator uses the above line as-is. An interface isn't included |
|
|
1212 | in AnyEvent because it doesn't make sense outside the embedded |
|
|
1213 | interpreter inside *rxvt-unicode*, and it is updated and maintained as |
|
|
1214 | part of the *rxvt-unicode* distribution. |
|
|
1215 | |
|
|
1216 | *rxvt-unicode* also cheats a bit by not providing blocking access to |
|
|
1217 | condition variables: code blocking while waiting for a condition will |
|
|
1218 | "die". This still works with most modules/usages, and blocking calls |
|
|
1219 | must not be done in an interactive application, so it makes sense. |
|
|
1220 | |
|
|
1221 | EXAMPLE PROGRAM |
|
|
1222 | The following program uses an I/O watcher to read data from STDIN, a |
|
|
1223 | timer to display a message once per second, and a condition variable to |
|
|
1224 | quit the program when the user enters quit: |
|
|
1225 | |
|
|
1226 | use AnyEvent; |
|
|
1227 | |
|
|
1228 | my $cv = AnyEvent->condvar; |
|
|
1229 | |
|
|
1230 | my $io_watcher = AnyEvent->io ( |
|
|
1231 | fh => \*STDIN, |
|
|
1232 | poll => 'r', |
|
|
1233 | cb => sub { |
|
|
1234 | warn "io event <$_[0]>\n"; # will always output <r> |
|
|
1235 | chomp (my $input = <STDIN>); # read a line |
|
|
1236 | warn "read: $input\n"; # output what has been read |
|
|
1237 | $cv->send if $input =~ /^q/i; # quit program if /^q/i |
|
|
1238 | }, |
|
|
1239 | ); |
|
|
1240 | |
|
|
1241 | my $time_watcher = AnyEvent->timer (after => 1, interval => 1, cb => sub { |
|
|
1242 | warn "timeout\n"; # print 'timeout' at most every second |
|
|
1243 | }); |
|
|
1244 | |
|
|
1245 | $cv->recv; # wait until user enters /^q/i |
|
|
1246 | |
|
|
1247 | REAL-WORLD EXAMPLE |
|
|
1248 | Consider the Net::FCP module. It features (among others) the following |
|
|
1249 | API calls, which are to freenet what HTTP GET requests are to http: |
|
|
1250 | |
|
|
1251 | my $data = $fcp->client_get ($url); # blocks |
|
|
1252 | |
|
|
1253 | my $transaction = $fcp->txn_client_get ($url); # does not block |
|
|
1254 | $transaction->cb ( sub { ... } ); # set optional result callback |
|
|
1255 | my $data = $transaction->result; # possibly blocks |
|
|
1256 | |
|
|
1257 | The "client_get" method works like "LWP::Simple::get": it requests the |
|
|
1258 | given URL and waits till the data has arrived. It is defined to be: |
|
|
1259 | |
|
|
1260 | sub client_get { $_[0]->txn_client_get ($_[1])->result } |
|
|
1261 | |
|
|
1262 | And in fact is automatically generated. This is the blocking API of |
|
|
1263 | Net::FCP, and it works as simple as in any other, similar, module. |
|
|
1264 | |
|
|
1265 | More complicated is "txn_client_get": It only creates a transaction |
|
|
1266 | (completion, result, ...) object and initiates the transaction. |
|
|
1267 | |
|
|
1268 | my $txn = bless { }, Net::FCP::Txn::; |
|
|
1269 | |
|
|
1270 | It also creates a condition variable that is used to signal the |
|
|
1271 | completion of the request: |
|
|
1272 | |
|
|
1273 | $txn->{finished} = AnyAvent->condvar; |
|
|
1274 | |
|
|
1275 | It then creates a socket in non-blocking mode. |
|
|
1276 | |
|
|
1277 | socket $txn->{fh}, ...; |
|
|
1278 | fcntl $txn->{fh}, F_SETFL, O_NONBLOCK; |
|
|
1279 | connect $txn->{fh}, ... |
|
|
1280 | and !$!{EWOULDBLOCK} |
|
|
1281 | and !$!{EINPROGRESS} |
|
|
1282 | and Carp::croak "unable to connect: $!\n"; |
|
|
1283 | |
|
|
1284 | Then it creates a write-watcher which gets called whenever an error |
|
|
1285 | occurs or the connection succeeds: |
|
|
1286 | |
|
|
1287 | $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'w', cb => sub { $txn->fh_ready_w }); |
|
|
1288 | |
|
|
1289 | And returns this transaction object. The "fh_ready_w" callback gets |
|
|
1290 | called as soon as the event loop detects that the socket is ready for |
|
|
1291 | writing. |
|
|
1292 | |
|
|
1293 | The "fh_ready_w" method makes the socket blocking again, writes the |
|
|
1294 | request data and replaces the watcher by a read watcher (waiting for |
|
|
1295 | reply data). The actual code is more complicated, but that doesn't |
|
|
1296 | matter for this example: |
|
|
1297 | |
|
|
1298 | fcntl $txn->{fh}, F_SETFL, 0; |
|
|
1299 | syswrite $txn->{fh}, $txn->{request} |
|
|
1300 | or die "connection or write error"; |
|
|
1301 | $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r }); |
|
|
1302 | |
|
|
1303 | Again, "fh_ready_r" waits till all data has arrived, and then stores the |
|
|
1304 | result and signals any possible waiters that the request has finished: |
|
|
1305 | |
|
|
1306 | sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf}; |
|
|
1307 | |
|
|
1308 | if (end-of-file or data complete) { |
|
|
1309 | $txn->{result} = $txn->{buf}; |
|
|
1310 | $txn->{finished}->send; |
|
|
1311 | $txb->{cb}->($txn) of $txn->{cb}; # also call callback |
|
|
1312 | } |
|
|
1313 | |
|
|
1314 | The "result" method, finally, just waits for the finished signal (if the |
|
|
1315 | request was already finished, it doesn't wait, of course, and returns |
|
|
1316 | the data: |
|
|
1317 | |
|
|
1318 | $txn->{finished}->recv; |
|
|
1319 | return $txn->{result}; |
|
|
1320 | |
|
|
1321 | The actual code goes further and collects all errors ("die"s, |
|
|
1322 | exceptions) that occurred during request processing. The "result" method |
|
|
1323 | detects whether an exception as thrown (it is stored inside the $txn |
|
|
1324 | object) and just throws the exception, which means connection errors and |
|
|
1325 | other problems get reported tot he code that tries to use the result, |
|
|
1326 | not in a random callback. |
|
|
1327 | |
|
|
1328 | All of this enables the following usage styles: |
|
|
1329 | |
|
|
1330 | 1. Blocking: |
|
|
1331 | |
|
|
1332 | my $data = $fcp->client_get ($url); |
|
|
1333 | |
|
|
1334 | 2. Blocking, but running in parallel: |
|
|
1335 | |
|
|
1336 | my @datas = map $_->result, |
|
|
1337 | map $fcp->txn_client_get ($_), |
|
|
1338 | @urls; |
|
|
1339 | |
|
|
1340 | Both blocking examples work without the module user having to know |
|
|
1341 | anything about events. |
|
|
1342 | |
|
|
1343 | 3a. Event-based in a main program, using any supported event module: |
|
|
1344 | |
|
|
1345 | use EV; |
|
|
1346 | |
|
|
1347 | $fcp->txn_client_get ($url)->cb (sub { |
|
|
1348 | my $txn = shift; |
|
|
1349 | my $data = $txn->result; |
|
|
1350 | ... |
|
|
1351 | }); |
|
|
1352 | |
|
|
1353 | EV::loop; |
|
|
1354 | |
|
|
1355 | 3b. The module user could use AnyEvent, too: |
|
|
1356 | |
|
|
1357 | use AnyEvent; |
|
|
1358 | |
|
|
1359 | my $quit = AnyEvent->condvar; |
|
|
1360 | |
|
|
1361 | $fcp->txn_client_get ($url)->cb (sub { |
|
|
1362 | ... |
|
|
1363 | $quit->send; |
|
|
1364 | }); |
|
|
1365 | |
|
|
1366 | $quit->recv; |
|
|
1367 | |
|
|
1368 | BENCHMARKS |
|
|
1369 | To give you an idea of the performance and overheads that AnyEvent adds |
|
|
1370 | over the event loops themselves and to give you an impression of the |
|
|
1371 | speed of various event loops I prepared some benchmarks. |
|
|
1372 | |
|
|
1373 | BENCHMARKING ANYEVENT OVERHEAD |
|
|
1374 | Here is a benchmark of various supported event models used natively and |
|
|
1375 | through AnyEvent. The benchmark creates a lot of timers (with a zero |
|
|
1376 | timeout) and I/O watchers (watching STDOUT, a pty, to become writable, |
|
|
1377 | which it is), lets them fire exactly once and destroys them again. |
|
|
1378 | |
|
|
1379 | Source code for this benchmark is found as eg/bench in the AnyEvent |
|
|
1380 | distribution. It uses the AE interface, which makes a real difference |
|
|
1381 | for the EV and Perl backends only. |
|
|
1382 | |
|
|
1383 | Explanation of the columns |
|
|
1384 | *watcher* is the number of event watchers created/destroyed. Since |
|
|
1385 | different event models feature vastly different performances, each event |
|
|
1386 | loop was given a number of watchers so that overall runtime is |
|
|
1387 | acceptable and similar between tested event loop (and keep them from |
|
|
1388 | crashing): Glib would probably take thousands of years if asked to |
|
|
1389 | process the same number of watchers as EV in this benchmark. |
|
|
1390 | |
|
|
1391 | *bytes* is the number of bytes (as measured by the resident set size, |
|
|
1392 | RSS) consumed by each watcher. This method of measuring captures both C |
|
|
1393 | and Perl-based overheads. |
|
|
1394 | |
|
|
1395 | *create* is the time, in microseconds (millionths of seconds), that it |
|
|
1396 | takes to create a single watcher. The callback is a closure shared |
|
|
1397 | between all watchers, to avoid adding memory overhead. That means |
|
|
1398 | closure creation and memory usage is not included in the figures. |
|
|
1399 | |
|
|
1400 | *invoke* is the time, in microseconds, used to invoke a simple callback. |
|
|
1401 | The callback simply counts down a Perl variable and after it was invoked |
|
|
1402 | "watcher" times, it would "->send" a condvar once to signal the end of |
|
|
1403 | this phase. |
|
|
1404 | |
|
|
1405 | *destroy* is the time, in microseconds, that it takes to destroy a |
|
|
1406 | single watcher. |
|
|
1407 | |
|
|
1408 | Results |
|
|
1409 | name watchers bytes create invoke destroy comment |
|
|
1410 | EV/EV 100000 223 0.47 0.43 0.27 EV native interface |
|
|
1411 | EV/Any 100000 223 0.48 0.42 0.26 EV + AnyEvent watchers |
|
|
1412 | Coro::EV/Any 100000 223 0.47 0.42 0.26 coroutines + Coro::Signal |
|
|
1413 | Perl/Any 100000 431 2.70 0.74 0.92 pure perl implementation |
|
|
1414 | Event/Event 16000 516 31.16 31.84 0.82 Event native interface |
|
|
1415 | Event/Any 16000 1203 42.61 34.79 1.80 Event + AnyEvent watchers |
|
|
1416 | IOAsync/Any 16000 1911 41.92 27.45 16.81 via IO::Async::Loop::IO_Poll |
|
|
1417 | IOAsync/Any 16000 1726 40.69 26.37 15.25 via IO::Async::Loop::Epoll |
|
|
1418 | Glib/Any 16000 1118 89.00 12.57 51.17 quadratic behaviour |
|
|
1419 | Tk/Any 2000 1346 20.96 10.75 8.00 SEGV with >> 2000 watchers |
|
|
1420 | POE/Any 2000 6951 108.97 795.32 14.24 via POE::Loop::Event |
|
|
1421 | POE/Any 2000 6648 94.79 774.40 575.51 via POE::Loop::Select |
|
|
1422 | |
|
|
1423 | Discussion |
|
|
1424 | The benchmark does *not* measure scalability of the event loop very |
|
|
1425 | well. For example, a select-based event loop (such as the pure perl one) |
|
|
1426 | can never compete with an event loop that uses epoll when the number of |
|
|
1427 | file descriptors grows high. In this benchmark, all events become ready |
|
|
1428 | at the same time, so select/poll-based implementations get an unnatural |
|
|
1429 | speed boost. |
|
|
1430 | |
|
|
1431 | Also, note that the number of watchers usually has a nonlinear effect on |
|
|
1432 | overall speed, that is, creating twice as many watchers doesn't take |
|
|
1433 | twice the time - usually it takes longer. This puts event loops tested |
|
|
1434 | with a higher number of watchers at a disadvantage. |
|
|
1435 | |
|
|
1436 | To put the range of results into perspective, consider that on the |
|
|
1437 | benchmark machine, handling an event takes roughly 1600 CPU cycles with |
|
|
1438 | EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 |
|
|
1439 | CPU cycles with POE. |
|
|
1440 | |
|
|
1441 | "EV" is the sole leader regarding speed and memory use, which are both |
|
|
1442 | maximal/minimal, respectively. When using the AE API there is zero |
|
|
1443 | overhead (when going through the AnyEvent API create is about 5-6 times |
|
|
1444 | slower, with other times being equal, so still uses far less memory than |
|
|
1445 | any other event loop and is still faster than Event natively). |
|
|
1446 | |
|
|
1447 | The pure perl implementation is hit in a few sweet spots (both the |
|
|
1448 | constant timeout and the use of a single fd hit optimisations in the |
|
|
1449 | perl interpreter and the backend itself). Nevertheless this shows that |
|
|
1450 | it adds very little overhead in itself. Like any select-based backend |
|
|
1451 | its performance becomes really bad with lots of file descriptors (and |
|
|
1452 | few of them active), of course, but this was not subject of this |
|
|
1453 | benchmark. |
|
|
1454 | |
|
|
1455 | The "Event" module has a relatively high setup and callback invocation |
|
|
1456 | cost, but overall scores in on the third place. |
|
|
1457 | |
|
|
1458 | "IO::Async" performs admirably well, about on par with "Event", even |
|
|
1459 | when using its pure perl backend. |
|
|
1460 | |
|
|
1461 | "Glib"'s memory usage is quite a bit higher, but it features a faster |
|
|
1462 | callback invocation and overall ends up in the same class as "Event". |
|
|
1463 | However, Glib scales extremely badly, doubling the number of watchers |
|
|
1464 | increases the processing time by more than a factor of four, making it |
|
|
1465 | completely unusable when using larger numbers of watchers (note that |
|
|
1466 | only a single file descriptor was used in the benchmark, so |
|
|
1467 | inefficiencies of "poll" do not account for this). |
|
|
1468 | |
|
|
1469 | The "Tk" adaptor works relatively well. The fact that it crashes with |
|
|
1470 | more than 2000 watchers is a big setback, however, as correctness takes |
|
|
1471 | precedence over speed. Nevertheless, its performance is surprising, as |
|
|
1472 | the file descriptor is dup()ed for each watcher. This shows that the |
|
|
1473 | dup() employed by some adaptors is not a big performance issue (it does |
|
|
1474 | incur a hidden memory cost inside the kernel which is not reflected in |
|
|
1475 | the figures above). |
|
|
1476 | |
|
|
1477 | "POE", regardless of underlying event loop (whether using its pure perl |
|
|
1478 | select-based backend or the Event module, the POE-EV backend couldn't be |
|
|
1479 | tested because it wasn't working) shows abysmal performance and memory |
|
|
1480 | usage with AnyEvent: Watchers use almost 30 times as much memory as EV |
|
|
1481 | watchers, and 10 times as much memory as Event (the high memory |
|
|
1482 | requirements are caused by requiring a session for each watcher). |
|
|
1483 | Watcher invocation speed is almost 900 times slower than with AnyEvent's |
|
|
1484 | pure perl implementation. |
|
|
1485 | |
|
|
1486 | The design of the POE adaptor class in AnyEvent can not really account |
|
|
1487 | for the performance issues, though, as session creation overhead is |
|
|
1488 | small compared to execution of the state machine, which is coded pretty |
|
|
1489 | optimally within AnyEvent::Impl::POE (and while everybody agrees that |
|
|
1490 | using multiple sessions is not a good approach, especially regarding |
|
|
1491 | memory usage, even the author of POE could not come up with a faster |
|
|
1492 | design). |
|
|
1493 | |
|
|
1494 | Summary |
|
|
1495 | * Using EV through AnyEvent is faster than any other event loop (even |
|
|
1496 | when used without AnyEvent), but most event loops have acceptable |
|
|
1497 | performance with or without AnyEvent. |
|
|
1498 | |
|
|
1499 | * The overhead AnyEvent adds is usually much smaller than the overhead |
|
|
1500 | of the actual event loop, only with extremely fast event loops such |
|
|
1501 | as EV adds AnyEvent significant overhead. |
|
|
1502 | |
|
|
1503 | * You should avoid POE like the plague if you want performance or |
|
|
1504 | reasonable memory usage. |
|
|
1505 | |
|
|
1506 | BENCHMARKING THE LARGE SERVER CASE |
|
|
1507 | This benchmark actually benchmarks the event loop itself. It works by |
|
|
1508 | creating a number of "servers": each server consists of a socket pair, a |
|
|
1509 | timeout watcher that gets reset on activity (but never fires), and an |
|
|
1510 | I/O watcher waiting for input on one side of the socket. Each time the |
|
|
1511 | socket watcher reads a byte it will write that byte to a random other |
|
|
1512 | "server". |
|
|
1513 | |
|
|
1514 | The effect is that there will be a lot of I/O watchers, only part of |
|
|
1515 | which are active at any one point (so there is a constant number of |
|
|
1516 | active fds for each loop iteration, but which fds these are is random). |
|
|
1517 | The timeout is reset each time something is read because that reflects |
|
|
1518 | how most timeouts work (and puts extra pressure on the event loops). |
|
|
1519 | |
|
|
1520 | In this benchmark, we use 10000 socket pairs (20000 sockets), of which |
|
|
1521 | 100 (1%) are active. This mirrors the activity of large servers with |
|
|
1522 | many connections, most of which are idle at any one point in time. |
|
|
1523 | |
|
|
1524 | Source code for this benchmark is found as eg/bench2 in the AnyEvent |
|
|
1525 | distribution. It uses the AE interface, which makes a real difference |
|
|
1526 | for the EV and Perl backends only. |
|
|
1527 | |
|
|
1528 | Explanation of the columns |
|
|
1529 | *sockets* is the number of sockets, and twice the number of "servers" |
|
|
1530 | (as each server has a read and write socket end). |
|
|
1531 | |
|
|
1532 | *create* is the time it takes to create a socket pair (which is |
|
|
1533 | nontrivial) and two watchers: an I/O watcher and a timeout watcher. |
|
|
1534 | |
|
|
1535 | *request*, the most important value, is the time it takes to handle a |
|
|
1536 | single "request", that is, reading the token from the pipe and |
|
|
1537 | forwarding it to another server. This includes deleting the old timeout |
|
|
1538 | and creating a new one that moves the timeout into the future. |
|
|
1539 | |
|
|
1540 | Results |
|
|
1541 | name sockets create request |
|
|
1542 | EV 20000 62.66 7.99 |
|
|
1543 | Perl 20000 68.32 32.64 |
|
|
1544 | IOAsync 20000 174.06 101.15 epoll |
|
|
1545 | IOAsync 20000 174.67 610.84 poll |
|
|
1546 | Event 20000 202.69 242.91 |
|
|
1547 | Glib 20000 557.01 1689.52 |
|
|
1548 | POE 20000 341.54 12086.32 uses POE::Loop::Event |
|
|
1549 | |
|
|
1550 | Discussion |
|
|
1551 | This benchmark *does* measure scalability and overall performance of the |
|
|
1552 | particular event loop. |
|
|
1553 | |
|
|
1554 | EV is again fastest. Since it is using epoll on my system, the setup |
|
|
1555 | time is relatively high, though. |
|
|
1556 | |
|
|
1557 | Perl surprisingly comes second. It is much faster than the C-based event |
|
|
1558 | loops Event and Glib. |
|
|
1559 | |
|
|
1560 | IO::Async performs very well when using its epoll backend, and still |
|
|
1561 | quite good compared to Glib when using its pure perl backend. |
|
|
1562 | |
|
|
1563 | Event suffers from high setup time as well (look at its code and you |
|
|
1564 | will understand why). Callback invocation also has a high overhead |
|
|
1565 | compared to the "$_->() for .."-style loop that the Perl event loop |
|
|
1566 | uses. Event uses select or poll in basically all documented |
|
|
1567 | configurations. |
|
|
1568 | |
|
|
1569 | Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It |
|
|
1570 | clearly fails to perform with many filehandles or in busy servers. |
|
|
1571 | |
|
|
1572 | POE is still completely out of the picture, taking over 1000 times as |
|
|
1573 | long as EV, and over 100 times as long as the Perl implementation, even |
|
|
1574 | though it uses a C-based event loop in this case. |
|
|
1575 | |
|
|
1576 | Summary |
|
|
1577 | * The pure perl implementation performs extremely well. |
|
|
1578 | |
|
|
1579 | * Avoid Glib or POE in large projects where performance matters. |
|
|
1580 | |
|
|
1581 | BENCHMARKING SMALL SERVERS |
|
|
1582 | While event loops should scale (and select-based ones do not...) even to |
|
|
1583 | large servers, most programs we (or I :) actually write have only a few |
|
|
1584 | I/O watchers. |
|
|
1585 | |
|
|
1586 | In this benchmark, I use the same benchmark program as in the large |
|
|
1587 | server case, but it uses only eight "servers", of which three are active |
|
|
1588 | at any one time. This should reflect performance for a small server |
|
|
1589 | relatively well. |
|
|
1590 | |
|
|
1591 | The columns are identical to the previous table. |
|
|
1592 | |
|
|
1593 | Results |
|
|
1594 | name sockets create request |
|
|
1595 | EV 16 20.00 6.54 |
|
|
1596 | Perl 16 25.75 12.62 |
|
|
1597 | Event 16 81.27 35.86 |
|
|
1598 | Glib 16 32.63 15.48 |
|
|
1599 | POE 16 261.87 276.28 uses POE::Loop::Event |
|
|
1600 | |
|
|
1601 | Discussion |
|
|
1602 | The benchmark tries to test the performance of a typical small server. |
|
|
1603 | While knowing how various event loops perform is interesting, keep in |
|
|
1604 | mind that their overhead in this case is usually not as important, due |
|
|
1605 | to the small absolute number of watchers (that is, you need efficiency |
|
|
1606 | and speed most when you have lots of watchers, not when you only have a |
|
|
1607 | few of them). |
|
|
1608 | |
|
|
1609 | EV is again fastest. |
|
|
1610 | |
|
|
1611 | Perl again comes second. It is noticeably faster than the C-based event |
|
|
1612 | loops Event and Glib, although the difference is too small to really |
|
|
1613 | matter. |
|
|
1614 | |
|
|
1615 | POE also performs much better in this case, but is is still far behind |
|
|
1616 | the others. |
|
|
1617 | |
|
|
1618 | Summary |
|
|
1619 | * C-based event loops perform very well with small number of watchers, |
|
|
1620 | as the management overhead dominates. |
|
|
1621 | |
|
|
1622 | THE IO::Lambda BENCHMARK |
|
|
1623 | Recently I was told about the benchmark in the IO::Lambda manpage, which |
|
|
1624 | could be misinterpreted to make AnyEvent look bad. In fact, the |
|
|
1625 | benchmark simply compares IO::Lambda with POE, and IO::Lambda looks |
|
|
1626 | better (which shouldn't come as a surprise to anybody). As such, the |
|
|
1627 | benchmark is fine, and mostly shows that the AnyEvent backend from |
|
|
1628 | IO::Lambda isn't very optimal. But how would AnyEvent compare when used |
|
|
1629 | without the extra baggage? To explore this, I wrote the equivalent |
|
|
1630 | benchmark for AnyEvent. |
|
|
1631 | |
|
|
1632 | The benchmark itself creates an echo-server, and then, for 500 times, |
|
|
1633 | connects to the echo server, sends a line, waits for the reply, and then |
|
|
1634 | creates the next connection. This is a rather bad benchmark, as it |
|
|
1635 | doesn't test the efficiency of the framework or much non-blocking I/O, |
|
|
1636 | but it is a benchmark nevertheless. |
|
|
1637 | |
|
|
1638 | name runtime |
|
|
1639 | Lambda/select 0.330 sec |
|
|
1640 | + optimized 0.122 sec |
|
|
1641 | Lambda/AnyEvent 0.327 sec |
|
|
1642 | + optimized 0.138 sec |
|
|
1643 | Raw sockets/select 0.077 sec |
|
|
1644 | POE/select, components 0.662 sec |
|
|
1645 | POE/select, raw sockets 0.226 sec |
|
|
1646 | POE/select, optimized 0.404 sec |
|
|
1647 | |
|
|
1648 | AnyEvent/select/nb 0.085 sec |
|
|
1649 | AnyEvent/EV/nb 0.068 sec |
|
|
1650 | +state machine 0.134 sec |
|
|
1651 | |
|
|
1652 | The benchmark is also a bit unfair (my fault): the IO::Lambda/POE |
|
|
1653 | benchmarks actually make blocking connects and use 100% blocking I/O, |
|
|
1654 | defeating the purpose of an event-based solution. All of the newly |
|
|
1655 | written AnyEvent benchmarks use 100% non-blocking connects (using |
|
|
1656 | AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS |
|
|
1657 | resolver), so AnyEvent is at a disadvantage here, as non-blocking |
|
|
1658 | connects generally require a lot more bookkeeping and event handling |
|
|
1659 | than blocking connects (which involve a single syscall only). |
|
|
1660 | |
|
|
1661 | The last AnyEvent benchmark additionally uses AnyEvent::Handle, which |
|
|
1662 | offers similar expressive power as POE and IO::Lambda, using |
|
|
1663 | conventional Perl syntax. This means that both the echo server and the |
|
|
1664 | client are 100% non-blocking, further placing it at a disadvantage. |
|
|
1665 | |
|
|
1666 | As you can see, the AnyEvent + EV combination even beats the |
|
|
1667 | hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl |
|
|
1668 | backend easily beats IO::Lambda and POE. |
|
|
1669 | |
|
|
1670 | And even the 100% non-blocking version written using the high-level (and |
|
|
1671 | slow :) AnyEvent::Handle abstraction beats both POE and IO::Lambda |
|
|
1672 | higher level ("unoptimised") abstractions by a large margin, even though |
|
|
1673 | it does all of DNS, tcp-connect and socket I/O in a non-blocking way. |
|
|
1674 | |
|
|
1675 | The two AnyEvent benchmarks programs can be found as eg/ae0.pl and |
|
|
1676 | eg/ae2.pl in the AnyEvent distribution, the remaining benchmarks are |
|
|
1677 | part of the IO::Lambda distribution and were used without any changes. |
|
|
1678 | |
|
|
1679 | SIGNALS |
|
|
1680 | AnyEvent currently installs handlers for these signals: |
|
|
1681 | |
|
|
1682 | SIGCHLD |
|
|
1683 | A handler for "SIGCHLD" is installed by AnyEvent's child watcher |
|
|
1684 | emulation for event loops that do not support them natively. Also, |
|
|
1685 | some event loops install a similar handler. |
|
|
1686 | |
|
|
1687 | Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE, |
|
|
1688 | then AnyEvent will reset it to default, to avoid losing child exit |
|
|
1689 | statuses. |
|
|
1690 | |
|
|
1691 | SIGPIPE |
|
|
1692 | A no-op handler is installed for "SIGPIPE" when $SIG{PIPE} is |
|
|
1693 | "undef" when AnyEvent gets loaded. |
|
|
1694 | |
|
|
1695 | The rationale for this is that AnyEvent users usually do not really |
|
|
1696 | depend on SIGPIPE delivery (which is purely an optimisation for |
|
|
1697 | shell use, or badly-written programs), but "SIGPIPE" can cause |
|
|
1698 | spurious and rare program exits as a lot of people do not expect |
|
|
1699 | "SIGPIPE" when writing to some random socket. |
|
|
1700 | |
|
|
1701 | The rationale for installing a no-op handler as opposed to ignoring |
|
|
1702 | it is that this way, the handler will be restored to defaults on |
|
|
1703 | exec. |
|
|
1704 | |
|
|
1705 | Feel free to install your own handler, or reset it to defaults. |
|
|
1706 | |
|
|
1707 | RECOMMENDED/OPTIONAL MODULES |
|
|
1708 | One of AnyEvent's main goals is to be 100% Pure-Perl(tm): only perl (and |
|
|
1709 | it's built-in modules) are required to use it. |
|
|
1710 | |
|
|
1711 | That does not mean that AnyEvent won't take advantage of some additional |
|
|
1712 | modules if they are installed. |
|
|
1713 | |
|
|
1714 | This section epxlains which additional modules will be used, and how |
|
|
1715 | they affect AnyEvent's operetion. |
|
|
1716 | |
|
|
1717 | Async::Interrupt |
|
|
1718 | This slightly arcane module is used to implement fast signal |
|
|
1719 | handling: To my knowledge, there is no way to do completely |
|
|
1720 | race-free and quick signal handling in pure perl. To ensure that |
|
|
1721 | signals still get delivered, AnyEvent will start an interval timer |
|
|
1722 | to wake up perl (and catch the signals) with some delay (default is |
|
|
1723 | 10 seconds, look for $AnyEvent::MAX_SIGNAL_LATENCY). |
|
|
1724 | |
|
|
1725 | If this module is available, then it will be used to implement |
|
|
1726 | signal catching, which means that signals will not be delayed, and |
|
|
1727 | the event loop will not be interrupted regularly, which is more |
|
|
1728 | efficient (And good for battery life on laptops). |
|
|
1729 | |
|
|
1730 | This affects not just the pure-perl event loop, but also other event |
|
|
1731 | loops that have no signal handling on their own (e.g. Glib, Tk, Qt). |
|
|
1732 | |
|
|
1733 | Some event loops (POE, Event, Event::Lib) offer signal watchers |
|
|
1734 | natively, and either employ their own workarounds (POE) or use |
|
|
1735 | AnyEvent's workaround (using $AnyEvent::MAX_SIGNAL_LATENCY). |
|
|
1736 | Installing Async::Interrupt does nothing for those backends. |
|
|
1737 | |
|
|
1738 | EV This module isn't really "optional", as it is simply one of the |
|
|
1739 | backend event loops that AnyEvent can use. However, it is simply the |
|
|
1740 | best event loop available in terms of features, speed and stability: |
|
|
1741 | It supports the AnyEvent API optimally, implements all the watcher |
|
|
1742 | types in XS, does automatic timer adjustments even when no monotonic |
|
|
1743 | clock is available, can take avdantage of advanced kernel interfaces |
|
|
1744 | such as "epoll" and "kqueue", and is the fastest backend *by far*. |
|
|
1745 | You can even embed Glib/Gtk2 in it (or vice versa, see EV::Glib and |
|
|
1746 | Glib::EV). |
|
|
1747 | |
|
|
1748 | Guard |
|
|
1749 | The guard module, when used, will be used to implement |
|
|
1750 | "AnyEvent::Util::guard". This speeds up guards considerably (and |
|
|
1751 | uses a lot less memory), but otherwise doesn't affect guard |
|
|
1752 | operation much. It is purely used for performance. |
|
|
1753 | |
|
|
1754 | JSON and JSON::XS |
|
|
1755 | One of these modules is required when you want to read or write JSON |
|
|
1756 | data via AnyEvent::Handle. It is also written in pure-perl, but can |
|
|
1757 | take advantage of the ultra-high-speed JSON::XS module when it is |
|
|
1758 | installed. |
|
|
1759 | |
|
|
1760 | In fact, AnyEvent::Handle will use JSON::XS by default if it is |
|
|
1761 | installed. |
|
|
1762 | |
|
|
1763 | Net::SSLeay |
|
|
1764 | Implementing TLS/SSL in Perl is certainly interesting, but not very |
|
|
1765 | worthwhile: If this module is installed, then AnyEvent::Handle (with |
|
|
1766 | the help of AnyEvent::TLS), gains the ability to do TLS/SSL. |
|
|
1767 | |
|
|
1768 | Time::HiRes |
|
|
1769 | This module is part of perl since release 5.008. It will be used |
|
|
1770 | when the chosen event library does not come with a timing source on |
|
|
1771 | it's own. The pure-perl event loop (AnyEvent::Impl::Perl) will |
|
|
1772 | additionally use it to try to use a monotonic clock for timing |
|
|
1773 | stability. |
|
|
1774 | |
|
|
1775 | FORK |
|
|
1776 | Most event libraries are not fork-safe. The ones who are usually are |
|
|
1777 | because they rely on inefficient but fork-safe "select" or "poll" calls. |
|
|
1778 | Only EV is fully fork-aware. |
|
|
1779 | |
|
|
1780 | If you have to fork, you must either do so *before* creating your first |
|
|
1781 | watcher OR you must not use AnyEvent at all in the child OR you must do |
|
|
1782 | something completely out of the scope of AnyEvent. |
|
|
1783 | |
|
|
1784 | SECURITY CONSIDERATIONS |
|
|
1785 | AnyEvent can be forced to load any event model via |
|
|
1786 | $ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used |
|
|
1787 | to execute arbitrary code or directly gain access, it can easily be used |
|
|
1788 | to make the program hang or malfunction in subtle ways, as AnyEvent |
|
|
1789 | watchers will not be active when the program uses a different event |
|
|
1790 | model than specified in the variable. |
|
|
1791 | |
|
|
1792 | You can make AnyEvent completely ignore this variable by deleting it |
|
|
1793 | before the first watcher gets created, e.g. with a "BEGIN" block: |
|
|
1794 | |
|
|
1795 | BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} } |
|
|
1796 | |
|
|
1797 | use AnyEvent; |
|
|
1798 | |
|
|
1799 | Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can |
|
|
1800 | be used to probe what backend is used and gain other information (which |
|
|
1801 | is probably even less useful to an attacker than PERL_ANYEVENT_MODEL), |
|
|
1802 | and $ENV{PERL_ANYEVENT_STRICT}. |
|
|
1803 | |
|
|
1804 | Note that AnyEvent will remove *all* environment variables starting with |
|
|
1805 | "PERL_ANYEVENT_" from %ENV when it is loaded while taint mode is |
|
|
1806 | enabled. |
|
|
1807 | |
|
|
1808 | BUGS |
|
|
1809 | Perl 5.8 has numerous memleaks that sometimes hit this module and are |
|
|
1810 | hard to work around. If you suffer from memleaks, first upgrade to Perl |
|
|
1811 | 5.10 and check wether the leaks still show up. (Perl 5.10.0 has other |
|
|
1812 | annoying memleaks, such as leaking on "map" and "grep" but it is usually |
|
|
1813 | not as pronounced). |
|
|
1814 | |
|
|
1815 | SEE ALSO |
|
|
1816 | Utility functions: AnyEvent::Util. |
|
|
1817 | |
|
|
1818 | Event modules: EV, EV::Glib, Glib::EV, Event, Glib::Event, Glib, Tk, |
|
|
1819 | Event::Lib, Qt, POE. |
|
|
1820 | |
|
|
1821 | Implementations: AnyEvent::Impl::EV, AnyEvent::Impl::Event, |
|
|
1822 | AnyEvent::Impl::Glib, AnyEvent::Impl::Tk, AnyEvent::Impl::Perl, |
|
|
1823 | AnyEvent::Impl::EventLib, AnyEvent::Impl::Qt, AnyEvent::Impl::POE, |
|
|
1824 | AnyEvent::Impl::IOAsync, Anyevent::Impl::Irssi. |
|
|
1825 | |
|
|
1826 | Non-blocking file handles, sockets, TCP clients and servers: |
|
|
1827 | AnyEvent::Handle, AnyEvent::Socket, AnyEvent::TLS. |
|
|
1828 | |
|
|
1829 | Asynchronous DNS: AnyEvent::DNS. |
|
|
1830 | |
|
|
1831 | Coroutine support: Coro, Coro::AnyEvent, Coro::EV, Coro::Event, |
|
|
1832 | |
|
|
1833 | Nontrivial usage examples: AnyEvent::GPSD, AnyEvent::XMPP, |
|
|
1834 | AnyEvent::HTTP. |
|
|
1835 | |
|
|
1836 | AUTHOR |
|
|
1837 | Marc Lehmann <schmorp@schmorp.de> |
|
|
1838 | http://home.schmorp.de/ |
|
|
1839 | |