1 | NAME |
1 | NAME |
2 | AnyEvent - provide framework for multiple event loops |
2 | AnyEvent - provide framework for multiple event loops |
3 | |
3 | |
4 | EV, Event, Coro::EV, Coro::Event, Glib, Tk, Perl - various supported |
4 | EV, Event, Glib, Tk, Perl, Event::Lib, Qt, POE - various supported event |
5 | event loops |
5 | loops |
6 | |
6 | |
7 | SYNOPSIS |
7 | SYNOPSIS |
8 | use AnyEvent; |
8 | use AnyEvent; |
9 | |
9 | |
10 | my $w = AnyEvent->io (fh => $fh, poll => "r|w", cb => sub { |
10 | my $w = AnyEvent->io (fh => $fh, poll => "r|w", cb => sub { |
… | |
… | |
14 | my $w = AnyEvent->timer (after => $seconds, cb => sub { |
14 | my $w = AnyEvent->timer (after => $seconds, cb => sub { |
15 | ... |
15 | ... |
16 | }); |
16 | }); |
17 | |
17 | |
18 | my $w = AnyEvent->condvar; # stores whether a condition was flagged |
18 | my $w = AnyEvent->condvar; # stores whether a condition was flagged |
|
|
19 | $w->send; # wake up current and all future recv's |
19 | $w->wait; # enters "main loop" till $condvar gets ->broadcast |
20 | $w->recv; # enters "main loop" till $condvar gets ->send |
20 | $w->broadcast; # wake up current and all future wait's |
|
|
21 | |
21 | |
22 | WHY YOU SHOULD USE THIS MODULE (OR NOT) |
22 | WHY YOU SHOULD USE THIS MODULE (OR NOT) |
23 | Glib, POE, IO::Async, Event... CPAN offers event models by the dozen |
23 | Glib, POE, IO::Async, Event... CPAN offers event models by the dozen |
24 | nowadays. So what is different about AnyEvent? |
24 | nowadays. So what is different about AnyEvent? |
25 | |
25 | |
… | |
… | |
75 | The interface itself is vaguely similar, but not identical to the Event |
75 | The interface itself is vaguely similar, but not identical to the Event |
76 | module. |
76 | module. |
77 | |
77 | |
78 | During the first call of any watcher-creation method, the module tries |
78 | During the first call of any watcher-creation method, the module tries |
79 | to detect the currently loaded event loop by probing whether one of the |
79 | to detect the currently loaded event loop by probing whether one of the |
80 | following modules is already loaded: Coro::EV, Coro::Event, EV, Event, |
80 | following modules is already loaded: EV, Event, Glib, |
81 | Glib, Tk. The first one found is used. If none are found, the module |
81 | AnyEvent::Impl::Perl, Tk, Event::Lib, Qt, POE. The first one found is |
82 | tries to load these modules in the stated order. The first one that can |
82 | used. If none are found, the module tries to load these modules |
|
|
83 | (excluding Tk, Event::Lib, Qt and POE as the pure perl adaptor should |
|
|
84 | always succeed) in the order given. The first one that can be |
83 | be successfully loaded will be used. If, after this, still none could be |
85 | successfully loaded will be used. If, after this, still none could be |
84 | found, AnyEvent will fall back to a pure-perl event loop, which is not |
86 | found, AnyEvent will fall back to a pure-perl event loop, which is not |
85 | very efficient, but should work everywhere. |
87 | very efficient, but should work everywhere. |
86 | |
88 | |
87 | Because AnyEvent first checks for modules that are already loaded, |
89 | Because AnyEvent first checks for modules that are already loaded, |
88 | loading an event model explicitly before first using AnyEvent will |
90 | loading an event model explicitly before first using AnyEvent will |
… | |
… | |
129 | |
131 | |
130 | Note that "my $w; $w =" combination. This is necessary because in Perl, |
132 | Note that "my $w; $w =" combination. This is necessary because in Perl, |
131 | my variables are only visible after the statement in which they are |
133 | my variables are only visible after the statement in which they are |
132 | declared. |
134 | declared. |
133 | |
135 | |
134 | IO WATCHERS |
136 | I/O WATCHERS |
135 | You can create an I/O watcher by calling the "AnyEvent->io" method with |
137 | You can create an I/O watcher by calling the "AnyEvent->io" method with |
136 | the following mandatory key-value pairs as arguments: |
138 | the following mandatory key-value pairs as arguments: |
137 | |
139 | |
138 | "fh" the Perl *file handle* (*not* file descriptor) to watch for events. |
140 | "fh" the Perl *file handle* (*not* file descriptor) to watch for events. |
139 | "poll" must be a string that is either "r" or "w", which creates a |
141 | "poll" must be a string that is either "r" or "w", which creates a |
140 | watcher waiting for "r"eadable or "w"ritable events, respectively. "cb" |
142 | watcher waiting for "r"eadable or "w"ritable events, respectively. "cb" |
141 | is the callback to invoke each time the file handle becomes ready. |
143 | is the callback to invoke each time the file handle becomes ready. |
142 | |
144 | |
143 | File handles will be kept alive, so as long as the watcher exists, the |
145 | Although the callback might get passed parameters, their value and |
144 | file handle exists, too. |
146 | presence is undefined and you cannot rely on them. Portable AnyEvent |
|
|
147 | callbacks cannot use arguments passed to I/O watcher callbacks. |
145 | |
148 | |
|
|
149 | The I/O watcher might use the underlying file descriptor or a copy of |
146 | It is not allowed to close a file handle as long as any watcher is |
150 | it. You must not close a file handle as long as any watcher is active on |
147 | active on the underlying file descriptor. |
151 | the underlying file descriptor. |
148 | |
152 | |
149 | Some event loops issue spurious readyness notifications, so you should |
153 | Some event loops issue spurious readyness notifications, so you should |
150 | always use non-blocking calls when reading/writing from/to your file |
154 | always use non-blocking calls when reading/writing from/to your file |
151 | handles. |
155 | handles. |
152 | |
156 | |
… | |
… | |
162 | TIME WATCHERS |
166 | TIME WATCHERS |
163 | You can create a time watcher by calling the "AnyEvent->timer" method |
167 | You can create a time watcher by calling the "AnyEvent->timer" method |
164 | with the following mandatory arguments: |
168 | with the following mandatory arguments: |
165 | |
169 | |
166 | "after" specifies after how many seconds (fractional values are |
170 | "after" specifies after how many seconds (fractional values are |
167 | supported) should the timer activate. "cb" the callback to invoke in |
171 | supported) the callback should be invoked. "cb" is the callback to |
168 | that case. |
172 | invoke in that case. |
|
|
173 | |
|
|
174 | Although the callback might get passed parameters, their value and |
|
|
175 | presence is undefined and you cannot rely on them. Portable AnyEvent |
|
|
176 | callbacks cannot use arguments passed to time watcher callbacks. |
169 | |
177 | |
170 | The timer callback will be invoked at most once: if you want a repeating |
178 | The timer callback will be invoked at most once: if you want a repeating |
171 | timer you have to create a new watcher (this is a limitation by both Tk |
179 | timer you have to create a new watcher (this is a limitation by both Tk |
172 | and Glib). |
180 | and Glib). |
173 | |
181 | |
… | |
… | |
200 | o'clock"). |
208 | o'clock"). |
201 | |
209 | |
202 | While most event loops expect timers to specified in a relative way, |
210 | While most event loops expect timers to specified in a relative way, |
203 | they use absolute time internally. This makes a difference when your |
211 | they use absolute time internally. This makes a difference when your |
204 | clock "jumps", for example, when ntp decides to set your clock backwards |
212 | clock "jumps", for example, when ntp decides to set your clock backwards |
205 | from the wrong 2014-01-01 to 2008-01-01, a watcher that you created to |
213 | from the wrong date of 2014-01-01 to 2008-01-01, a watcher that is |
206 | fire "after" a second might actually take six years to finally fire. |
214 | supposed to fire "after" a second might actually take six years to |
|
|
215 | finally fire. |
207 | |
216 | |
208 | AnyEvent cannot compensate for this. The only event loop that is |
217 | AnyEvent cannot compensate for this. The only event loop that is |
209 | conscious about these issues is EV, which offers both relative |
218 | conscious about these issues is EV, which offers both relative |
210 | (ev_timer) and absolute (ev_periodic) timers. |
219 | (ev_timer, based on true relative time) and absolute (ev_periodic, based |
|
|
220 | on wallclock time) timers. |
211 | |
221 | |
212 | AnyEvent always prefers relative timers, if available, matching the |
222 | AnyEvent always prefers relative timers, if available, matching the |
213 | AnyEvent API. |
223 | AnyEvent API. |
214 | |
224 | |
215 | SIGNAL WATCHERS |
225 | SIGNAL WATCHERS |
216 | You can watch for signals using a signal watcher, "signal" is the signal |
226 | You can watch for signals using a signal watcher, "signal" is the signal |
217 | *name* without any "SIG" prefix, "cb" is the Perl callback to be invoked |
227 | *name* without any "SIG" prefix, "cb" is the Perl callback to be invoked |
218 | whenever a signal occurs. |
228 | whenever a signal occurs. |
219 | |
229 | |
|
|
230 | Although the callback might get passed parameters, their value and |
|
|
231 | presence is undefined and you cannot rely on them. Portable AnyEvent |
|
|
232 | callbacks cannot use arguments passed to signal watcher callbacks. |
|
|
233 | |
220 | Multiple signals occurances can be clumped together into one callback |
234 | Multiple signal occurances can be clumped together into one callback |
221 | invocation, and callback invocation will be synchronous. synchronous |
235 | invocation, and callback invocation will be synchronous. synchronous |
222 | means that it might take a while until the signal gets handled by the |
236 | means that it might take a while until the signal gets handled by the |
223 | process, but it is guarenteed not to interrupt any other callbacks. |
237 | process, but it is guarenteed not to interrupt any other callbacks. |
224 | |
238 | |
225 | The main advantage of using these watchers is that you can share a |
239 | The main advantage of using these watchers is that you can share a |
… | |
… | |
237 | |
251 | |
238 | The child process is specified by the "pid" argument (if set to 0, it |
252 | The child process is specified by the "pid" argument (if set to 0, it |
239 | watches for any child process exit). The watcher will trigger as often |
253 | watches for any child process exit). The watcher will trigger as often |
240 | as status change for the child are received. This works by installing a |
254 | as status change for the child are received. This works by installing a |
241 | signal handler for "SIGCHLD". The callback will be called with the pid |
255 | signal handler for "SIGCHLD". The callback will be called with the pid |
242 | and exit status (as returned by waitpid). |
256 | and exit status (as returned by waitpid), so unlike other watcher types, |
|
|
257 | you *can* rely on child watcher callback arguments. |
243 | |
258 | |
244 | Example: wait for pid 1333 |
259 | There is a slight catch to child watchers, however: you usually start |
|
|
260 | them *after* the child process was created, and this means the process |
|
|
261 | could have exited already (and no SIGCHLD will be sent anymore). |
|
|
262 | |
|
|
263 | Not all event models handle this correctly (POE doesn't), but even for |
|
|
264 | event models that *do* handle this correctly, they usually need to be |
|
|
265 | loaded before the process exits (i.e. before you fork in the first |
|
|
266 | place). |
|
|
267 | |
|
|
268 | This means you cannot create a child watcher as the very first thing in |
|
|
269 | an AnyEvent program, you *have* to create at least one watcher before |
|
|
270 | you "fork" the child (alternatively, you can call "AnyEvent::detect"). |
|
|
271 | |
|
|
272 | Example: fork a process and wait for it |
|
|
273 | |
|
|
274 | my $done = AnyEvent->condvar; |
|
|
275 | |
|
|
276 | my $pid = fork or exit 5; |
245 | |
277 | |
246 | my $w = AnyEvent->child ( |
278 | my $w = AnyEvent->child ( |
247 | pid => 1333, |
279 | pid => $pid, |
248 | cb => sub { |
280 | cb => sub { |
249 | my ($pid, $status) = @_; |
281 | my ($pid, $status) = @_; |
250 | warn "pid $pid exited with status $status"; |
282 | warn "pid $pid exited with status $status"; |
|
|
283 | $done->send; |
251 | }, |
284 | }, |
252 | ); |
285 | ); |
253 | |
286 | |
|
|
287 | # do something else, then wait for process exit |
|
|
288 | $done->recv; |
|
|
289 | |
254 | CONDITION VARIABLES |
290 | CONDITION VARIABLES |
|
|
291 | If you are familiar with some event loops you will know that all of them |
|
|
292 | require you to run some blocking "loop", "run" or similar function that |
|
|
293 | will actively watch for new events and call your callbacks. |
|
|
294 | |
|
|
295 | AnyEvent is different, it expects somebody else to run the event loop |
|
|
296 | and will only block when necessary (usually when told by the user). |
|
|
297 | |
|
|
298 | The instrument to do that is called a "condition variable", so called |
|
|
299 | because they represent a condition that must become true. |
|
|
300 | |
255 | Condition variables can be created by calling the "AnyEvent->condvar" |
301 | Condition variables can be created by calling the "AnyEvent->condvar" |
256 | method without any arguments. |
302 | method, usually without arguments. The only argument pair allowed is |
|
|
303 | "cb", which specifies a callback to be called when the condition |
|
|
304 | variable becomes true. |
257 | |
305 | |
258 | A condition variable waits for a condition - precisely that the |
306 | After creation, the conditon variable is "false" until it becomes "true" |
259 | "->broadcast" method has been called. |
307 | by calling the "send" method. |
260 | |
308 | |
261 | They are very useful to signal that a condition has been fulfilled, for |
309 | Condition variables are similar to callbacks, except that you can |
|
|
310 | optionally wait for them. They can also be called merge points - points |
|
|
311 | in time where multiple outstandign events have been processed. And yet |
|
|
312 | another way to call them is transations - each condition variable can be |
|
|
313 | used to represent a transaction, which finishes at some point and |
|
|
314 | delivers a result. |
|
|
315 | |
|
|
316 | Condition variables are very useful to signal that something has |
262 | example, if you write a module that does asynchronous http requests, |
317 | finished, for example, if you write a module that does asynchronous http |
263 | then a condition variable would be the ideal candidate to signal the |
318 | requests, then a condition variable would be the ideal candidate to |
264 | availability of results. |
319 | signal the availability of results. The user can either act when the |
|
|
320 | callback is called or can synchronously "->recv" for the results. |
265 | |
321 | |
266 | You can also use condition variables to block your main program until an |
322 | You can also use them to simulate traditional event loops - for example, |
267 | event occurs - for example, you could "->wait" in your main program |
323 | you can block your main program until an event occurs - for example, you |
268 | until the user clicks the Quit button in your app, which would |
324 | could "->recv" in your main program until the user clicks the Quit |
269 | "->broadcast" the "quit" event. |
325 | button of your app, which would "->send" the "quit" event. |
270 | |
326 | |
271 | Note that condition variables recurse into the event loop - if you have |
327 | Note that condition variables recurse into the event loop - if you have |
272 | two pirces of code that call "->wait" in a round-robbin fashion, you |
328 | two pieces of code that call "->recv" in a round-robbin fashion, you |
273 | lose. Therefore, condition variables are good to export to your caller, |
329 | lose. Therefore, condition variables are good to export to your caller, |
274 | but you should avoid making a blocking wait yourself, at least in |
330 | but you should avoid making a blocking wait yourself, at least in |
275 | callbacks, as this asks for trouble. |
331 | callbacks, as this asks for trouble. |
276 | |
332 | |
277 | This object has two methods: |
333 | Condition variables are represented by hash refs in perl, and the keys |
|
|
334 | used by AnyEvent itself are all named "_ae_XXX" to make subclassing easy |
|
|
335 | (it is often useful to build your own transaction class on top of |
|
|
336 | AnyEvent). To subclass, use "AnyEvent::CondVar" as base class and call |
|
|
337 | it's "new" method in your own "new" method. |
278 | |
338 | |
279 | $cv->wait |
339 | There are two "sides" to a condition variable - the "producer side" |
|
|
340 | which eventually calls "-> send", and the "consumer side", which waits |
|
|
341 | for the send to occur. |
|
|
342 | |
|
|
343 | Example: |
|
|
344 | |
|
|
345 | # wait till the result is ready |
|
|
346 | my $result_ready = AnyEvent->condvar; |
|
|
347 | |
|
|
348 | # do something such as adding a timer |
|
|
349 | # or socket watcher the calls $result_ready->send |
|
|
350 | # when the "result" is ready. |
|
|
351 | # in this case, we simply use a timer: |
|
|
352 | my $w = AnyEvent->timer ( |
|
|
353 | after => 1, |
|
|
354 | cb => sub { $result_ready->send }, |
|
|
355 | ); |
|
|
356 | |
|
|
357 | # this "blocks" (while handling events) till the callback |
|
|
358 | # calls send |
|
|
359 | $result_ready->recv; |
|
|
360 | |
|
|
361 | METHODS FOR PRODUCERS |
|
|
362 | These methods should only be used by the producing side, i.e. the |
|
|
363 | code/module that eventually sends the signal. Note that it is also the |
|
|
364 | producer side which creates the condvar in most cases, but it isn't |
|
|
365 | uncommon for the consumer to create it as well. |
|
|
366 | |
|
|
367 | $cv->send (...) |
|
|
368 | Flag the condition as ready - a running "->recv" and all further |
|
|
369 | calls to "recv" will (eventually) return after this method has been |
|
|
370 | called. If nobody is waiting the send will be remembered. |
|
|
371 | |
|
|
372 | If a callback has been set on the condition variable, it is called |
|
|
373 | immediately from within send. |
|
|
374 | |
|
|
375 | Any arguments passed to the "send" call will be returned by all |
|
|
376 | future "->recv" calls. |
|
|
377 | |
|
|
378 | $cv->croak ($error) |
|
|
379 | Similar to send, but causes all call's to "->recv" to invoke |
|
|
380 | "Carp::croak" with the given error message/object/scalar. |
|
|
381 | |
|
|
382 | This can be used to signal any errors to the condition variable |
|
|
383 | user/consumer. |
|
|
384 | |
|
|
385 | $cv->begin ([group callback]) |
|
|
386 | $cv->end |
|
|
387 | These two methods are EXPERIMENTAL and MIGHT CHANGE. |
|
|
388 | |
|
|
389 | These two methods can be used to combine many transactions/events |
|
|
390 | into one. For example, a function that pings many hosts in parallel |
|
|
391 | might want to use a condition variable for the whole process. |
|
|
392 | |
|
|
393 | Every call to "->begin" will increment a counter, and every call to |
|
|
394 | "->end" will decrement it. If the counter reaches 0 in "->end", the |
|
|
395 | (last) callback passed to "begin" will be executed. That callback is |
|
|
396 | *supposed* to call "->send", but that is not required. If no |
|
|
397 | callback was set, "send" will be called without any arguments. |
|
|
398 | |
|
|
399 | Let's clarify this with the ping example: |
|
|
400 | |
|
|
401 | my $cv = AnyEvent->condvar; |
|
|
402 | |
|
|
403 | my %result; |
|
|
404 | $cv->begin (sub { $cv->send (\%result) }); |
|
|
405 | |
|
|
406 | for my $host (@list_of_hosts) { |
|
|
407 | $cv->begin; |
|
|
408 | ping_host_then_call_callback $host, sub { |
|
|
409 | $result{$host} = ...; |
|
|
410 | $cv->end; |
|
|
411 | }; |
|
|
412 | } |
|
|
413 | |
|
|
414 | $cv->end; |
|
|
415 | |
|
|
416 | This code fragment supposedly pings a number of hosts and calls |
|
|
417 | "send" after results for all then have have been gathered - in any |
|
|
418 | order. To achieve this, the code issues a call to "begin" when it |
|
|
419 | starts each ping request and calls "end" when it has received some |
|
|
420 | result for it. Since "begin" and "end" only maintain a counter, the |
|
|
421 | order in which results arrive is not relevant. |
|
|
422 | |
|
|
423 | There is an additional bracketing call to "begin" and "end" outside |
|
|
424 | the loop, which serves two important purposes: first, it sets the |
|
|
425 | callback to be called once the counter reaches 0, and second, it |
|
|
426 | ensures that "send" is called even when "no" hosts are being pinged |
|
|
427 | (the loop doesn't execute once). |
|
|
428 | |
|
|
429 | This is the general pattern when you "fan out" into multiple |
|
|
430 | subrequests: use an outer "begin"/"end" pair to set the callback and |
|
|
431 | ensure "end" is called at least once, and then, for each subrequest |
|
|
432 | you start, call "begin" and for eahc subrequest you finish, call |
|
|
433 | "end". |
|
|
434 | |
|
|
435 | METHODS FOR CONSUMERS |
|
|
436 | These methods should only be used by the consuming side, i.e. the code |
|
|
437 | awaits the condition. |
|
|
438 | |
|
|
439 | $cv->recv |
280 | Wait (blocking if necessary) until the "->broadcast" method has been |
440 | Wait (blocking if necessary) until the "->send" or "->croak" methods |
281 | called on c<$cv>, while servicing other watchers normally. |
441 | have been called on c<$cv>, while servicing other watchers normally. |
282 | |
442 | |
283 | You can only wait once on a condition - additional calls will return |
443 | You can only wait once on a condition - additional calls are valid |
284 | immediately. |
444 | but will return immediately. |
|
|
445 | |
|
|
446 | If an error condition has been set by calling "->croak", then this |
|
|
447 | function will call "croak". |
|
|
448 | |
|
|
449 | In list context, all parameters passed to "send" will be returned, |
|
|
450 | in scalar context only the first one will be returned. |
285 | |
451 | |
286 | Not all event models support a blocking wait - some die in that case |
452 | Not all event models support a blocking wait - some die in that case |
287 | (programs might want to do that to stay interactive), so *if you are |
453 | (programs might want to do that to stay interactive), so *if you are |
288 | using this from a module, never require a blocking wait*, but let |
454 | using this from a module, never require a blocking wait*, but let |
289 | the caller decide whether the call will block or not (for example, |
455 | the caller decide whether the call will block or not (for example, |
290 | by coupling condition variables with some kind of request results |
456 | by coupling condition variables with some kind of request results |
291 | and supporting callbacks so the caller knows that getting the result |
457 | and supporting callbacks so the caller knows that getting the result |
292 | will not block, while still suppporting blocking waits if the caller |
458 | will not block, while still suppporting blocking waits if the caller |
293 | so desires). |
459 | so desires). |
294 | |
460 | |
295 | Another reason *never* to "->wait" in a module is that you cannot |
461 | Another reason *never* to "->recv" in a module is that you cannot |
296 | sensibly have two "->wait"'s in parallel, as that would require |
462 | sensibly have two "->recv"'s in parallel, as that would require |
297 | multiple interpreters or coroutines/threads, none of which |
463 | multiple interpreters or coroutines/threads, none of which |
298 | "AnyEvent" can supply (the coroutine-aware backends |
464 | "AnyEvent" can supply. |
299 | AnyEvent::Impl::CoroEV and AnyEvent::Impl::CoroEvent explicitly |
|
|
300 | support concurrent "->wait"'s from different coroutines, however). |
|
|
301 | |
465 | |
302 | $cv->broadcast |
466 | The Coro module, however, *can* and *does* supply coroutines and, in |
303 | Flag the condition as ready - a running "->wait" and all further |
467 | fact, Coro::AnyEvent replaces AnyEvent's condvars by coroutine-safe |
304 | calls to "wait" will (eventually) return after this method has been |
468 | versions and also integrates coroutines into AnyEvent, making |
305 | called. If nobody is waiting the broadcast will be remembered.. |
469 | blocking "->recv" calls perfectly safe as long as they are done from |
|
|
470 | another coroutine (one that doesn't run the event loop). |
306 | |
471 | |
307 | Example: |
472 | You can ensure that "-recv" never blocks by setting a callback and |
|
|
473 | only calling "->recv" from within that callback (or at a later |
|
|
474 | time). This will work even when the event loop does not support |
|
|
475 | blocking waits otherwise. |
308 | |
476 | |
309 | # wait till the result is ready |
477 | $bool = $cv->ready |
310 | my $result_ready = AnyEvent->condvar; |
478 | Returns true when the condition is "true", i.e. whether "send" or |
|
|
479 | "croak" have been called. |
311 | |
480 | |
312 | # do something such as adding a timer |
481 | $cb = $cv->cb ([new callback]) |
313 | # or socket watcher the calls $result_ready->broadcast |
482 | This is a mutator function that returns the callback set and |
314 | # when the "result" is ready. |
483 | optionally replaces it before doing so. |
315 | # in this case, we simply use a timer: |
|
|
316 | my $w = AnyEvent->timer ( |
|
|
317 | after => 1, |
|
|
318 | cb => sub { $result_ready->broadcast }, |
|
|
319 | ); |
|
|
320 | |
484 | |
321 | # this "blocks" (while handling events) till the watcher |
485 | The callback will be called when the condition becomes "true", i.e. |
322 | # calls broadcast |
486 | when "send" or "croak" are called. Calling "recv" inside the |
323 | $result_ready->wait; |
487 | callback or at any later time is guaranteed not to block. |
324 | |
488 | |
325 | GLOBAL VARIABLES AND FUNCTIONS |
489 | GLOBAL VARIABLES AND FUNCTIONS |
326 | $AnyEvent::MODEL |
490 | $AnyEvent::MODEL |
327 | Contains "undef" until the first watcher is being created. Then it |
491 | Contains "undef" until the first watcher is being created. Then it |
328 | contains the event model that is being used, which is the name of |
492 | contains the event model that is being used, which is the name of |
… | |
… | |
330 | the "AnyEvent::Impl:xxx" modules, but can be any other class in the |
494 | the "AnyEvent::Impl:xxx" modules, but can be any other class in the |
331 | case AnyEvent has been extended at runtime (e.g. in *rxvt-unicode*). |
495 | case AnyEvent has been extended at runtime (e.g. in *rxvt-unicode*). |
332 | |
496 | |
333 | The known classes so far are: |
497 | The known classes so far are: |
334 | |
498 | |
335 | AnyEvent::Impl::CoroEV based on Coro::EV, best choice. |
|
|
336 | AnyEvent::Impl::CoroEvent based on Coro::Event, second best choice. |
|
|
337 | AnyEvent::Impl::EV based on EV (an interface to libev, also best choice). |
499 | AnyEvent::Impl::EV based on EV (an interface to libev, best choice). |
338 | AnyEvent::Impl::Event based on Event, also second best choice :) |
500 | AnyEvent::Impl::Event based on Event, second best choice. |
|
|
501 | AnyEvent::Impl::Perl pure-perl implementation, fast and portable. |
339 | AnyEvent::Impl::Glib based on Glib, third-best choice. |
502 | AnyEvent::Impl::Glib based on Glib, third-best choice. |
340 | AnyEvent::Impl::Tk based on Tk, very bad choice. |
503 | AnyEvent::Impl::Tk based on Tk, very bad choice. |
341 | AnyEvent::Impl::Perl pure-perl implementation, inefficient but portable. |
504 | AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs). |
|
|
505 | AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse. |
|
|
506 | AnyEvent::Impl::POE based on POE, not generic enough for full support. |
|
|
507 | |
|
|
508 | There is no support for WxWidgets, as WxWidgets has no support for |
|
|
509 | watching file handles. However, you can use WxWidgets through the |
|
|
510 | POE Adaptor, as POE has a Wx backend that simply polls 20 times per |
|
|
511 | second, which was considered to be too horrible to even consider for |
|
|
512 | AnyEvent. Likewise, other POE backends can be used by AnyEvent by |
|
|
513 | using it's adaptor. |
|
|
514 | |
|
|
515 | AnyEvent knows about Prima and Wx and will try to use POE when |
|
|
516 | autodetecting them. |
342 | |
517 | |
343 | AnyEvent::detect |
518 | AnyEvent::detect |
344 | Returns $AnyEvent::MODEL, forcing autodetection of the event model |
519 | Returns $AnyEvent::MODEL, forcing autodetection of the event model |
345 | if necessary. You should only call this function right before you |
520 | if necessary. You should only call this function right before you |
346 | would have created an AnyEvent watcher anyway, that is, as late as |
521 | would have created an AnyEvent watcher anyway, that is, as late as |
347 | possible at runtime. |
522 | possible at runtime. |
348 | |
523 | |
|
|
524 | $guard = AnyEvent::post_detect { BLOCK } |
|
|
525 | Arranges for the code block to be executed as soon as the event |
|
|
526 | model is autodetected (or immediately if this has already happened). |
|
|
527 | |
|
|
528 | If called in scalar or list context, then it creates and returns an |
|
|
529 | object that automatically removes the callback again when it is |
|
|
530 | destroyed. See Coro::BDB for a case where this is useful. |
|
|
531 | |
|
|
532 | @AnyEvent::post_detect |
|
|
533 | If there are any code references in this array (you can "push" to it |
|
|
534 | before or after loading AnyEvent), then they will called directly |
|
|
535 | after the event loop has been chosen. |
|
|
536 | |
|
|
537 | You should check $AnyEvent::MODEL before adding to this array, |
|
|
538 | though: if it contains a true value then the event loop has already |
|
|
539 | been detected, and the array will be ignored. |
|
|
540 | |
|
|
541 | Best use "AnyEvent::post_detect { BLOCK }" instead. |
|
|
542 | |
349 | WHAT TO DO IN A MODULE |
543 | WHAT TO DO IN A MODULE |
350 | As a module author, you should "use AnyEvent" and call AnyEvent methods |
544 | As a module author, you should "use AnyEvent" and call AnyEvent methods |
351 | freely, but you should not load a specific event module or rely on it. |
545 | freely, but you should not load a specific event module or rely on it. |
352 | |
546 | |
353 | Be careful when you create watchers in the module body - AnyEvent will |
547 | Be careful when you create watchers in the module body - AnyEvent will |
354 | decide which event module to use as soon as the first method is called, |
548 | decide which event module to use as soon as the first method is called, |
355 | so by calling AnyEvent in your module body you force the user of your |
549 | so by calling AnyEvent in your module body you force the user of your |
356 | module to load the event module first. |
550 | module to load the event module first. |
357 | |
551 | |
358 | Never call "->wait" on a condition variable unless you *know* that the |
552 | Never call "->recv" on a condition variable unless you *know* that the |
359 | "->broadcast" method has been called on it already. This is because it |
553 | "->send" method has been called on it already. This is because it will |
360 | will stall the whole program, and the whole point of using events is to |
554 | stall the whole program, and the whole point of using events is to stay |
361 | stay interactive. |
555 | interactive. |
362 | |
556 | |
363 | It is fine, however, to call "->wait" when the user of your module |
557 | It is fine, however, to call "->recv" when the user of your module |
364 | requests it (i.e. if you create a http request object ad have a method |
558 | requests it (i.e. if you create a http request object ad have a method |
365 | called "results" that returns the results, it should call "->wait" |
559 | called "results" that returns the results, it should call "->recv" |
366 | freely, as the user of your module knows what she is doing. always). |
560 | freely, as the user of your module knows what she is doing. always). |
367 | |
561 | |
368 | WHAT TO DO IN THE MAIN PROGRAM |
562 | WHAT TO DO IN THE MAIN PROGRAM |
369 | There will always be a single main program - the only place that should |
563 | There will always be a single main program - the only place that should |
370 | dictate which event model to use. |
564 | dictate which event model to use. |
… | |
… | |
385 | |
579 | |
386 | You can chose to use a rather inefficient pure-perl implementation by |
580 | You can chose to use a rather inefficient pure-perl implementation by |
387 | loading the "AnyEvent::Impl::Perl" module, which gives you similar |
581 | loading the "AnyEvent::Impl::Perl" module, which gives you similar |
388 | behaviour everywhere, but letting AnyEvent chose is generally better. |
582 | behaviour everywhere, but letting AnyEvent chose is generally better. |
389 | |
583 | |
|
|
584 | OTHER MODULES |
|
|
585 | The following is a non-exhaustive list of additional modules that use |
|
|
586 | AnyEvent and can therefore be mixed easily with other AnyEvent modules |
|
|
587 | in the same program. Some of the modules come with AnyEvent, some are |
|
|
588 | available via CPAN. |
|
|
589 | |
|
|
590 | AnyEvent::Util |
|
|
591 | Contains various utility functions that replace often-used but |
|
|
592 | blocking functions such as "inet_aton" by event-/callback-based |
|
|
593 | versions. |
|
|
594 | |
|
|
595 | AnyEvent::Handle |
|
|
596 | Provide read and write buffers and manages watchers for reads and |
|
|
597 | writes. |
|
|
598 | |
|
|
599 | AnyEvent::HTTPD |
|
|
600 | Provides a simple web application server framework. |
|
|
601 | |
|
|
602 | AnyEvent::DNS |
|
|
603 | Provides asynchronous DNS resolver capabilities, beyond what |
|
|
604 | AnyEvent::Util offers. |
|
|
605 | |
|
|
606 | AnyEvent::FastPing |
|
|
607 | The fastest ping in the west. |
|
|
608 | |
|
|
609 | Net::IRC3 |
|
|
610 | AnyEvent based IRC client module family. |
|
|
611 | |
|
|
612 | Net::XMPP2 |
|
|
613 | AnyEvent based XMPP (Jabber protocol) module family. |
|
|
614 | |
|
|
615 | Net::FCP |
|
|
616 | AnyEvent-based implementation of the Freenet Client Protocol, |
|
|
617 | birthplace of AnyEvent. |
|
|
618 | |
|
|
619 | Event::ExecFlow |
|
|
620 | High level API for event-based execution flow control. |
|
|
621 | |
|
|
622 | Coro |
|
|
623 | Has special support for AnyEvent via Coro::AnyEvent. |
|
|
624 | |
|
|
625 | AnyEvent::AIO, IO::AIO |
|
|
626 | Truly asynchronous I/O, should be in the toolbox of every event |
|
|
627 | programmer. AnyEvent::AIO transparently fuses IO::AIO and AnyEvent |
|
|
628 | together. |
|
|
629 | |
|
|
630 | AnyEvent::BDB, BDB |
|
|
631 | Truly asynchronous Berkeley DB access. AnyEvent::AIO transparently |
|
|
632 | fuses IO::AIO and AnyEvent together. |
|
|
633 | |
|
|
634 | IO::Lambda |
|
|
635 | The lambda approach to I/O - don't ask, look there. Can use |
|
|
636 | AnyEvent. |
|
|
637 | |
390 | SUPPLYING YOUR OWN EVENT MODEL INTERFACE |
638 | SUPPLYING YOUR OWN EVENT MODEL INTERFACE |
391 | This is an advanced topic that you do not normally need to use AnyEvent |
639 | This is an advanced topic that you do not normally need to use AnyEvent |
392 | in a module. This section is only of use to event loop authors who want |
640 | in a module. This section is only of use to event loop authors who want |
393 | to provide AnyEvent compatibility. |
641 | to provide AnyEvent compatibility. |
394 | |
642 | |
… | |
… | |
431 | must not be done in an interactive application, so it makes sense. |
679 | must not be done in an interactive application, so it makes sense. |
432 | |
680 | |
433 | ENVIRONMENT VARIABLES |
681 | ENVIRONMENT VARIABLES |
434 | The following environment variables are used by this module: |
682 | The following environment variables are used by this module: |
435 | |
683 | |
436 | "PERL_ANYEVENT_VERBOSE" when set to 2 or higher, cause AnyEvent to |
684 | "PERL_ANYEVENT_VERBOSE" |
437 | report to STDERR which event model it chooses. |
685 | By default, AnyEvent will be completely silent except in fatal |
|
|
686 | conditions. You can set this environment variable to make AnyEvent |
|
|
687 | more talkative. |
|
|
688 | |
|
|
689 | When set to 1 or higher, causes AnyEvent to warn about unexpected |
|
|
690 | conditions, such as not being able to load the event model specified |
|
|
691 | by "PERL_ANYEVENT_MODEL". |
|
|
692 | |
|
|
693 | When set to 2 or higher, cause AnyEvent to report to STDERR which |
|
|
694 | event model it chooses. |
|
|
695 | |
|
|
696 | "PERL_ANYEVENT_MODEL" |
|
|
697 | This can be used to specify the event model to be used by AnyEvent, |
|
|
698 | before autodetection and -probing kicks in. It must be a string |
|
|
699 | consisting entirely of ASCII letters. The string "AnyEvent::Impl::" |
|
|
700 | gets prepended and the resulting module name is loaded and if the |
|
|
701 | load was successful, used as event model. If it fails to load |
|
|
702 | AnyEvent will proceed with autodetection and -probing. |
|
|
703 | |
|
|
704 | This functionality might change in future versions. |
|
|
705 | |
|
|
706 | For example, to force the pure perl model (AnyEvent::Impl::Perl) you |
|
|
707 | could start your program like this: |
|
|
708 | |
|
|
709 | PERL_ANYEVENT_MODEL=Perl perl ... |
438 | |
710 | |
439 | EXAMPLE PROGRAM |
711 | EXAMPLE PROGRAM |
440 | The following program uses an IO watcher to read data from STDIN, a |
712 | The following program uses an I/O watcher to read data from STDIN, a |
441 | timer to display a message once per second, and a condition variable to |
713 | timer to display a message once per second, and a condition variable to |
442 | quit the program when the user enters quit: |
714 | quit the program when the user enters quit: |
443 | |
715 | |
444 | use AnyEvent; |
716 | use AnyEvent; |
445 | |
717 | |
… | |
… | |
450 | poll => 'r', |
722 | poll => 'r', |
451 | cb => sub { |
723 | cb => sub { |
452 | warn "io event <$_[0]>\n"; # will always output <r> |
724 | warn "io event <$_[0]>\n"; # will always output <r> |
453 | chomp (my $input = <STDIN>); # read a line |
725 | chomp (my $input = <STDIN>); # read a line |
454 | warn "read: $input\n"; # output what has been read |
726 | warn "read: $input\n"; # output what has been read |
455 | $cv->broadcast if $input =~ /^q/i; # quit program if /^q/i |
727 | $cv->send if $input =~ /^q/i; # quit program if /^q/i |
456 | }, |
728 | }, |
457 | ); |
729 | ); |
458 | |
730 | |
459 | my $time_watcher; # can only be used once |
731 | my $time_watcher; # can only be used once |
460 | |
732 | |
… | |
… | |
465 | }); |
737 | }); |
466 | } |
738 | } |
467 | |
739 | |
468 | new_timer; # create first timer |
740 | new_timer; # create first timer |
469 | |
741 | |
470 | $cv->wait; # wait until user enters /^q/i |
742 | $cv->recv; # wait until user enters /^q/i |
471 | |
743 | |
472 | REAL-WORLD EXAMPLE |
744 | REAL-WORLD EXAMPLE |
473 | Consider the Net::FCP module. It features (among others) the following |
745 | Consider the Net::FCP module. It features (among others) the following |
474 | API calls, which are to freenet what HTTP GET requests are to http: |
746 | API calls, which are to freenet what HTTP GET requests are to http: |
475 | |
747 | |
… | |
… | |
530 | |
802 | |
531 | sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf}; |
803 | sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf}; |
532 | |
804 | |
533 | if (end-of-file or data complete) { |
805 | if (end-of-file or data complete) { |
534 | $txn->{result} = $txn->{buf}; |
806 | $txn->{result} = $txn->{buf}; |
535 | $txn->{finished}->broadcast; |
807 | $txn->{finished}->send; |
536 | $txb->{cb}->($txn) of $txn->{cb}; # also call callback |
808 | $txb->{cb}->($txn) of $txn->{cb}; # also call callback |
537 | } |
809 | } |
538 | |
810 | |
539 | The "result" method, finally, just waits for the finished signal (if the |
811 | The "result" method, finally, just waits for the finished signal (if the |
540 | request was already finished, it doesn't wait, of course, and returns |
812 | request was already finished, it doesn't wait, of course, and returns |
541 | the data: |
813 | the data: |
542 | |
814 | |
543 | $txn->{finished}->wait; |
815 | $txn->{finished}->recv; |
544 | return $txn->{result}; |
816 | return $txn->{result}; |
545 | |
817 | |
546 | The actual code goes further and collects all errors ("die"s, |
818 | The actual code goes further and collects all errors ("die"s, |
547 | exceptions) that occured during request processing. The "result" method |
819 | exceptions) that occured during request processing. The "result" method |
548 | detects whether an exception as thrown (it is stored inside the $txn |
820 | detects whether an exception as thrown (it is stored inside the $txn |
… | |
… | |
583 | |
855 | |
584 | my $quit = AnyEvent->condvar; |
856 | my $quit = AnyEvent->condvar; |
585 | |
857 | |
586 | $fcp->txn_client_get ($url)->cb (sub { |
858 | $fcp->txn_client_get ($url)->cb (sub { |
587 | ... |
859 | ... |
588 | $quit->broadcast; |
860 | $quit->send; |
589 | }); |
861 | }); |
590 | |
862 | |
591 | $quit->wait; |
863 | $quit->recv; |
|
|
864 | |
|
|
865 | BENCHMARKS |
|
|
866 | To give you an idea of the performance and overheads that AnyEvent adds |
|
|
867 | over the event loops themselves and to give you an impression of the |
|
|
868 | speed of various event loops I prepared some benchmarks. |
|
|
869 | |
|
|
870 | BENCHMARKING ANYEVENT OVERHEAD |
|
|
871 | Here is a benchmark of various supported event models used natively and |
|
|
872 | through anyevent. The benchmark creates a lot of timers (with a zero |
|
|
873 | timeout) and I/O watchers (watching STDOUT, a pty, to become writable, |
|
|
874 | which it is), lets them fire exactly once and destroys them again. |
|
|
875 | |
|
|
876 | Source code for this benchmark is found as eg/bench in the AnyEvent |
|
|
877 | distribution. |
|
|
878 | |
|
|
879 | Explanation of the columns |
|
|
880 | *watcher* is the number of event watchers created/destroyed. Since |
|
|
881 | different event models feature vastly different performances, each event |
|
|
882 | loop was given a number of watchers so that overall runtime is |
|
|
883 | acceptable and similar between tested event loop (and keep them from |
|
|
884 | crashing): Glib would probably take thousands of years if asked to |
|
|
885 | process the same number of watchers as EV in this benchmark. |
|
|
886 | |
|
|
887 | *bytes* is the number of bytes (as measured by the resident set size, |
|
|
888 | RSS) consumed by each watcher. This method of measuring captures both C |
|
|
889 | and Perl-based overheads. |
|
|
890 | |
|
|
891 | *create* is the time, in microseconds (millionths of seconds), that it |
|
|
892 | takes to create a single watcher. The callback is a closure shared |
|
|
893 | between all watchers, to avoid adding memory overhead. That means |
|
|
894 | closure creation and memory usage is not included in the figures. |
|
|
895 | |
|
|
896 | *invoke* is the time, in microseconds, used to invoke a simple callback. |
|
|
897 | The callback simply counts down a Perl variable and after it was invoked |
|
|
898 | "watcher" times, it would "->send" a condvar once to signal the end of |
|
|
899 | this phase. |
|
|
900 | |
|
|
901 | *destroy* is the time, in microseconds, that it takes to destroy a |
|
|
902 | single watcher. |
|
|
903 | |
|
|
904 | Results |
|
|
905 | name watchers bytes create invoke destroy comment |
|
|
906 | EV/EV 400000 244 0.56 0.46 0.31 EV native interface |
|
|
907 | EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers |
|
|
908 | CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal |
|
|
909 | Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation |
|
|
910 | Event/Event 16000 516 31.88 31.30 0.85 Event native interface |
|
|
911 | Event/Any 16000 590 35.75 31.42 1.08 Event + AnyEvent watchers |
|
|
912 | Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour |
|
|
913 | Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers |
|
|
914 | POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event |
|
|
915 | POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select |
|
|
916 | |
|
|
917 | Discussion |
|
|
918 | The benchmark does *not* measure scalability of the event loop very |
|
|
919 | well. For example, a select-based event loop (such as the pure perl one) |
|
|
920 | can never compete with an event loop that uses epoll when the number of |
|
|
921 | file descriptors grows high. In this benchmark, all events become ready |
|
|
922 | at the same time, so select/poll-based implementations get an unnatural |
|
|
923 | speed boost. |
|
|
924 | |
|
|
925 | Also, note that the number of watchers usually has a nonlinear effect on |
|
|
926 | overall speed, that is, creating twice as many watchers doesn't take |
|
|
927 | twice the time - usually it takes longer. This puts event loops tested |
|
|
928 | with a higher number of watchers at a disadvantage. |
|
|
929 | |
|
|
930 | To put the range of results into perspective, consider that on the |
|
|
931 | benchmark machine, handling an event takes roughly 1600 CPU cycles with |
|
|
932 | EV, 3100 CPU cycles with AnyEvent's pure perl loop and almost 3000000 |
|
|
933 | CPU cycles with POE. |
|
|
934 | |
|
|
935 | "EV" is the sole leader regarding speed and memory use, which are both |
|
|
936 | maximal/minimal, respectively. Even when going through AnyEvent, it uses |
|
|
937 | far less memory than any other event loop and is still faster than Event |
|
|
938 | natively. |
|
|
939 | |
|
|
940 | The pure perl implementation is hit in a few sweet spots (both the |
|
|
941 | constant timeout and the use of a single fd hit optimisations in the |
|
|
942 | perl interpreter and the backend itself). Nevertheless this shows that |
|
|
943 | it adds very little overhead in itself. Like any select-based backend |
|
|
944 | its performance becomes really bad with lots of file descriptors (and |
|
|
945 | few of them active), of course, but this was not subject of this |
|
|
946 | benchmark. |
|
|
947 | |
|
|
948 | The "Event" module has a relatively high setup and callback invocation |
|
|
949 | cost, but overall scores in on the third place. |
|
|
950 | |
|
|
951 | "Glib"'s memory usage is quite a bit higher, but it features a faster |
|
|
952 | callback invocation and overall ends up in the same class as "Event". |
|
|
953 | However, Glib scales extremely badly, doubling the number of watchers |
|
|
954 | increases the processing time by more than a factor of four, making it |
|
|
955 | completely unusable when using larger numbers of watchers (note that |
|
|
956 | only a single file descriptor was used in the benchmark, so |
|
|
957 | inefficiencies of "poll" do not account for this). |
|
|
958 | |
|
|
959 | The "Tk" adaptor works relatively well. The fact that it crashes with |
|
|
960 | more than 2000 watchers is a big setback, however, as correctness takes |
|
|
961 | precedence over speed. Nevertheless, its performance is surprising, as |
|
|
962 | the file descriptor is dup()ed for each watcher. This shows that the |
|
|
963 | dup() employed by some adaptors is not a big performance issue (it does |
|
|
964 | incur a hidden memory cost inside the kernel which is not reflected in |
|
|
965 | the figures above). |
|
|
966 | |
|
|
967 | "POE", regardless of underlying event loop (whether using its pure perl |
|
|
968 | select-based backend or the Event module, the POE-EV backend couldn't be |
|
|
969 | tested because it wasn't working) shows abysmal performance and memory |
|
|
970 | usage with AnyEvent: Watchers use almost 30 times as much memory as EV |
|
|
971 | watchers, and 10 times as much memory as Event (the high memory |
|
|
972 | requirements are caused by requiring a session for each watcher). |
|
|
973 | Watcher invocation speed is almost 900 times slower than with AnyEvent's |
|
|
974 | pure perl implementation. |
|
|
975 | |
|
|
976 | The design of the POE adaptor class in AnyEvent can not really account |
|
|
977 | for the performance issues, though, as session creation overhead is |
|
|
978 | small compared to execution of the state machine, which is coded pretty |
|
|
979 | optimally within AnyEvent::Impl::POE (and while everybody agrees that |
|
|
980 | using multiple sessions is not a good approach, especially regarding |
|
|
981 | memory usage, even the author of POE could not come up with a faster |
|
|
982 | design). |
|
|
983 | |
|
|
984 | Summary |
|
|
985 | * Using EV through AnyEvent is faster than any other event loop (even |
|
|
986 | when used without AnyEvent), but most event loops have acceptable |
|
|
987 | performance with or without AnyEvent. |
|
|
988 | |
|
|
989 | * The overhead AnyEvent adds is usually much smaller than the overhead |
|
|
990 | of the actual event loop, only with extremely fast event loops such |
|
|
991 | as EV adds AnyEvent significant overhead. |
|
|
992 | |
|
|
993 | * You should avoid POE like the plague if you want performance or |
|
|
994 | reasonable memory usage. |
|
|
995 | |
|
|
996 | BENCHMARKING THE LARGE SERVER CASE |
|
|
997 | This benchmark atcually benchmarks the event loop itself. It works by |
|
|
998 | creating a number of "servers": each server consists of a socketpair, a |
|
|
999 | timeout watcher that gets reset on activity (but never fires), and an |
|
|
1000 | I/O watcher waiting for input on one side of the socket. Each time the |
|
|
1001 | socket watcher reads a byte it will write that byte to a random other |
|
|
1002 | "server". |
|
|
1003 | |
|
|
1004 | The effect is that there will be a lot of I/O watchers, only part of |
|
|
1005 | which are active at any one point (so there is a constant number of |
|
|
1006 | active fds for each loop iterstaion, but which fds these are is random). |
|
|
1007 | The timeout is reset each time something is read because that reflects |
|
|
1008 | how most timeouts work (and puts extra pressure on the event loops). |
|
|
1009 | |
|
|
1010 | In this benchmark, we use 10000 socketpairs (20000 sockets), of which |
|
|
1011 | 100 (1%) are active. This mirrors the activity of large servers with |
|
|
1012 | many connections, most of which are idle at any one point in time. |
|
|
1013 | |
|
|
1014 | Source code for this benchmark is found as eg/bench2 in the AnyEvent |
|
|
1015 | distribution. |
|
|
1016 | |
|
|
1017 | Explanation of the columns |
|
|
1018 | *sockets* is the number of sockets, and twice the number of "servers" |
|
|
1019 | (as each server has a read and write socket end). |
|
|
1020 | |
|
|
1021 | *create* is the time it takes to create a socketpair (which is |
|
|
1022 | nontrivial) and two watchers: an I/O watcher and a timeout watcher. |
|
|
1023 | |
|
|
1024 | *request*, the most important value, is the time it takes to handle a |
|
|
1025 | single "request", that is, reading the token from the pipe and |
|
|
1026 | forwarding it to another server. This includes deleting the old timeout |
|
|
1027 | and creating a new one that moves the timeout into the future. |
|
|
1028 | |
|
|
1029 | Results |
|
|
1030 | name sockets create request |
|
|
1031 | EV 20000 69.01 11.16 |
|
|
1032 | Perl 20000 73.32 35.87 |
|
|
1033 | Event 20000 212.62 257.32 |
|
|
1034 | Glib 20000 651.16 1896.30 |
|
|
1035 | POE 20000 349.67 12317.24 uses POE::Loop::Event |
|
|
1036 | |
|
|
1037 | Discussion |
|
|
1038 | This benchmark *does* measure scalability and overall performance of the |
|
|
1039 | particular event loop. |
|
|
1040 | |
|
|
1041 | EV is again fastest. Since it is using epoll on my system, the setup |
|
|
1042 | time is relatively high, though. |
|
|
1043 | |
|
|
1044 | Perl surprisingly comes second. It is much faster than the C-based event |
|
|
1045 | loops Event and Glib. |
|
|
1046 | |
|
|
1047 | Event suffers from high setup time as well (look at its code and you |
|
|
1048 | will understand why). Callback invocation also has a high overhead |
|
|
1049 | compared to the "$_->() for .."-style loop that the Perl event loop |
|
|
1050 | uses. Event uses select or poll in basically all documented |
|
|
1051 | configurations. |
|
|
1052 | |
|
|
1053 | Glib is hit hard by its quadratic behaviour w.r.t. many watchers. It |
|
|
1054 | clearly fails to perform with many filehandles or in busy servers. |
|
|
1055 | |
|
|
1056 | POE is still completely out of the picture, taking over 1000 times as |
|
|
1057 | long as EV, and over 100 times as long as the Perl implementation, even |
|
|
1058 | though it uses a C-based event loop in this case. |
|
|
1059 | |
|
|
1060 | Summary |
|
|
1061 | * The pure perl implementation performs extremely well. |
|
|
1062 | |
|
|
1063 | * Avoid Glib or POE in large projects where performance matters. |
|
|
1064 | |
|
|
1065 | BENCHMARKING SMALL SERVERS |
|
|
1066 | While event loops should scale (and select-based ones do not...) even to |
|
|
1067 | large servers, most programs we (or I :) actually write have only a few |
|
|
1068 | I/O watchers. |
|
|
1069 | |
|
|
1070 | In this benchmark, I use the same benchmark program as in the large |
|
|
1071 | server case, but it uses only eight "servers", of which three are active |
|
|
1072 | at any one time. This should reflect performance for a small server |
|
|
1073 | relatively well. |
|
|
1074 | |
|
|
1075 | The columns are identical to the previous table. |
|
|
1076 | |
|
|
1077 | Results |
|
|
1078 | name sockets create request |
|
|
1079 | EV 16 20.00 6.54 |
|
|
1080 | Perl 16 25.75 12.62 |
|
|
1081 | Event 16 81.27 35.86 |
|
|
1082 | Glib 16 32.63 15.48 |
|
|
1083 | POE 16 261.87 276.28 uses POE::Loop::Event |
|
|
1084 | |
|
|
1085 | Discussion |
|
|
1086 | The benchmark tries to test the performance of a typical small server. |
|
|
1087 | While knowing how various event loops perform is interesting, keep in |
|
|
1088 | mind that their overhead in this case is usually not as important, due |
|
|
1089 | to the small absolute number of watchers (that is, you need efficiency |
|
|
1090 | and speed most when you have lots of watchers, not when you only have a |
|
|
1091 | few of them). |
|
|
1092 | |
|
|
1093 | EV is again fastest. |
|
|
1094 | |
|
|
1095 | Perl again comes second. It is noticably faster than the C-based event |
|
|
1096 | loops Event and Glib, although the difference is too small to really |
|
|
1097 | matter. |
|
|
1098 | |
|
|
1099 | POE also performs much better in this case, but is is still far behind |
|
|
1100 | the others. |
|
|
1101 | |
|
|
1102 | Summary |
|
|
1103 | * C-based event loops perform very well with small number of watchers, |
|
|
1104 | as the management overhead dominates. |
|
|
1105 | |
|
|
1106 | FORK |
|
|
1107 | Most event libraries are not fork-safe. The ones who are usually are |
|
|
1108 | because they rely on inefficient but fork-safe "select" or "poll" calls. |
|
|
1109 | Only EV is fully fork-aware. |
|
|
1110 | |
|
|
1111 | If you have to fork, you must either do so *before* creating your first |
|
|
1112 | watcher OR you must not use AnyEvent at all in the child. |
|
|
1113 | |
|
|
1114 | SECURITY CONSIDERATIONS |
|
|
1115 | AnyEvent can be forced to load any event model via |
|
|
1116 | $ENV{PERL_ANYEVENT_MODEL}. While this cannot (to my knowledge) be used |
|
|
1117 | to execute arbitrary code or directly gain access, it can easily be used |
|
|
1118 | to make the program hang or malfunction in subtle ways, as AnyEvent |
|
|
1119 | watchers will not be active when the program uses a different event |
|
|
1120 | model than specified in the variable. |
|
|
1121 | |
|
|
1122 | You can make AnyEvent completely ignore this variable by deleting it |
|
|
1123 | before the first watcher gets created, e.g. with a "BEGIN" block: |
|
|
1124 | |
|
|
1125 | BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} } |
|
|
1126 | |
|
|
1127 | use AnyEvent; |
|
|
1128 | |
|
|
1129 | Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can |
|
|
1130 | be used to probe what backend is used and gain other information (which |
|
|
1131 | is probably even less useful to an attacker than PERL_ANYEVENT_MODEL). |
592 | |
1132 | |
593 | SEE ALSO |
1133 | SEE ALSO |
594 | Event modules: Coro::EV, EV, EV::Glib, Glib::EV, Coro::Event, Event, |
1134 | Event modules: EV, EV::Glib, Glib::EV, Event, Glib::Event, Glib, Tk, |
595 | Glib::Event, Glib, Coro, Tk. |
1135 | Event::Lib, Qt, POE. |
596 | |
1136 | |
597 | Implementations: AnyEvent::Impl::CoroEV, AnyEvent::Impl::EV, |
1137 | Implementations: AnyEvent::Impl::EV, AnyEvent::Impl::Event, |
598 | AnyEvent::Impl::CoroEvent, AnyEvent::Impl::Event, AnyEvent::Impl::Glib, |
|
|
599 | AnyEvent::Impl::Tk, AnyEvent::Impl::Perl. |
1138 | AnyEvent::Impl::Glib, AnyEvent::Impl::Tk, AnyEvent::Impl::Perl, |
|
|
1139 | AnyEvent::Impl::EventLib, AnyEvent::Impl::Qt, AnyEvent::Impl::POE. |
|
|
1140 | |
|
|
1141 | Coroutine support: Coro, Coro::AnyEvent, Coro::EV, Coro::Event, |
600 | |
1142 | |
601 | Nontrivial usage examples: Net::FCP, Net::XMPP2. |
1143 | Nontrivial usage examples: Net::FCP, Net::XMPP2. |
602 | |
1144 | |
603 | AUTHOR |
1145 | AUTHOR |
604 | Marc Lehmann <schmorp@schmorp.de> |
1146 | Marc Lehmann <schmorp@schmorp.de> |