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

Comparing AnyEvent/lib/AnyEvent.pm (file contents):
Revision 1.117 by root, Sun May 11 17:54:13 2008 UTC vs.
Revision 1.254 by root, Fri Jul 24 08:40:35 2009 UTC

1=head1 NAME 1=head1 NAME
2 2
3AnyEvent - provide framework for multiple event loops 3AnyEvent - events independent of event loop implementation
4 4
5EV, Event, Glib, Tk, Perl, Event::Lib, Qt, POE - various supported event loops 5EV, Event, Glib, Tk, Perl, Event::Lib, Qt and POE are various supported
6event loops.
6 7
7=head1 SYNOPSIS 8=head1 SYNOPSIS
8 9
9 use AnyEvent; 10 use AnyEvent;
10 11
12 # file descriptor readable
11 my $w = AnyEvent->io (fh => $fh, poll => "r|w", cb => sub { 13 my $w = AnyEvent->io (fh => $fh, poll => "r", cb => sub { ... });
14
15 # one-shot or repeating timers
16 my $w = AnyEvent->timer (after => $seconds, cb => sub { ... });
17 my $w = AnyEvent->timer (after => $seconds, interval => $seconds, cb => ...
18
19 print AnyEvent->now; # prints current event loop time
20 print AnyEvent->time; # think Time::HiRes::time or simply CORE::time.
21
22 # POSIX signal
23 my $w = AnyEvent->signal (signal => "TERM", cb => sub { ... });
24
25 # child process exit
26 my $w = AnyEvent->child (pid => $pid, cb => sub {
27 my ($pid, $status) = @_;
12 ... 28 ...
13 }); 29 });
14 30
15 my $w = AnyEvent->timer (after => $seconds, cb => sub { 31 # called when event loop idle (if applicable)
16 ... 32 my $w = AnyEvent->idle (cb => sub { ... });
17 });
18 33
19 my $w = AnyEvent->condvar; # stores whether a condition was flagged 34 my $w = AnyEvent->condvar; # stores whether a condition was flagged
20 $w->send; # wake up current and all future recv's 35 $w->send; # wake up current and all future recv's
21 $w->recv; # enters "main loop" till $condvar gets ->send 36 $w->recv; # enters "main loop" till $condvar gets ->send
37 # use a condvar in callback mode:
38 $w->cb (sub { $_[0]->recv });
39
40=head1 INTRODUCTION/TUTORIAL
41
42This manpage is mainly a reference manual. If you are interested
43in a tutorial or some gentle introduction, have a look at the
44L<AnyEvent::Intro> manpage.
45
46=head1 SUPPORT
47
48There is a mailinglist for discussing all things AnyEvent, and an IRC
49channel, too.
50
51See the AnyEvent project page at the B<Schmorpforge Ta-Sa Software
52Respository>, at L<http://anyevent.schmorp.de>, for more info.
22 53
23=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT) 54=head1 WHY YOU SHOULD USE THIS MODULE (OR NOT)
24 55
25Glib, POE, IO::Async, Event... CPAN offers event models by the dozen 56Glib, POE, IO::Async, Event... CPAN offers event models by the dozen
26nowadays. So what is different about AnyEvent? 57nowadays. So what is different about AnyEvent?
27 58
28Executive Summary: AnyEvent is I<compatible>, AnyEvent is I<free of 59Executive Summary: AnyEvent is I<compatible>, AnyEvent is I<free of
29policy> and AnyEvent is I<small and efficient>. 60policy> and AnyEvent is I<small and efficient>.
30 61
31First and foremost, I<AnyEvent is not an event model> itself, it only 62First and foremost, I<AnyEvent is not an event model> itself, it only
32interfaces to whatever event model the main program happens to use in a 63interfaces to whatever event model the main program happens to use, in a
33pragmatic way. For event models and certain classes of immortals alike, 64pragmatic way. For event models and certain classes of immortals alike,
34the statement "there can only be one" is a bitter reality: In general, 65the statement "there can only be one" is a bitter reality: In general,
35only one event loop can be active at the same time in a process. AnyEvent 66only one event loop can be active at the same time in a process. AnyEvent
36helps hiding the differences between those event loops. 67cannot change this, but it can hide the differences between those event
68loops.
37 69
38The goal of AnyEvent is to offer module authors the ability to do event 70The goal of AnyEvent is to offer module authors the ability to do event
39programming (waiting for I/O or timer events) without subscribing to a 71programming (waiting for I/O or timer events) without subscribing to a
40religion, a way of living, and most importantly: without forcing your 72religion, a way of living, and most importantly: without forcing your
41module users into the same thing by forcing them to use the same event 73module users into the same thing by forcing them to use the same event
42model you use. 74model you use.
43 75
44For modules like POE or IO::Async (which is a total misnomer as it is 76For modules like POE or IO::Async (which is a total misnomer as it is
45actually doing all I/O I<synchronously>...), using them in your module is 77actually doing all I/O I<synchronously>...), using them in your module is
46like joining a cult: After you joined, you are dependent on them and you 78like joining a cult: After you joined, you are dependent on them and you
47cannot use anything else, as it is simply incompatible to everything that 79cannot use anything else, as they are simply incompatible to everything
48isn't itself. What's worse, all the potential users of your module are 80that isn't them. What's worse, all the potential users of your
49I<also> forced to use the same event loop you use. 81module are I<also> forced to use the same event loop you use.
50 82
51AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works 83AnyEvent is different: AnyEvent + POE works fine. AnyEvent + Glib works
52fine. AnyEvent + Tk works fine etc. etc. but none of these work together 84fine. AnyEvent + Tk works fine etc. etc. but none of these work together
53with the rest: POE + IO::Async? no go. Tk + Event? no go. Again: if 85with the rest: POE + IO::Async? No go. Tk + Event? No go. Again: if
54your module uses one of those, every user of your module has to use it, 86your module uses one of those, every user of your module has to use it,
55too. But if your module uses AnyEvent, it works transparently with all 87too. But if your module uses AnyEvent, it works transparently with all
56event models it supports (including stuff like POE and IO::Async, as long 88event models it supports (including stuff like IO::Async, as long as those
57as those use one of the supported event loops. It is trivial to add new 89use one of the supported event loops. It is trivial to add new event loops
58event loops to AnyEvent, too, so it is future-proof). 90to AnyEvent, too, so it is future-proof).
59 91
60In addition to being free of having to use I<the one and only true event 92In addition to being free of having to use I<the one and only true event
61model>, AnyEvent also is free of bloat and policy: with POE or similar 93model>, AnyEvent also is free of bloat and policy: with POE or similar
62modules, you get an enourmous amount of code and strict rules you have to 94modules, you get an enormous amount of code and strict rules you have to
63follow. AnyEvent, on the other hand, is lean and up to the point, by only 95follow. AnyEvent, on the other hand, is lean and up to the point, by only
64offering the functionality that is necessary, in as thin as a wrapper as 96offering the functionality that is necessary, in as thin as a wrapper as
65technically possible. 97technically possible.
66 98
99Of course, AnyEvent comes with a big (and fully optional!) toolbox
100of useful functionality, such as an asynchronous DNS resolver, 100%
101non-blocking connects (even with TLS/SSL, IPv6 and on broken platforms
102such as Windows) and lots of real-world knowledge and workarounds for
103platform bugs and differences.
104
67Of course, if you want lots of policy (this can arguably be somewhat 105Now, if you I<do want> lots of policy (this can arguably be somewhat
68useful) and you want to force your users to use the one and only event 106useful) and you want to force your users to use the one and only event
69model, you should I<not> use this module. 107model, you should I<not> use this module.
70 108
71=head1 DESCRIPTION 109=head1 DESCRIPTION
72 110
102starts using it, all bets are off. Maybe you should tell their authors to 140starts using it, all bets are off. Maybe you should tell their authors to
103use AnyEvent so their modules work together with others seamlessly... 141use AnyEvent so their modules work together with others seamlessly...
104 142
105The pure-perl implementation of AnyEvent is called 143The pure-perl implementation of AnyEvent is called
106C<AnyEvent::Impl::Perl>. Like other event modules you can load it 144C<AnyEvent::Impl::Perl>. Like other event modules you can load it
107explicitly. 145explicitly and enjoy the high availability of that event loop :)
108 146
109=head1 WATCHERS 147=head1 WATCHERS
110 148
111AnyEvent has the central concept of a I<watcher>, which is an object that 149AnyEvent has the central concept of a I<watcher>, which is an object that
112stores relevant data for each kind of event you are waiting for, such as 150stores relevant data for each kind of event you are waiting for, such as
113the callback to call, the filehandle to watch, etc. 151the callback to call, the file handle to watch, etc.
114 152
115These watchers are normal Perl objects with normal Perl lifetime. After 153These watchers are normal Perl objects with normal Perl lifetime. After
116creating a watcher it will immediately "watch" for events and invoke the 154creating a watcher it will immediately "watch" for events and invoke the
117callback when the event occurs (of course, only when the event model 155callback when the event occurs (of course, only when the event model
118is in control). 156is in control).
119 157
158Note that B<callbacks must not permanently change global variables>
159potentially in use by the event loop (such as C<$_> or C<$[>) and that B<<
160callbacks must not C<die> >>. The former is good programming practise in
161Perl and the latter stems from the fact that exception handling differs
162widely between event loops.
163
120To disable the watcher you have to destroy it (e.g. by setting the 164To disable the watcher you have to destroy it (e.g. by setting the
121variable you store it in to C<undef> or otherwise deleting all references 165variable you store it in to C<undef> or otherwise deleting all references
122to it). 166to it).
123 167
124All watchers are created by calling a method on the C<AnyEvent> class. 168All watchers are created by calling a method on the C<AnyEvent> class.
126Many watchers either are used with "recursion" (repeating timers for 170Many watchers either are used with "recursion" (repeating timers for
127example), or need to refer to their watcher object in other ways. 171example), or need to refer to their watcher object in other ways.
128 172
129An any way to achieve that is this pattern: 173An any way to achieve that is this pattern:
130 174
131 my $w; $w = AnyEvent->type (arg => value ..., cb => sub { 175 my $w; $w = AnyEvent->type (arg => value ..., cb => sub {
132 # you can use $w here, for example to undef it 176 # you can use $w here, for example to undef it
133 undef $w; 177 undef $w;
134 }); 178 });
135 179
136Note that C<my $w; $w => combination. This is necessary because in Perl, 180Note that C<my $w; $w => combination. This is necessary because in Perl,
137my variables are only visible after the statement in which they are 181my variables are only visible after the statement in which they are
138declared. 182declared.
139 183
140=head2 I/O WATCHERS 184=head2 I/O WATCHERS
141 185
142You can create an I/O watcher by calling the C<< AnyEvent->io >> method 186You can create an I/O watcher by calling the C<< AnyEvent->io >> method
143with the following mandatory key-value pairs as arguments: 187with the following mandatory key-value pairs as arguments:
144 188
145C<fh> the Perl I<file handle> (I<not> file descriptor) to watch 189C<fh> is the Perl I<file handle> (or a naked file descriptor) to watch
190for events (AnyEvent might or might not keep a reference to this file
191handle). Note that only file handles pointing to things for which
192non-blocking operation makes sense are allowed. This includes sockets,
193most character devices, pipes, fifos and so on, but not for example files
194or block devices.
195
146for events. C<poll> must be a string that is either C<r> or C<w>, 196C<poll> must be a string that is either C<r> or C<w>, which creates a
147which creates a watcher waiting for "r"eadable or "w"ritable events, 197watcher waiting for "r"eadable or "w"ritable events, respectively.
198
148respectively. C<cb> is the callback to invoke each time the file handle 199C<cb> is the callback to invoke each time the file handle becomes ready.
149becomes ready.
150 200
151Although the callback might get passed parameters, their value and 201Although the callback might get passed parameters, their value and
152presence is undefined and you cannot rely on them. Portable AnyEvent 202presence is undefined and you cannot rely on them. Portable AnyEvent
153callbacks cannot use arguments passed to I/O watcher callbacks. 203callbacks cannot use arguments passed to I/O watcher callbacks.
154 204
158 208
159Some event loops issue spurious readyness notifications, so you should 209Some event loops issue spurious readyness notifications, so you should
160always use non-blocking calls when reading/writing from/to your file 210always use non-blocking calls when reading/writing from/to your file
161handles. 211handles.
162 212
163Example:
164
165 # wait for readability of STDIN, then read a line and disable the watcher 213Example: wait for readability of STDIN, then read a line and disable the
214watcher.
215
166 my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub { 216 my $w; $w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
167 chomp (my $input = <STDIN>); 217 chomp (my $input = <STDIN>);
168 warn "read: $input\n"; 218 warn "read: $input\n";
169 undef $w; 219 undef $w;
170 }); 220 });
180 230
181Although the callback might get passed parameters, their value and 231Although the callback might get passed parameters, their value and
182presence is undefined and you cannot rely on them. Portable AnyEvent 232presence is undefined and you cannot rely on them. Portable AnyEvent
183callbacks cannot use arguments passed to time watcher callbacks. 233callbacks cannot use arguments passed to time watcher callbacks.
184 234
185The timer callback will be invoked at most once: if you want a repeating 235The callback will normally be invoked once only. If you specify another
186timer you have to create a new watcher (this is a limitation by both Tk 236parameter, C<interval>, as a strictly positive number (> 0), then the
187and Glib). 237callback will be invoked regularly at that interval (in fractional
238seconds) after the first invocation. If C<interval> is specified with a
239false value, then it is treated as if it were missing.
188 240
189Example: 241The callback will be rescheduled before invoking the callback, but no
242attempt is done to avoid timer drift in most backends, so the interval is
243only approximate.
190 244
191 # fire an event after 7.7 seconds 245Example: fire an event after 7.7 seconds.
246
192 my $w = AnyEvent->timer (after => 7.7, cb => sub { 247 my $w = AnyEvent->timer (after => 7.7, cb => sub {
193 warn "timeout\n"; 248 warn "timeout\n";
194 }); 249 });
195 250
196 # to cancel the timer: 251 # to cancel the timer:
197 undef $w; 252 undef $w;
198 253
199Example 2:
200
201 # fire an event after 0.5 seconds, then roughly every second 254Example 2: fire an event after 0.5 seconds, then roughly every second.
202 my $w;
203 255
204 my $cb = sub {
205 # cancel the old timer while creating a new one
206 $w = AnyEvent->timer (after => 1, cb => $cb); 256 my $w = AnyEvent->timer (after => 0.5, interval => 1, cb => sub {
257 warn "timeout\n";
207 }; 258 };
208
209 # start the "loop" by creating the first watcher
210 $w = AnyEvent->timer (after => 0.5, cb => $cb);
211 259
212=head3 TIMING ISSUES 260=head3 TIMING ISSUES
213 261
214There are two ways to handle timers: based on real time (relative, "fire 262There are two ways to handle timers: based on real time (relative, "fire
215in 10 seconds") and based on wallclock time (absolute, "fire at 12 263in 10 seconds") and based on wallclock time (absolute, "fire at 12
227timers. 275timers.
228 276
229AnyEvent always prefers relative timers, if available, matching the 277AnyEvent always prefers relative timers, if available, matching the
230AnyEvent API. 278AnyEvent API.
231 279
280AnyEvent has two additional methods that return the "current time":
281
282=over 4
283
284=item AnyEvent->time
285
286This returns the "current wallclock time" as a fractional number of
287seconds since the Epoch (the same thing as C<time> or C<Time::HiRes::time>
288return, and the result is guaranteed to be compatible with those).
289
290It progresses independently of any event loop processing, i.e. each call
291will check the system clock, which usually gets updated frequently.
292
293=item AnyEvent->now
294
295This also returns the "current wallclock time", but unlike C<time>, above,
296this value might change only once per event loop iteration, depending on
297the event loop (most return the same time as C<time>, above). This is the
298time that AnyEvent's timers get scheduled against.
299
300I<In almost all cases (in all cases if you don't care), this is the
301function to call when you want to know the current time.>
302
303This function is also often faster then C<< AnyEvent->time >>, and
304thus the preferred method if you want some timestamp (for example,
305L<AnyEvent::Handle> uses this to update it's activity timeouts).
306
307The rest of this section is only of relevance if you try to be very exact
308with your timing, you can skip it without bad conscience.
309
310For a practical example of when these times differ, consider L<Event::Lib>
311and L<EV> and the following set-up:
312
313The event loop is running and has just invoked one of your callback at
314time=500 (assume no other callbacks delay processing). In your callback,
315you wait a second by executing C<sleep 1> (blocking the process for a
316second) and then (at time=501) you create a relative timer that fires
317after three seconds.
318
319With L<Event::Lib>, C<< AnyEvent->time >> and C<< AnyEvent->now >> will
320both return C<501>, because that is the current time, and the timer will
321be scheduled to fire at time=504 (C<501> + C<3>).
322
323With L<EV>, C<< AnyEvent->time >> returns C<501> (as that is the current
324time), but C<< AnyEvent->now >> returns C<500>, as that is the time the
325last event processing phase started. With L<EV>, your timer gets scheduled
326to run at time=503 (C<500> + C<3>).
327
328In one sense, L<Event::Lib> is more exact, as it uses the current time
329regardless of any delays introduced by event processing. However, most
330callbacks do not expect large delays in processing, so this causes a
331higher drift (and a lot more system calls to get the current time).
332
333In another sense, L<EV> is more exact, as your timer will be scheduled at
334the same time, regardless of how long event processing actually took.
335
336In either case, if you care (and in most cases, you don't), then you
337can get whatever behaviour you want with any event loop, by taking the
338difference between C<< AnyEvent->time >> and C<< AnyEvent->now >> into
339account.
340
341=item AnyEvent->now_update
342
343Some event loops (such as L<EV> or L<AnyEvent::Impl::Perl>) cache
344the current time for each loop iteration (see the discussion of L<<
345AnyEvent->now >>, above).
346
347When a callback runs for a long time (or when the process sleeps), then
348this "current" time will differ substantially from the real time, which
349might affect timers and time-outs.
350
351When this is the case, you can call this method, which will update the
352event loop's idea of "current time".
353
354Note that updating the time I<might> cause some events to be handled.
355
356=back
357
232=head2 SIGNAL WATCHERS 358=head2 SIGNAL WATCHERS
233 359
234You can watch for signals using a signal watcher, C<signal> is the signal 360You can watch for signals using a signal watcher, C<signal> is the signal
235I<name> without any C<SIG> prefix, C<cb> is the Perl callback to 361I<name> in uppercase and without any C<SIG> prefix, C<cb> is the Perl
236be invoked whenever a signal occurs. 362callback to be invoked whenever a signal occurs.
237 363
238Although the callback might get passed parameters, their value and 364Although the callback might get passed parameters, their value and
239presence is undefined and you cannot rely on them. Portable AnyEvent 365presence is undefined and you cannot rely on them. Portable AnyEvent
240callbacks cannot use arguments passed to signal watcher callbacks. 366callbacks cannot use arguments passed to signal watcher callbacks.
241 367
242Multiple signal occurances can be clumped together into one callback 368Multiple signal occurrences can be clumped together into one callback
243invocation, and callback invocation will be synchronous. synchronous means 369invocation, and callback invocation will be synchronous. Synchronous means
244that it might take a while until the signal gets handled by the process, 370that it might take a while until the signal gets handled by the process,
245but it is guarenteed not to interrupt any other callbacks. 371but it is guaranteed not to interrupt any other callbacks.
246 372
247The main advantage of using these watchers is that you can share a signal 373The main advantage of using these watchers is that you can share a signal
248between multiple watchers. 374between multiple watchers, and AnyEvent will ensure that signals will not
375interrupt your program at bad times.
249 376
250This watcher might use C<%SIG>, so programs overwriting those signals 377This watcher might use C<%SIG> (depending on the event loop used),
251directly will likely not work correctly. 378so programs overwriting those signals directly will likely not work
379correctly.
252 380
253Example: exit on SIGINT 381Example: exit on SIGINT
254 382
255 my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 }); 383 my $w = AnyEvent->signal (signal => "INT", cb => sub { exit 1 });
256 384
385=head3 Signal Races, Delays and Workarounds
386
387Many event loops (e.g. Glib, Tk, Qt, IO::Async) do not support attaching
388callbacks to signals in a generic way, which is a pity, as you cannot do
389race-free signal handling in perl. AnyEvent will try to do it's best, but
390in some cases, signals will be delayed. The maximum time a signal might
391be delayed is specified in C<$AnyEvent::MAX_SIGNAL_LATENCY> (default: 10
392seconds). This variable can be changed only before the first signal
393watcher is created, and should be left alone otherwise. Higher values
394will cause fewer spurious wake-ups, which is better for power and CPU
395saving. All these problems can be avoided by installing the optional
396L<Async::Interrupt> module. This will not work with inherently broken
397event loops such as L<Event> or L<Event::Lib> (and not with L<POE>
398currently, as POE does it's own workaround with one-second latency). With
399those, you just have to suffer the delays.
400
257=head2 CHILD PROCESS WATCHERS 401=head2 CHILD PROCESS WATCHERS
258 402
259You can also watch on a child process exit and catch its exit status. 403You can also watch on a child process exit and catch its exit status.
260 404
261The child process is specified by the C<pid> argument (if set to C<0>, it 405The child process is specified by the C<pid> argument (one some backends,
262watches for any child process exit). The watcher will trigger as often 406using C<0> watches for any child process exit, on others this will
263as status change for the child are received. This works by installing a 407croak). The watcher will be triggered only when the child process has
264signal handler for C<SIGCHLD>. The callback will be called with the pid 408finished and an exit status is available, not on any trace events
265and exit status (as returned by waitpid), so unlike other watcher types, 409(stopped/continued).
266you I<can> rely on child watcher callback arguments. 410
411The callback will be called with the pid and exit status (as returned by
412waitpid), so unlike other watcher types, you I<can> rely on child watcher
413callback arguments.
414
415This watcher type works by installing a signal handler for C<SIGCHLD>,
416and since it cannot be shared, nothing else should use SIGCHLD or reap
417random child processes (waiting for specific child processes, e.g. inside
418C<system>, is just fine).
267 419
268There is a slight catch to child watchers, however: you usually start them 420There is a slight catch to child watchers, however: you usually start them
269I<after> the child process was created, and this means the process could 421I<after> the child process was created, and this means the process could
270have exited already (and no SIGCHLD will be sent anymore). 422have exited already (and no SIGCHLD will be sent anymore).
271 423
272Not all event models handle this correctly (POE doesn't), but even for 424Not all event models handle this correctly (neither POE nor IO::Async do,
425see their AnyEvent::Impl manpages for details), but even for event models
273event models that I<do> handle this correctly, they usually need to be 426that I<do> handle this correctly, they usually need to be loaded before
274loaded before the process exits (i.e. before you fork in the first place). 427the process exits (i.e. before you fork in the first place). AnyEvent's
428pure perl event loop handles all cases correctly regardless of when you
429start the watcher.
275 430
276This means you cannot create a child watcher as the very first thing in an 431This means you cannot create a child watcher as the very first
277AnyEvent program, you I<have> to create at least one watcher before you 432thing in an AnyEvent program, you I<have> to create at least one
278C<fork> the child (alternatively, you can call C<AnyEvent::detect>). 433watcher before you C<fork> the child (alternatively, you can call
434C<AnyEvent::detect>).
435
436As most event loops do not support waiting for child events, they will be
437emulated by AnyEvent in most cases, in which the latency and race problems
438mentioned in the description of signal watchers apply.
279 439
280Example: fork a process and wait for it 440Example: fork a process and wait for it
281 441
282 my $done = AnyEvent->condvar; 442 my $done = AnyEvent->condvar;
283 443
284 my $pid = fork or exit 5; 444 my $pid = fork or exit 5;
285 445
286 my $w = AnyEvent->child ( 446 my $w = AnyEvent->child (
287 pid => $pid, 447 pid => $pid,
288 cb => sub { 448 cb => sub {
289 my ($pid, $status) = @_; 449 my ($pid, $status) = @_;
290 warn "pid $pid exited with status $status"; 450 warn "pid $pid exited with status $status";
291 $done->send; 451 $done->send;
292 }, 452 },
293 ); 453 );
294 454
295 # do something else, then wait for process exit 455 # do something else, then wait for process exit
296 $done->recv; 456 $done->recv;
457
458=head2 IDLE WATCHERS
459
460Sometimes there is a need to do something, but it is not so important
461to do it instantly, but only when there is nothing better to do. This
462"nothing better to do" is usually defined to be "no other events need
463attention by the event loop".
464
465Idle watchers ideally get invoked when the event loop has nothing
466better to do, just before it would block the process to wait for new
467events. Instead of blocking, the idle watcher is invoked.
468
469Most event loops unfortunately do not really support idle watchers (only
470EV, Event and Glib do it in a usable fashion) - for the rest, AnyEvent
471will simply call the callback "from time to time".
472
473Example: read lines from STDIN, but only process them when the
474program is otherwise idle:
475
476 my @lines; # read data
477 my $idle_w;
478 my $io_w = AnyEvent->io (fh => \*STDIN, poll => 'r', cb => sub {
479 push @lines, scalar <STDIN>;
480
481 # start an idle watcher, if not already done
482 $idle_w ||= AnyEvent->idle (cb => sub {
483 # handle only one line, when there are lines left
484 if (my $line = shift @lines) {
485 print "handled when idle: $line";
486 } else {
487 # otherwise disable the idle watcher again
488 undef $idle_w;
489 }
490 });
491 });
297 492
298=head2 CONDITION VARIABLES 493=head2 CONDITION VARIABLES
299 494
300If you are familiar with some event loops you will know that all of them 495If you are familiar with some event loops you will know that all of them
301require you to run some blocking "loop", "run" or similar function that 496require you to run some blocking "loop", "run" or similar function that
302will actively watch for new events and call your callbacks. 497will actively watch for new events and call your callbacks.
303 498
304AnyEvent is different, it expects somebody else to run the event loop and 499AnyEvent is slightly different: it expects somebody else to run the event
305will only block when necessary (usually when told by the user). 500loop and will only block when necessary (usually when told by the user).
306 501
307The instrument to do that is called a "condition variable", so called 502The instrument to do that is called a "condition variable", so called
308because they represent a condition that must become true. 503because they represent a condition that must become true.
504
505Now is probably a good time to look at the examples further below.
309 506
310Condition variables can be created by calling the C<< AnyEvent->condvar 507Condition variables can be created by calling the C<< AnyEvent->condvar
311>> method, usually without arguments. The only argument pair allowed is 508>> method, usually without arguments. The only argument pair allowed is
312C<cb>, which specifies a callback to be called when the condition variable 509C<cb>, which specifies a callback to be called when the condition variable
313becomes true. 510becomes true, with the condition variable as the first argument (but not
511the results).
314 512
315After creation, the conditon variable is "false" until it becomes "true" 513After creation, the condition variable is "false" until it becomes "true"
316by calling the C<send> method. 514by calling the C<send> method (or calling the condition variable as if it
515were a callback, read about the caveats in the description for the C<<
516->send >> method).
317 517
318Condition variables are similar to callbacks, except that you can 518Condition variables are similar to callbacks, except that you can
319optionally wait for them. They can also be called merge points - points 519optionally wait for them. They can also be called merge points - points
320in time where multiple outstandign events have been processed. And yet 520in time where multiple outstanding events have been processed. And yet
321another way to call them is transations - each condition variable can be 521another way to call them is transactions - each condition variable can be
322used to represent a transaction, which finishes at some point and delivers 522used to represent a transaction, which finishes at some point and delivers
323a result. 523a result. And yet some people know them as "futures" - a promise to
524compute/deliver something that you can wait for.
324 525
325Condition variables are very useful to signal that something has finished, 526Condition variables are very useful to signal that something has finished,
326for example, if you write a module that does asynchronous http requests, 527for example, if you write a module that does asynchronous http requests,
327then a condition variable would be the ideal candidate to signal the 528then a condition variable would be the ideal candidate to signal the
328availability of results. The user can either act when the callback is 529availability of results. The user can either act when the callback is
332you can block your main program until an event occurs - for example, you 533you can block your main program until an event occurs - for example, you
333could C<< ->recv >> in your main program until the user clicks the Quit 534could C<< ->recv >> in your main program until the user clicks the Quit
334button of your app, which would C<< ->send >> the "quit" event. 535button of your app, which would C<< ->send >> the "quit" event.
335 536
336Note that condition variables recurse into the event loop - if you have 537Note that condition variables recurse into the event loop - if you have
337two pieces of code that call C<< ->recv >> in a round-robbin fashion, you 538two pieces of code that call C<< ->recv >> in a round-robin fashion, you
338lose. Therefore, condition variables are good to export to your caller, but 539lose. Therefore, condition variables are good to export to your caller, but
339you should avoid making a blocking wait yourself, at least in callbacks, 540you should avoid making a blocking wait yourself, at least in callbacks,
340as this asks for trouble. 541as this asks for trouble.
341 542
342Condition variables are represented by hash refs in perl, and the keys 543Condition variables are represented by hash refs in perl, and the keys
347 548
348There are two "sides" to a condition variable - the "producer side" which 549There are two "sides" to a condition variable - the "producer side" which
349eventually calls C<< -> send >>, and the "consumer side", which waits 550eventually calls C<< -> send >>, and the "consumer side", which waits
350for the send to occur. 551for the send to occur.
351 552
352Example: 553Example: wait for a timer.
353 554
354 # wait till the result is ready 555 # wait till the result is ready
355 my $result_ready = AnyEvent->condvar; 556 my $result_ready = AnyEvent->condvar;
356 557
357 # do something such as adding a timer 558 # do something such as adding a timer
362 after => 1, 563 after => 1,
363 cb => sub { $result_ready->send }, 564 cb => sub { $result_ready->send },
364 ); 565 );
365 566
366 # this "blocks" (while handling events) till the callback 567 # this "blocks" (while handling events) till the callback
367 # calls send 568 # calls -<send
368 $result_ready->recv; 569 $result_ready->recv;
570
571Example: wait for a timer, but take advantage of the fact that condition
572variables are also callable directly.
573
574 my $done = AnyEvent->condvar;
575 my $delay = AnyEvent->timer (after => 5, cb => $done);
576 $done->recv;
577
578Example: Imagine an API that returns a condvar and doesn't support
579callbacks. This is how you make a synchronous call, for example from
580the main program:
581
582 use AnyEvent::CouchDB;
583
584 ...
585
586 my @info = $couchdb->info->recv;
587
588And this is how you would just set a callback to be called whenever the
589results are available:
590
591 $couchdb->info->cb (sub {
592 my @info = $_[0]->recv;
593 });
369 594
370=head3 METHODS FOR PRODUCERS 595=head3 METHODS FOR PRODUCERS
371 596
372These methods should only be used by the producing side, i.e. the 597These methods should only be used by the producing side, i.e. the
373code/module that eventually sends the signal. Note that it is also 598code/module that eventually sends the signal. Note that it is also
386immediately from within send. 611immediately from within send.
387 612
388Any arguments passed to the C<send> call will be returned by all 613Any arguments passed to the C<send> call will be returned by all
389future C<< ->recv >> calls. 614future C<< ->recv >> calls.
390 615
616Condition variables are overloaded so one can call them directly (as if
617they were a code reference). Calling them directly is the same as calling
618C<send>.
619
391=item $cv->croak ($error) 620=item $cv->croak ($error)
392 621
393Similar to send, but causes all call's to C<< ->recv >> to invoke 622Similar to send, but causes all call's to C<< ->recv >> to invoke
394C<Carp::croak> with the given error message/object/scalar. 623C<Carp::croak> with the given error message/object/scalar.
395 624
396This can be used to signal any errors to the condition variable 625This can be used to signal any errors to the condition variable
397user/consumer. 626user/consumer. Doing it this way instead of calling C<croak> directly
627delays the error detetcion, but has the overwhelmign advantage that it
628diagnoses the error at the place where the result is expected, and not
629deep in some event clalback without connection to the actual code causing
630the problem.
398 631
399=item $cv->begin ([group callback]) 632=item $cv->begin ([group callback])
400 633
401=item $cv->end 634=item $cv->end
402
403These two methods are EXPERIMENTAL and MIGHT CHANGE.
404 635
405These two methods can be used to combine many transactions/events into 636These two methods can be used to combine many transactions/events into
406one. For example, a function that pings many hosts in parallel might want 637one. For example, a function that pings many hosts in parallel might want
407to use a condition variable for the whole process. 638to use a condition variable for the whole process.
408 639
410C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end 641C<< ->end >> will decrement it. If the counter reaches C<0> in C<< ->end
411>>, the (last) callback passed to C<begin> will be executed. That callback 642>>, the (last) callback passed to C<begin> will be executed. That callback
412is I<supposed> to call C<< ->send >>, but that is not required. If no 643is I<supposed> to call C<< ->send >>, but that is not required. If no
413callback was set, C<send> will be called without any arguments. 644callback was set, C<send> will be called without any arguments.
414 645
415Let's clarify this with the ping example: 646You can think of C<< $cv->send >> giving you an OR condition (one call
647sends), while C<< $cv->begin >> and C<< $cv->end >> giving you an AND
648condition (all C<begin> calls must be C<end>'ed before the condvar sends).
649
650Let's start with a simple example: you have two I/O watchers (for example,
651STDOUT and STDERR for a program), and you want to wait for both streams to
652close before activating a condvar:
653
654 my $cv = AnyEvent->condvar;
655
656 $cv->begin; # first watcher
657 my $w1 = AnyEvent->io (fh => $fh1, cb => sub {
658 defined sysread $fh1, my $buf, 4096
659 or $cv->end;
660 });
661
662 $cv->begin; # second watcher
663 my $w2 = AnyEvent->io (fh => $fh2, cb => sub {
664 defined sysread $fh2, my $buf, 4096
665 or $cv->end;
666 });
667
668 $cv->recv;
669
670This works because for every event source (EOF on file handle), there is
671one call to C<begin>, so the condvar waits for all calls to C<end> before
672sending.
673
674The ping example mentioned above is slightly more complicated, as the
675there are results to be passwd back, and the number of tasks that are
676begung can potentially be zero:
416 677
417 my $cv = AnyEvent->condvar; 678 my $cv = AnyEvent->condvar;
418 679
419 my %result; 680 my %result;
420 $cv->begin (sub { $cv->send (\%result) }); 681 $cv->begin (sub { $cv->send (\%result) });
440loop, which serves two important purposes: first, it sets the callback 701loop, which serves two important purposes: first, it sets the callback
441to be called once the counter reaches C<0>, and second, it ensures that 702to be called once the counter reaches C<0>, and second, it ensures that
442C<send> is called even when C<no> hosts are being pinged (the loop 703C<send> is called even when C<no> hosts are being pinged (the loop
443doesn't execute once). 704doesn't execute once).
444 705
445This is the general pattern when you "fan out" into multiple subrequests: 706This is the general pattern when you "fan out" into multiple (but
446use an outer C<begin>/C<end> pair to set the callback and ensure C<end> 707potentially none) subrequests: use an outer C<begin>/C<end> pair to set
447is called at least once, and then, for each subrequest you start, call 708the callback and ensure C<end> is called at least once, and then, for each
448C<begin> and for eahc subrequest you finish, call C<end>. 709subrequest you start, call C<begin> and for each subrequest you finish,
710call C<end>.
449 711
450=back 712=back
451 713
452=head3 METHODS FOR CONSUMERS 714=head3 METHODS FOR CONSUMERS
453 715
469function will call C<croak>. 731function will call C<croak>.
470 732
471In list context, all parameters passed to C<send> will be returned, 733In list context, all parameters passed to C<send> will be returned,
472in scalar context only the first one will be returned. 734in scalar context only the first one will be returned.
473 735
736Note that doing a blocking wait in a callback is not supported by any
737event loop, that is, recursive invocation of a blocking C<< ->recv
738>> is not allowed, and the C<recv> call will C<croak> if such a
739condition is detected. This condition can be slightly loosened by using
740L<Coro::AnyEvent>, which allows you to do a blocking C<< ->recv >> from
741any thread that doesn't run the event loop itself.
742
474Not all event models support a blocking wait - some die in that case 743Not all event models support a blocking wait - some die in that case
475(programs might want to do that to stay interactive), so I<if you are 744(programs might want to do that to stay interactive), so I<if you are
476using this from a module, never require a blocking wait>, but let the 745using this from a module, never require a blocking wait>. Instead, let the
477caller decide whether the call will block or not (for example, by coupling 746caller decide whether the call will block or not (for example, by coupling
478condition variables with some kind of request results and supporting 747condition variables with some kind of request results and supporting
479callbacks so the caller knows that getting the result will not block, 748callbacks so the caller knows that getting the result will not block,
480while still suppporting blocking waits if the caller so desires). 749while still supporting blocking waits if the caller so desires).
481
482Another reason I<never> to C<< ->recv >> in a module is that you cannot
483sensibly have two C<< ->recv >>'s in parallel, as that would require
484multiple interpreters or coroutines/threads, none of which C<AnyEvent>
485can supply.
486
487The L<Coro> module, however, I<can> and I<does> supply coroutines and, in
488fact, L<Coro::AnyEvent> replaces AnyEvent's condvars by coroutine-safe
489versions and also integrates coroutines into AnyEvent, making blocking
490C<< ->recv >> calls perfectly safe as long as they are done from another
491coroutine (one that doesn't run the event loop).
492 750
493You can ensure that C<< -recv >> never blocks by setting a callback and 751You can ensure that C<< -recv >> never blocks by setting a callback and
494only calling C<< ->recv >> from within that callback (or at a later 752only calling C<< ->recv >> from within that callback (or at a later
495time). This will work even when the event loop does not support blocking 753time). This will work even when the event loop does not support blocking
496waits otherwise. 754waits otherwise.
498=item $bool = $cv->ready 756=item $bool = $cv->ready
499 757
500Returns true when the condition is "true", i.e. whether C<send> or 758Returns true when the condition is "true", i.e. whether C<send> or
501C<croak> have been called. 759C<croak> have been called.
502 760
503=item $cb = $cv->cb ([new callback]) 761=item $cb = $cv->cb ($cb->($cv))
504 762
505This is a mutator function that returns the callback set and optionally 763This is a mutator function that returns the callback set and optionally
506replaces it before doing so. 764replaces it before doing so.
507 765
508The callback will be called when the condition becomes "true", i.e. when 766The callback will be called when the condition becomes "true", i.e. when
509C<send> or C<croak> are called. Calling C<recv> inside the callback 767C<send> or C<croak> are called, with the only argument being the condition
510or at any later time is guaranteed not to block. 768variable itself. Calling C<recv> inside the callback or at any later time
769is guaranteed not to block.
511 770
512=back 771=back
513 772
773=head1 SUPPORTED EVENT LOOPS/BACKENDS
774
775The available backend classes are (every class has its own manpage):
776
777=over 4
778
779=item Backends that are autoprobed when no other event loop can be found.
780
781EV is the preferred backend when no other event loop seems to be in
782use. If EV is not installed, then AnyEvent will try Event, and, failing
783that, will fall back to its own pure-perl implementation, which is
784available everywhere as it comes with AnyEvent itself.
785
786 AnyEvent::Impl::EV based on EV (interface to libev, best choice).
787 AnyEvent::Impl::Event based on Event, very stable, few glitches.
788 AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
789
790=item Backends that are transparently being picked up when they are used.
791
792These will be used when they are currently loaded when the first watcher
793is created, in which case it is assumed that the application is using
794them. This means that AnyEvent will automatically pick the right backend
795when the main program loads an event module before anything starts to
796create watchers. Nothing special needs to be done by the main program.
797
798 AnyEvent::Impl::Glib based on Glib, slow but very stable.
799 AnyEvent::Impl::Tk based on Tk, very broken.
800 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
801 AnyEvent::Impl::POE based on POE, very slow, some limitations.
802 AnyEvent::Impl::Irssi used when running within irssi.
803
804=item Backends with special needs.
805
806Qt requires the Qt::Application to be instantiated first, but will
807otherwise be picked up automatically. As long as the main program
808instantiates the application before any AnyEvent watchers are created,
809everything should just work.
810
811 AnyEvent::Impl::Qt based on Qt.
812
813Support for IO::Async can only be partial, as it is too broken and
814architecturally limited to even support the AnyEvent API. It also
815is the only event loop that needs the loop to be set explicitly, so
816it can only be used by a main program knowing about AnyEvent. See
817L<AnyEvent::Impl::Async> for the gory details.
818
819 AnyEvent::Impl::IOAsync based on IO::Async, cannot be autoprobed.
820
821=item Event loops that are indirectly supported via other backends.
822
823Some event loops can be supported via other modules:
824
825There is no direct support for WxWidgets (L<Wx>) or L<Prima>.
826
827B<WxWidgets> has no support for watching file handles. However, you can
828use WxWidgets through the POE adaptor, as POE has a Wx backend that simply
829polls 20 times per second, which was considered to be too horrible to even
830consider for AnyEvent.
831
832B<Prima> is not supported as nobody seems to be using it, but it has a POE
833backend, so it can be supported through POE.
834
835AnyEvent knows about both L<Prima> and L<Wx>, however, and will try to
836load L<POE> when detecting them, in the hope that POE will pick them up,
837in which case everything will be automatic.
838
839=back
840
514=head1 GLOBAL VARIABLES AND FUNCTIONS 841=head1 GLOBAL VARIABLES AND FUNCTIONS
515 842
843These are not normally required to use AnyEvent, but can be useful to
844write AnyEvent extension modules.
845
516=over 4 846=over 4
517 847
518=item $AnyEvent::MODEL 848=item $AnyEvent::MODEL
519 849
520Contains C<undef> until the first watcher is being created. Then it 850Contains C<undef> until the first watcher is being created, before the
851backend has been autodetected.
852
521contains the event model that is being used, which is the name of the 853Afterwards it contains the event model that is being used, which is the
522Perl class implementing the model. This class is usually one of the 854name of the Perl class implementing the model. This class is usually one
523C<AnyEvent::Impl:xxx> modules, but can be any other class in the case 855of the C<AnyEvent::Impl:xxx> modules, but can be any other class in the
524AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode>). 856case AnyEvent has been extended at runtime (e.g. in I<rxvt-unicode> it
525 857will be C<urxvt::anyevent>).
526The known classes so far are:
527
528 AnyEvent::Impl::EV based on EV (an interface to libev, best choice).
529 AnyEvent::Impl::Event based on Event, second best choice.
530 AnyEvent::Impl::Perl pure-perl implementation, fast and portable.
531 AnyEvent::Impl::Glib based on Glib, third-best choice.
532 AnyEvent::Impl::Tk based on Tk, very bad choice.
533 AnyEvent::Impl::Qt based on Qt, cannot be autoprobed (see its docs).
534 AnyEvent::Impl::EventLib based on Event::Lib, leaks memory and worse.
535 AnyEvent::Impl::POE based on POE, not generic enough for full support.
536
537There is no support for WxWidgets, as WxWidgets has no support for
538watching file handles. However, you can use WxWidgets through the
539POE Adaptor, as POE has a Wx backend that simply polls 20 times per
540second, which was considered to be too horrible to even consider for
541AnyEvent. Likewise, other POE backends can be used by AnyEvent by using
542it's adaptor.
543
544AnyEvent knows about L<Prima> and L<Wx> and will try to use L<POE> when
545autodetecting them.
546 858
547=item AnyEvent::detect 859=item AnyEvent::detect
548 860
549Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model 861Returns C<$AnyEvent::MODEL>, forcing autodetection of the event model
550if necessary. You should only call this function right before you would 862if necessary. You should only call this function right before you would
551have created an AnyEvent watcher anyway, that is, as late as possible at 863have created an AnyEvent watcher anyway, that is, as late as possible at
552runtime. 864runtime, and not e.g. while initialising of your module.
865
866If you need to do some initialisation before AnyEvent watchers are
867created, use C<post_detect>.
553 868
554=item $guard = AnyEvent::post_detect { BLOCK } 869=item $guard = AnyEvent::post_detect { BLOCK }
555 870
556Arranges for the code block to be executed as soon as the event model is 871Arranges for the code block to be executed as soon as the event model is
557autodetected (or immediately if this has already happened). 872autodetected (or immediately if this has already happened).
558 873
874The block will be executed I<after> the actual backend has been detected
875(C<$AnyEvent::MODEL> is set), but I<before> any watchers have been
876created, so it is possible to e.g. patch C<@AnyEvent::ISA> or do
877other initialisations - see the sources of L<AnyEvent::Strict> or
878L<AnyEvent::AIO> to see how this is used.
879
880The most common usage is to create some global watchers, without forcing
881event module detection too early, for example, L<AnyEvent::AIO> creates
882and installs the global L<IO::AIO> watcher in a C<post_detect> block to
883avoid autodetecting the event module at load time.
884
559If called in scalar or list context, then it creates and returns an object 885If called in scalar or list context, then it creates and returns an object
560that automatically removes the callback again when it is destroyed. See 886that automatically removes the callback again when it is destroyed (or
887C<undef> when the hook was immediately executed). See L<AnyEvent::AIO> for
561L<Coro::BDB> for a case where this is useful. 888a case where this is useful.
889
890Example: Create a watcher for the IO::AIO module and store it in
891C<$WATCHER>. Only do so after the event loop is initialised, though.
892
893 our WATCHER;
894
895 my $guard = AnyEvent::post_detect {
896 $WATCHER = AnyEvent->io (fh => IO::AIO::poll_fileno, poll => 'r', cb => \&IO::AIO::poll_cb);
897 };
898
899 # the ||= is important in case post_detect immediately runs the block,
900 # as to not clobber the newly-created watcher. assigning both watcher and
901 # post_detect guard to the same variable has the advantage of users being
902 # able to just C<undef $WATCHER> if the watcher causes them grief.
903
904 $WATCHER ||= $guard;
562 905
563=item @AnyEvent::post_detect 906=item @AnyEvent::post_detect
564 907
565If there are any code references in this array (you can C<push> to it 908If there are any code references in this array (you can C<push> to it
566before or after loading AnyEvent), then they will called directly after 909before or after loading AnyEvent), then they will called directly after
567the event loop has been chosen. 910the event loop has been chosen.
568 911
569You should check C<$AnyEvent::MODEL> before adding to this array, though: 912You should check C<$AnyEvent::MODEL> before adding to this array, though:
570if it contains a true value then the event loop has already been detected, 913if it is defined then the event loop has already been detected, and the
571and the array will be ignored. 914array will be ignored.
572 915
573Best use C<AnyEvent::post_detect { BLOCK }> instead. 916Best use C<AnyEvent::post_detect { BLOCK }> when your application allows
917it,as it takes care of these details.
918
919This variable is mainly useful for modules that can do something useful
920when AnyEvent is used and thus want to know when it is initialised, but do
921not need to even load it by default. This array provides the means to hook
922into AnyEvent passively, without loading it.
574 923
575=back 924=back
576 925
577=head1 WHAT TO DO IN A MODULE 926=head1 WHAT TO DO IN A MODULE
578 927
601 950
602If it doesn't care, it can just "use AnyEvent" and use it itself, or not 951If it doesn't care, it can just "use AnyEvent" and use it itself, or not
603do anything special (it does not need to be event-based) and let AnyEvent 952do anything special (it does not need to be event-based) and let AnyEvent
604decide which implementation to chose if some module relies on it. 953decide which implementation to chose if some module relies on it.
605 954
606If the main program relies on a specific event model. For example, in 955If the main program relies on a specific event model - for example, in
607Gtk2 programs you have to rely on the Glib module. You should load the 956Gtk2 programs you have to rely on the Glib module - you should load the
608event module before loading AnyEvent or any module that uses it: generally 957event module before loading AnyEvent or any module that uses it: generally
609speaking, you should load it as early as possible. The reason is that 958speaking, you should load it as early as possible. The reason is that
610modules might create watchers when they are loaded, and AnyEvent will 959modules might create watchers when they are loaded, and AnyEvent will
611decide on the event model to use as soon as it creates watchers, and it 960decide on the event model to use as soon as it creates watchers, and it
612might chose the wrong one unless you load the correct one yourself. 961might chose the wrong one unless you load the correct one yourself.
613 962
614You can chose to use a rather inefficient pure-perl implementation by 963You can chose to use a pure-perl implementation by loading the
615loading the C<AnyEvent::Impl::Perl> module, which gives you similar 964C<AnyEvent::Impl::Perl> module, which gives you similar behaviour
616behaviour everywhere, but letting AnyEvent chose is generally better. 965everywhere, but letting AnyEvent chose the model is generally better.
966
967=head2 MAINLOOP EMULATION
968
969Sometimes (often for short test scripts, or even standalone programs who
970only want to use AnyEvent), you do not want to run a specific event loop.
971
972In that case, you can use a condition variable like this:
973
974 AnyEvent->condvar->recv;
975
976This has the effect of entering the event loop and looping forever.
977
978Note that usually your program has some exit condition, in which case
979it is better to use the "traditional" approach of storing a condition
980variable somewhere, waiting for it, and sending it when the program should
981exit cleanly.
982
617 983
618=head1 OTHER MODULES 984=head1 OTHER MODULES
619 985
620The following is a non-exhaustive list of additional modules that use 986The following is a non-exhaustive list of additional modules that use
621AnyEvent and can therefore be mixed easily with other AnyEvent modules 987AnyEvent as a client and can therefore be mixed easily with other AnyEvent
622in the same program. Some of the modules come with AnyEvent, some are 988modules and other event loops in the same program. Some of the modules
623available via CPAN. 989come with AnyEvent, most are available via CPAN.
624 990
625=over 4 991=over 4
626 992
627=item L<AnyEvent::Util> 993=item L<AnyEvent::Util>
628 994
629Contains various utility functions that replace often-used but blocking 995Contains various utility functions that replace often-used but blocking
630functions such as C<inet_aton> by event-/callback-based versions. 996functions such as C<inet_aton> by event-/callback-based versions.
631 997
998=item L<AnyEvent::Socket>
999
1000Provides various utility functions for (internet protocol) sockets,
1001addresses and name resolution. Also functions to create non-blocking tcp
1002connections or tcp servers, with IPv6 and SRV record support and more.
1003
632=item L<AnyEvent::Handle> 1004=item L<AnyEvent::Handle>
633 1005
634Provide read and write buffers and manages watchers for reads and writes. 1006Provide read and write buffers, manages watchers for reads and writes,
1007supports raw and formatted I/O, I/O queued and fully transparent and
1008non-blocking SSL/TLS (via L<AnyEvent::TLS>.
1009
1010=item L<AnyEvent::DNS>
1011
1012Provides rich asynchronous DNS resolver capabilities.
1013
1014=item L<AnyEvent::HTTP>
1015
1016A simple-to-use HTTP library that is capable of making a lot of concurrent
1017HTTP requests.
635 1018
636=item L<AnyEvent::HTTPD> 1019=item L<AnyEvent::HTTPD>
637 1020
638Provides a simple web application server framework. 1021Provides a simple web application server framework.
639 1022
640=item L<AnyEvent::DNS>
641
642Provides asynchronous DNS resolver capabilities, beyond what
643L<AnyEvent::Util> offers.
644
645=item L<AnyEvent::FastPing> 1023=item L<AnyEvent::FastPing>
646 1024
647The fastest ping in the west. 1025The fastest ping in the west.
648 1026
1027=item L<AnyEvent::DBI>
1028
1029Executes L<DBI> requests asynchronously in a proxy process.
1030
1031=item L<AnyEvent::AIO>
1032
1033Truly asynchronous I/O, should be in the toolbox of every event
1034programmer. AnyEvent::AIO transparently fuses L<IO::AIO> and AnyEvent
1035together.
1036
1037=item L<AnyEvent::BDB>
1038
1039Truly asynchronous Berkeley DB access. AnyEvent::BDB transparently fuses
1040L<BDB> and AnyEvent together.
1041
1042=item L<AnyEvent::GPSD>
1043
1044A non-blocking interface to gpsd, a daemon delivering GPS information.
1045
649=item L<Net::IRC3> 1046=item L<AnyEvent::IRC>
650 1047
651AnyEvent based IRC client module family. 1048AnyEvent based IRC client module family (replacing the older Net::IRC3).
652 1049
653=item L<Net::XMPP2> 1050=item L<AnyEvent::XMPP>
654 1051
655AnyEvent based XMPP (Jabber protocol) module family. 1052AnyEvent based XMPP (Jabber protocol) module family (replacing the older
1053Net::XMPP2>.
1054
1055=item L<AnyEvent::IGS>
1056
1057A non-blocking interface to the Internet Go Server protocol (used by
1058L<App::IGS>).
656 1059
657=item L<Net::FCP> 1060=item L<Net::FCP>
658 1061
659AnyEvent-based implementation of the Freenet Client Protocol, birthplace 1062AnyEvent-based implementation of the Freenet Client Protocol, birthplace
660of AnyEvent. 1063of AnyEvent.
665 1068
666=item L<Coro> 1069=item L<Coro>
667 1070
668Has special support for AnyEvent via L<Coro::AnyEvent>. 1071Has special support for AnyEvent via L<Coro::AnyEvent>.
669 1072
670=item L<AnyEvent::AIO>, L<IO::AIO>
671
672Truly asynchronous I/O, should be in the toolbox of every event
673programmer. AnyEvent::AIO transparently fuses IO::AIO and AnyEvent
674together.
675
676=item L<AnyEvent::BDB>, L<BDB>
677
678Truly asynchronous Berkeley DB access. AnyEvent::AIO transparently fuses
679IO::AIO and AnyEvent together.
680
681=item L<IO::Lambda>
682
683The lambda approach to I/O - don't ask, look there. Can use AnyEvent.
684
685=back 1073=back
686 1074
687=cut 1075=cut
688 1076
689package AnyEvent; 1077package AnyEvent;
690 1078
1079# basically a tuned-down version of common::sense
1080sub common_sense {
691no warnings; 1081 # no warnings
692use strict; 1082 ${^WARNING_BITS} ^= ${^WARNING_BITS};
1083 # use strict vars subs
1084 $^H |= 0x00000600;
1085}
693 1086
1087BEGIN { AnyEvent::common_sense }
1088
694use Carp; 1089use Carp ();
695 1090
696our $VERSION = '3.41'; 1091our $VERSION = 4.86;
697our $MODEL; 1092our $MODEL;
698 1093
699our $AUTOLOAD; 1094our $AUTOLOAD;
700our @ISA; 1095our @ISA;
701 1096
702our $verbose = $ENV{PERL_ANYEVENT_VERBOSE}*1;
703
704our @REGISTRY; 1097our @REGISTRY;
705 1098
1099our $WIN32;
1100
1101our $VERBOSE;
1102
1103BEGIN {
1104 eval "sub WIN32(){ " . (($^O =~ /mswin32/i)*1) ." }";
1105 eval "sub TAINT(){ " . (${^TAINT}*1) . " }";
1106
1107 delete @ENV{grep /^PERL_ANYEVENT_/, keys %ENV}
1108 if ${^TAINT};
1109
1110 $VERBOSE = $ENV{PERL_ANYEVENT_VERBOSE}*1;
1111
1112}
1113
1114our $MAX_SIGNAL_LATENCY = 10;
1115
1116our %PROTOCOL; # (ipv4|ipv6) => (1|2), higher numbers are preferred
1117
1118{
1119 my $idx;
1120 $PROTOCOL{$_} = ++$idx
1121 for reverse split /\s*,\s*/,
1122 $ENV{PERL_ANYEVENT_PROTOCOLS} || "ipv4,ipv6";
1123}
1124
706my @models = ( 1125my @models = (
707 [EV:: => AnyEvent::Impl::EV::], 1126 [EV:: => AnyEvent::Impl::EV:: , 1],
708 [Event:: => AnyEvent::Impl::Event::], 1127 [Event:: => AnyEvent::Impl::Event::, 1],
1128 [AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl:: , 1],
1129 # everything below here will not (normally) be autoprobed
1130 # as the pureperl backend should work everywhere
1131 # and is usually faster
1132 [Glib:: => AnyEvent::Impl::Glib:: , 1], # becomes extremely slow with many watchers
1133 [Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy
1134 [Irssi:: => AnyEvent::Impl::Irssi::], # Irssi has a bogus "Event" package
709 [Tk:: => AnyEvent::Impl::Tk::], 1135 [Tk:: => AnyEvent::Impl::Tk::], # crashes with many handles
1136 [Qt:: => AnyEvent::Impl::Qt::], # requires special main program
1137 [POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza
710 [Wx:: => AnyEvent::Impl::POE::], 1138 [Wx:: => AnyEvent::Impl::POE::],
711 [Prima:: => AnyEvent::Impl::POE::], 1139 [Prima:: => AnyEvent::Impl::POE::],
712 [AnyEvent::Impl::Perl:: => AnyEvent::Impl::Perl::], 1140 # IO::Async is just too broken - we would need workarounds for its
713 # everything below here will not be autoprobed as the pureperl backend should work everywhere 1141 # byzantine signal and broken child handling, among others.
714 [Glib:: => AnyEvent::Impl::Glib::], 1142 # IO::Async is rather hard to detect, as it doesn't have any
715 [Event::Lib:: => AnyEvent::Impl::EventLib::], # too buggy 1143 # obvious default class.
716 [Qt:: => AnyEvent::Impl::Qt::], # requires special main program 1144# [0, IO::Async:: => AnyEvent::Impl::IOAsync::], # requires special main program
717 [POE::Kernel:: => AnyEvent::Impl::POE::], # lasciate ogni speranza 1145# [0, IO::Async::Loop:: => AnyEvent::Impl::IOAsync::], # requires special main program
1146# [0, IO::Async::Notifier:: => AnyEvent::Impl::IOAsync::], # requires special main program
718); 1147);
719 1148
720our %method = map +($_ => 1), qw(io timer signal child condvar one_event DESTROY); 1149our %method = map +($_ => 1),
1150 qw(io timer time now now_update signal child idle condvar one_event DESTROY);
721 1151
722our @post_detect; 1152our @post_detect;
723 1153
724sub post_detect(&) { 1154sub post_detect(&) {
725 my ($cb) = @_; 1155 my ($cb) = @_;
726 1156
727 if ($MODEL) { 1157 if ($MODEL) {
728 $cb->(); 1158 $cb->();
729 1159
730 1 1160 undef
731 } else { 1161 } else {
732 push @post_detect, $cb; 1162 push @post_detect, $cb;
733 1163
734 defined wantarray 1164 defined wantarray
735 ? bless \$cb, "AnyEvent::Util::Guard" 1165 ? bless \$cb, "AnyEvent::Util::postdetect"
736 : () 1166 : ()
737 } 1167 }
738} 1168}
739 1169
740sub AnyEvent::Util::Guard::DESTROY { 1170sub AnyEvent::Util::postdetect::DESTROY {
741 @post_detect = grep $_ != ${$_[0]}, @post_detect; 1171 @post_detect = grep $_ != ${$_[0]}, @post_detect;
742} 1172}
743 1173
744sub detect() { 1174sub detect() {
745 unless ($MODEL) { 1175 unless ($MODEL) {
746 no strict 'refs'; 1176 local $SIG{__DIE__};
747 1177
748 if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) { 1178 if ($ENV{PERL_ANYEVENT_MODEL} =~ /^([a-zA-Z]+)$/) {
749 my $model = "AnyEvent::Impl::$1"; 1179 my $model = "AnyEvent::Impl::$1";
750 if (eval "require $model") { 1180 if (eval "require $model") {
751 $MODEL = $model; 1181 $MODEL = $model;
752 warn "AnyEvent: loaded model '$model' (forced by \$PERL_ANYEVENT_MODEL), using it.\n" if $verbose > 1; 1182 warn "AnyEvent: loaded model '$model' (forced by \$ENV{PERL_ANYEVENT_MODEL}), using it.\n" if $VERBOSE >= 2;
753 } else { 1183 } else {
754 warn "AnyEvent: unable to load model '$model' (from \$PERL_ANYEVENT_MODEL):\n$@" if $verbose; 1184 warn "AnyEvent: unable to load model '$model' (from \$ENV{PERL_ANYEVENT_MODEL}):\n$@" if $VERBOSE;
755 } 1185 }
756 } 1186 }
757 1187
758 # check for already loaded models 1188 # check for already loaded models
759 unless ($MODEL) { 1189 unless ($MODEL) {
760 for (@REGISTRY, @models) { 1190 for (@REGISTRY, @models) {
761 my ($package, $model) = @$_; 1191 my ($package, $model) = @$_;
762 if (${"$package\::VERSION"} > 0) { 1192 if (${"$package\::VERSION"} > 0) {
763 if (eval "require $model") { 1193 if (eval "require $model") {
764 $MODEL = $model; 1194 $MODEL = $model;
765 warn "AnyEvent: autodetected model '$model', using it.\n" if $verbose > 1; 1195 warn "AnyEvent: autodetected model '$model', using it.\n" if $VERBOSE >= 2;
766 last; 1196 last;
767 } 1197 }
768 } 1198 }
769 } 1199 }
770 1200
771 unless ($MODEL) { 1201 unless ($MODEL) {
772 # try to load a model 1202 # try to autoload a model
773
774 for (@REGISTRY, @models) { 1203 for (@REGISTRY, @models) {
775 my ($package, $model) = @$_; 1204 my ($package, $model, $autoload) = @$_;
1205 if (
1206 $autoload
776 if (eval "require $package" 1207 and eval "require $package"
777 and ${"$package\::VERSION"} > 0 1208 and ${"$package\::VERSION"} > 0
778 and eval "require $model") { 1209 and eval "require $model"
1210 ) {
779 $MODEL = $model; 1211 $MODEL = $model;
780 warn "AnyEvent: autoprobed model '$model', using it.\n" if $verbose > 1; 1212 warn "AnyEvent: autoloaded model '$model', using it.\n" if $VERBOSE >= 2;
781 last; 1213 last;
782 } 1214 }
783 } 1215 }
784 1216
785 $MODEL 1217 $MODEL
786 or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib."; 1218 or die "No event module selected for AnyEvent and autodetect failed. Install any one of these modules: EV, Event or Glib.\n";
787 } 1219 }
788 } 1220 }
789 1221
1222 push @{"$MODEL\::ISA"}, "AnyEvent::Base";
1223
790 unshift @ISA, $MODEL; 1224 unshift @ISA, $MODEL;
791 push @{"$MODEL\::ISA"}, "AnyEvent::Base"; 1225
1226 require AnyEvent::Strict if $ENV{PERL_ANYEVENT_STRICT};
792 1227
793 (shift @post_detect)->() while @post_detect; 1228 (shift @post_detect)->() while @post_detect;
794 } 1229 }
795 1230
796 $MODEL 1231 $MODEL
798 1233
799sub AUTOLOAD { 1234sub AUTOLOAD {
800 (my $func = $AUTOLOAD) =~ s/.*://; 1235 (my $func = $AUTOLOAD) =~ s/.*://;
801 1236
802 $method{$func} 1237 $method{$func}
803 or croak "$func: not a valid method for AnyEvent objects"; 1238 or Carp::croak "$func: not a valid method for AnyEvent objects";
804 1239
805 detect unless $MODEL; 1240 detect unless $MODEL;
806 1241
807 my $class = shift; 1242 my $class = shift;
808 $class->$func (@_); 1243 $class->$func (@_);
809} 1244}
810 1245
1246# utility function to dup a filehandle. this is used by many backends
1247# to support binding more than one watcher per filehandle (they usually
1248# allow only one watcher per fd, so we dup it to get a different one).
1249sub _dupfh($$;$$) {
1250 my ($poll, $fh, $r, $w) = @_;
1251
1252 # cygwin requires the fh mode to be matching, unix doesn't
1253 my ($rw, $mode) = $poll eq "r" ? ($r, "<&") : ($w, ">&");
1254
1255 open my $fh2, $mode, $fh
1256 or die "AnyEvent->io: cannot dup() filehandle in mode '$poll': $!,";
1257
1258 # we assume CLOEXEC is already set by perl in all important cases
1259
1260 ($fh2, $rw)
1261}
1262
811package AnyEvent::Base; 1263package AnyEvent::Base;
812 1264
1265# default implementations for many methods
1266
1267sub _time {
1268 # probe for availability of Time::HiRes
1269 if (eval "use Time::HiRes (); Time::HiRes::time (); 1") {
1270 warn "AnyEvent: using Time::HiRes for sub-second timing accuracy.\n" if $VERBOSE >= 8;
1271 *_time = \&Time::HiRes::time;
1272 # if (eval "use POSIX (); (POSIX::times())...
1273 } else {
1274 warn "AnyEvent: using built-in time(), WARNING, no sub-second resolution!\n" if $VERBOSE;
1275 *_time = sub { time }; # epic fail
1276 }
1277
1278 &_time
1279}
1280
1281sub time { _time }
1282sub now { _time }
1283sub now_update { }
1284
813# default implementation for ->condvar 1285# default implementation for ->condvar
814 1286
815sub condvar { 1287sub condvar {
816 bless {}, AnyEvent::CondVar:: 1288 bless { @_ == 3 ? (_ae_cb => $_[2]) : () }, "AnyEvent::CondVar"
817} 1289}
818 1290
819# default implementation for ->signal 1291# default implementation for ->signal
820 1292
821our %SIG_CB; 1293our $HAVE_ASYNC_INTERRUPT;
1294our ($SIGPIPE_R, $SIGPIPE_W, %SIG_CB, %SIG_EV, $SIG_IO);
1295our (%SIG_ASY, %SIG_ASY_W);
1296our ($SIG_COUNT, $SIG_TW);
822 1297
1298sub _signal_exec {
1299 $HAVE_ASYNC_INTERRUPT
1300 ? $SIGPIPE_R->drain
1301 : sysread $SIGPIPE_R, my $dummy, 9;
1302
1303 while (%SIG_EV) {
1304 for (keys %SIG_EV) {
1305 delete $SIG_EV{$_};
1306 $_->() for values %{ $SIG_CB{$_} || {} };
1307 }
1308 }
1309}
1310
1311# install a dumym wakeupw atcher to reduce signal catching latency
1312sub _sig_add() {
1313 unless ($SIG_COUNT++) {
1314 # try to align timer on a full-second boundary, if possible
1315 my $NOW = AnyEvent->now;
1316
1317 $SIG_TW = AnyEvent->timer (
1318 after => $MAX_SIGNAL_LATENCY - ($NOW - int $NOW),
1319 interval => $MAX_SIGNAL_LATENCY,
1320 cb => sub { }, # just for the PERL_ASYNC_CHECK
1321 );
1322 }
1323}
1324
1325sub _sig_del {
1326 undef $SIG_TW
1327 unless --$SIG_COUNT;
1328}
1329
823sub signal { 1330sub _signal {
824 my (undef, %arg) = @_; 1331 my (undef, %arg) = @_;
825 1332
826 my $signal = uc $arg{signal} 1333 my $signal = uc $arg{signal}
827 or Carp::croak "required option 'signal' is missing"; 1334 or Carp::croak "required option 'signal' is missing";
828 1335
829 $SIG_CB{$signal}{$arg{cb}} = $arg{cb}; 1336 $SIG_CB{$signal}{$arg{cb}} = $arg{cb};
1337
1338 if ($HAVE_ASYNC_INTERRUPT) {
1339 # async::interrupt
1340
1341 $SIG_ASY{$signal} ||= do {
1342 my $asy = new Async::Interrupt
1343 cb => sub { undef $SIG_EV{$signal} },
1344 signal => $signal,
1345 pipe => [$SIGPIPE_R->filenos],
1346 ;
1347 $asy->pipe_autodrain (0);
1348
1349 $asy
1350 };
1351
1352 } else {
1353 # pure perl
1354
830 $SIG{$signal} ||= sub { 1355 $SIG{$signal} ||= sub {
831 $_->() for values %{ $SIG_CB{$signal} || {} }; 1356 local $!;
1357 syswrite $SIGPIPE_W, "\x00", 1 unless %SIG_EV;
1358 undef $SIG_EV{$signal};
1359 };
1360
1361 # can't do signal processing without introducing races in pure perl,
1362 # so limit the signal latency.
1363 _sig_add;
832 }; 1364 }
833 1365
834 bless [$signal, $arg{cb}], "AnyEvent::Base::Signal" 1366 bless [$signal, $arg{cb}], "AnyEvent::Base::signal"
835} 1367}
836 1368
1369sub signal {
1370 # probe for availability of Async::Interrupt
1371 if (!$ENV{PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT} && eval "use Async::Interrupt 0.6 (); 1") {
1372 warn "AnyEvent: using Async::Interrupt for race-free signal handling.\n" if $VERBOSE >= 8;
1373
1374 $HAVE_ASYNC_INTERRUPT = 1;
1375 $SIGPIPE_R = new Async::Interrupt::EventPipe;
1376 $SIG_IO = AnyEvent->io (fh => $SIGPIPE_R->fileno, poll => "r", cb => \&_signal_exec);
1377
1378 } else {
1379 warn "AnyEvent: using emulated perl signal handling with latency timer.\n" if $VERBOSE >= 8;
1380
1381 require Fcntl;
1382
1383 if (AnyEvent::WIN32) {
1384 require AnyEvent::Util;
1385
1386 ($SIGPIPE_R, $SIGPIPE_W) = AnyEvent::Util::portable_pipe ();
1387 AnyEvent::Util::fh_nonblocking ($SIGPIPE_R) if $SIGPIPE_R;
1388 AnyEvent::Util::fh_nonblocking ($SIGPIPE_W) if $SIGPIPE_W; # just in case
1389 } else {
1390 pipe $SIGPIPE_R, $SIGPIPE_W;
1391 fcntl $SIGPIPE_R, &Fcntl::F_SETFL, &Fcntl::O_NONBLOCK if $SIGPIPE_R;
1392 fcntl $SIGPIPE_W, &Fcntl::F_SETFL, &Fcntl::O_NONBLOCK if $SIGPIPE_W; # just in case
1393
1394 # not strictly required, as $^F is normally 2, but let's make sure...
1395 fcntl $SIGPIPE_R, &Fcntl::F_SETFD, &Fcntl::FD_CLOEXEC;
1396 fcntl $SIGPIPE_W, &Fcntl::F_SETFD, &Fcntl::FD_CLOEXEC;
1397 }
1398
1399 $SIGPIPE_R
1400 or Carp::croak "AnyEvent: unable to create a signal reporting pipe: $!\n";
1401
1402 $SIG_IO = AnyEvent->io (fh => $SIGPIPE_R, poll => "r", cb => \&_signal_exec);
1403 }
1404
1405 *signal = \&_signal;
1406 &signal
1407}
1408
837sub AnyEvent::Base::Signal::DESTROY { 1409sub AnyEvent::Base::signal::DESTROY {
838 my ($signal, $cb) = @{$_[0]}; 1410 my ($signal, $cb) = @{$_[0]};
839 1411
1412 _sig_del;
1413
840 delete $SIG_CB{$signal}{$cb}; 1414 delete $SIG_CB{$signal}{$cb};
841 1415
842 $SIG{$signal} = 'DEFAULT' unless keys %{ $SIG_CB{$signal} }; 1416 $HAVE_ASYNC_INTERRUPT
1417 ? delete $SIG_ASY{$signal}
1418 : # delete doesn't work with older perls - they then
1419 # print weird messages, or just unconditionally exit
1420 # instead of getting the default action.
1421 undef $SIG{$signal}
1422 unless keys %{ $SIG_CB{$signal} };
843} 1423}
844 1424
845# default implementation for ->child 1425# default implementation for ->child
846 1426
847our %PID_CB; 1427our %PID_CB;
848our $CHLD_W; 1428our $CHLD_W;
849our $CHLD_DELAY_W; 1429our $CHLD_DELAY_W;
850our $PID_IDLE;
851our $WNOHANG; 1430our $WNOHANG;
852 1431
853sub _child_wait { 1432sub _emit_childstatus($$) {
854 while (0 < (my $pid = waitpid -1, $WNOHANG)) { 1433 my (undef, $rpid, $rstatus) = @_;
1434
1435 $_->($rpid, $rstatus)
855 $_->($pid, $?) for (values %{ $PID_CB{$pid} || {} }), 1436 for values %{ $PID_CB{$rpid} || {} },
856 (values %{ $PID_CB{0} || {} }); 1437 values %{ $PID_CB{0} || {} };
857 }
858
859 undef $PID_IDLE;
860} 1438}
861 1439
862sub _sigchld { 1440sub _sigchld {
863 # make sure we deliver these changes "synchronous" with the event loop. 1441 my $pid;
864 $CHLD_DELAY_W ||= AnyEvent->timer (after => 0, cb => sub { 1442
865 undef $CHLD_DELAY_W; 1443 AnyEvent->_emit_childstatus ($pid, $?)
866 &_child_wait; 1444 while ($pid = waitpid -1, $WNOHANG) > 0;
867 });
868} 1445}
869 1446
870sub child { 1447sub child {
871 my (undef, %arg) = @_; 1448 my (undef, %arg) = @_;
872 1449
873 defined (my $pid = $arg{pid} + 0) 1450 defined (my $pid = $arg{pid} + 0)
874 or Carp::croak "required option 'pid' is missing"; 1451 or Carp::croak "required option 'pid' is missing";
875 1452
876 $PID_CB{$pid}{$arg{cb}} = $arg{cb}; 1453 $PID_CB{$pid}{$arg{cb}} = $arg{cb};
877 1454
878 unless ($WNOHANG) { 1455 # WNOHANG is almost cetrainly 1 everywhere
879 $WNOHANG = eval { require POSIX; &POSIX::WNOHANG } || 1; 1456 $WNOHANG ||= $^O =~ /^(?:openbsd|netbsd|linux|freebsd|cygwin|MSWin32)$/
880 } 1457 ? 1
1458 : eval { local $SIG{__DIE__}; require POSIX; &POSIX::WNOHANG } || 1;
881 1459
882 unless ($CHLD_W) { 1460 unless ($CHLD_W) {
883 $CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld); 1461 $CHLD_W = AnyEvent->signal (signal => 'CHLD', cb => \&_sigchld);
884 # child could be a zombie already, so make at least one round 1462 # child could be a zombie already, so make at least one round
885 &_sigchld; 1463 &_sigchld;
886 } 1464 }
887 1465
888 bless [$pid, $arg{cb}], "AnyEvent::Base::Child" 1466 bless [$pid, $arg{cb}], "AnyEvent::Base::child"
889} 1467}
890 1468
891sub AnyEvent::Base::Child::DESTROY { 1469sub AnyEvent::Base::child::DESTROY {
892 my ($pid, $cb) = @{$_[0]}; 1470 my ($pid, $cb) = @{$_[0]};
893 1471
894 delete $PID_CB{$pid}{$cb}; 1472 delete $PID_CB{$pid}{$cb};
895 delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} }; 1473 delete $PID_CB{$pid} unless keys %{ $PID_CB{$pid} };
896 1474
897 undef $CHLD_W unless keys %PID_CB; 1475 undef $CHLD_W unless keys %PID_CB;
898} 1476}
899 1477
1478# idle emulation is done by simply using a timer, regardless
1479# of whether the process is idle or not, and not letting
1480# the callback use more than 50% of the time.
1481sub idle {
1482 my (undef, %arg) = @_;
1483
1484 my ($cb, $w, $rcb) = $arg{cb};
1485
1486 $rcb = sub {
1487 if ($cb) {
1488 $w = _time;
1489 &$cb;
1490 $w = _time - $w;
1491
1492 # never use more then 50% of the time for the idle watcher,
1493 # within some limits
1494 $w = 0.0001 if $w < 0.0001;
1495 $w = 5 if $w > 5;
1496
1497 $w = AnyEvent->timer (after => $w, cb => $rcb);
1498 } else {
1499 # clean up...
1500 undef $w;
1501 undef $rcb;
1502 }
1503 };
1504
1505 $w = AnyEvent->timer (after => 0.05, cb => $rcb);
1506
1507 bless \\$cb, "AnyEvent::Base::idle"
1508}
1509
1510sub AnyEvent::Base::idle::DESTROY {
1511 undef $${$_[0]};
1512}
1513
900package AnyEvent::CondVar; 1514package AnyEvent::CondVar;
901 1515
902our @ISA = AnyEvent::CondVar::Base::; 1516our @ISA = AnyEvent::CondVar::Base::;
903 1517
904package AnyEvent::CondVar::Base; 1518package AnyEvent::CondVar::Base;
1519
1520#use overload
1521# '&{}' => sub { my $self = shift; sub { $self->send (@_) } },
1522# fallback => 1;
1523
1524# save 300+ kilobytes by dirtily hardcoding overloading
1525${"AnyEvent::CondVar::Base::OVERLOAD"}{dummy}++; # Register with magic by touching.
1526*{'AnyEvent::CondVar::Base::()'} = sub { }; # "Make it findable via fetchmethod."
1527*{'AnyEvent::CondVar::Base::(&{}'} = sub { my $self = shift; sub { $self->send (@_) } }; # &{}
1528${'AnyEvent::CondVar::Base::()'} = 1; # fallback
1529
1530our $WAITING;
905 1531
906sub _send { 1532sub _send {
907 # nop 1533 # nop
908} 1534}
909 1535
922sub ready { 1548sub ready {
923 $_[0]{_ae_sent} 1549 $_[0]{_ae_sent}
924} 1550}
925 1551
926sub _wait { 1552sub _wait {
1553 $WAITING
1554 and !$_[0]{_ae_sent}
1555 and Carp::croak "AnyEvent::CondVar: recursive blocking wait detected";
1556
1557 local $WAITING = 1;
927 AnyEvent->one_event while !$_[0]{_ae_sent}; 1558 AnyEvent->one_event while !$_[0]{_ae_sent};
928} 1559}
929 1560
930sub recv { 1561sub recv {
931 $_[0]->_wait; 1562 $_[0]->_wait;
944 $_[0]{_ae_end_cb} = $_[1] if @_ > 1; 1575 $_[0]{_ae_end_cb} = $_[1] if @_ > 1;
945} 1576}
946 1577
947sub end { 1578sub end {
948 return if --$_[0]{_ae_counter}; 1579 return if --$_[0]{_ae_counter};
949 &{ $_[0]{_ae_end_cb} } if $_[0]{_ae_end_cb}; 1580 &{ $_[0]{_ae_end_cb} || sub { $_[0]->send } };
950} 1581}
951 1582
952# undocumented/compatibility with pre-3.4 1583# undocumented/compatibility with pre-3.4
953*broadcast = \&send; 1584*broadcast = \&send;
954*wait = \&_wait; 1585*wait = \&_wait;
1586
1587=head1 ERROR AND EXCEPTION HANDLING
1588
1589In general, AnyEvent does not do any error handling - it relies on the
1590caller to do that if required. The L<AnyEvent::Strict> module (see also
1591the C<PERL_ANYEVENT_STRICT> environment variable, below) provides strict
1592checking of all AnyEvent methods, however, which is highly useful during
1593development.
1594
1595As for exception handling (i.e. runtime errors and exceptions thrown while
1596executing a callback), this is not only highly event-loop specific, but
1597also not in any way wrapped by this module, as this is the job of the main
1598program.
1599
1600The pure perl event loop simply re-throws the exception (usually
1601within C<< condvar->recv >>), the L<Event> and L<EV> modules call C<<
1602$Event/EV::DIED->() >>, L<Glib> uses C<< install_exception_handler >> and
1603so on.
1604
1605=head1 ENVIRONMENT VARIABLES
1606
1607The following environment variables are used by this module or its
1608submodules.
1609
1610Note that AnyEvent will remove I<all> environment variables starting with
1611C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
1612enabled.
1613
1614=over 4
1615
1616=item C<PERL_ANYEVENT_VERBOSE>
1617
1618By default, AnyEvent will be completely silent except in fatal
1619conditions. You can set this environment variable to make AnyEvent more
1620talkative.
1621
1622When set to C<1> or higher, causes AnyEvent to warn about unexpected
1623conditions, such as not being able to load the event model specified by
1624C<PERL_ANYEVENT_MODEL>.
1625
1626When set to C<2> or higher, cause AnyEvent to report to STDERR which event
1627model it chooses.
1628
1629When set to C<8> or higher, then AnyEvent will report extra information on
1630which optional modules it loads and how it implements certain features.
1631
1632=item C<PERL_ANYEVENT_STRICT>
1633
1634AnyEvent does not do much argument checking by default, as thorough
1635argument checking is very costly. Setting this variable to a true value
1636will cause AnyEvent to load C<AnyEvent::Strict> and then to thoroughly
1637check the arguments passed to most method calls. If it finds any problems,
1638it will croak.
1639
1640In other words, enables "strict" mode.
1641
1642Unlike C<use strict> (or it's modern cousin, C<< use L<common::sense>
1643>>, it is definitely recommended to keep it off in production. Keeping
1644C<PERL_ANYEVENT_STRICT=1> in your environment while developing programs
1645can be very useful, however.
1646
1647=item C<PERL_ANYEVENT_MODEL>
1648
1649This can be used to specify the event model to be used by AnyEvent, before
1650auto detection and -probing kicks in. It must be a string consisting
1651entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
1652and the resulting module name is loaded and if the load was successful,
1653used as event model. If it fails to load AnyEvent will proceed with
1654auto detection and -probing.
1655
1656This functionality might change in future versions.
1657
1658For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
1659could start your program like this:
1660
1661 PERL_ANYEVENT_MODEL=Perl perl ...
1662
1663=item C<PERL_ANYEVENT_PROTOCOLS>
1664
1665Used by both L<AnyEvent::DNS> and L<AnyEvent::Socket> to determine preferences
1666for IPv4 or IPv6. The default is unspecified (and might change, or be the result
1667of auto probing).
1668
1669Must be set to a comma-separated list of protocols or address families,
1670current supported: C<ipv4> and C<ipv6>. Only protocols mentioned will be
1671used, and preference will be given to protocols mentioned earlier in the
1672list.
1673
1674This variable can effectively be used for denial-of-service attacks
1675against local programs (e.g. when setuid), although the impact is likely
1676small, as the program has to handle conenction and other failures anyways.
1677
1678Examples: C<PERL_ANYEVENT_PROTOCOLS=ipv4,ipv6> - prefer IPv4 over IPv6,
1679but support both and try to use both. C<PERL_ANYEVENT_PROTOCOLS=ipv4>
1680- only support IPv4, never try to resolve or contact IPv6
1681addresses. C<PERL_ANYEVENT_PROTOCOLS=ipv6,ipv4> support either IPv4 or
1682IPv6, but prefer IPv6 over IPv4.
1683
1684=item C<PERL_ANYEVENT_EDNS0>
1685
1686Used by L<AnyEvent::DNS> to decide whether to use the EDNS0 extension
1687for DNS. This extension is generally useful to reduce DNS traffic, but
1688some (broken) firewalls drop such DNS packets, which is why it is off by
1689default.
1690
1691Setting this variable to C<1> will cause L<AnyEvent::DNS> to announce
1692EDNS0 in its DNS requests.
1693
1694=item C<PERL_ANYEVENT_MAX_FORKS>
1695
1696The maximum number of child processes that C<AnyEvent::Util::fork_call>
1697will create in parallel.
1698
1699=item C<PERL_ANYEVENT_MAX_OUTSTANDING_DNS>
1700
1701The default value for the C<max_outstanding> parameter for the default DNS
1702resolver - this is the maximum number of parallel DNS requests that are
1703sent to the DNS server.
1704
1705=item C<PERL_ANYEVENT_RESOLV_CONF>
1706
1707The file to use instead of F</etc/resolv.conf> (or OS-specific
1708configuration) in the default resolver. When set to the empty string, no
1709default config will be used.
1710
1711=item C<PERL_ANYEVENT_CA_FILE>, C<PERL_ANYEVENT_CA_PATH>.
1712
1713When neither C<ca_file> nor C<ca_path> was specified during
1714L<AnyEvent::TLS> context creation, and either of these environment
1715variables exist, they will be used to specify CA certificate locations
1716instead of a system-dependent default.
1717
1718=item C<PERL_ANYEVENT_AVOID_GUARD> and C<PERL_ANYEVENT_AVOID_ASYNC_INTERRUPT>
1719
1720When these are set to C<1>, then the respective modules are not
1721loaded. Mostly good for testing AnyEvent itself.
1722
1723=back
955 1724
956=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE 1725=head1 SUPPLYING YOUR OWN EVENT MODEL INTERFACE
957 1726
958This is an advanced topic that you do not normally need to use AnyEvent in 1727This is an advanced topic that you do not normally need to use AnyEvent in
959a module. This section is only of use to event loop authors who want to 1728a module. This section is only of use to event loop authors who want to
993 1762
994I<rxvt-unicode> also cheats a bit by not providing blocking access to 1763I<rxvt-unicode> also cheats a bit by not providing blocking access to
995condition variables: code blocking while waiting for a condition will 1764condition variables: code blocking while waiting for a condition will
996C<die>. This still works with most modules/usages, and blocking calls must 1765C<die>. This still works with most modules/usages, and blocking calls must
997not be done in an interactive application, so it makes sense. 1766not be done in an interactive application, so it makes sense.
998
999=head1 ENVIRONMENT VARIABLES
1000
1001The following environment variables are used by this module:
1002
1003=over 4
1004
1005=item C<PERL_ANYEVENT_VERBOSE>
1006
1007By default, AnyEvent will be completely silent except in fatal
1008conditions. You can set this environment variable to make AnyEvent more
1009talkative.
1010
1011When set to C<1> or higher, causes AnyEvent to warn about unexpected
1012conditions, such as not being able to load the event model specified by
1013C<PERL_ANYEVENT_MODEL>.
1014
1015When set to C<2> or higher, cause AnyEvent to report to STDERR which event
1016model it chooses.
1017
1018=item C<PERL_ANYEVENT_MODEL>
1019
1020This can be used to specify the event model to be used by AnyEvent, before
1021autodetection and -probing kicks in. It must be a string consisting
1022entirely of ASCII letters. The string C<AnyEvent::Impl::> gets prepended
1023and the resulting module name is loaded and if the load was successful,
1024used as event model. If it fails to load AnyEvent will proceed with
1025autodetection and -probing.
1026
1027This functionality might change in future versions.
1028
1029For example, to force the pure perl model (L<AnyEvent::Impl::Perl>) you
1030could start your program like this:
1031
1032 PERL_ANYEVENT_MODEL=Perl perl ...
1033
1034=back
1035 1767
1036=head1 EXAMPLE PROGRAM 1768=head1 EXAMPLE PROGRAM
1037 1769
1038The following program uses an I/O watcher to read data from STDIN, a timer 1770The following program uses an I/O watcher to read data from STDIN, a timer
1039to display a message once per second, and a condition variable to quit the 1771to display a message once per second, and a condition variable to quit the
1048 poll => 'r', 1780 poll => 'r',
1049 cb => sub { 1781 cb => sub {
1050 warn "io event <$_[0]>\n"; # will always output <r> 1782 warn "io event <$_[0]>\n"; # will always output <r>
1051 chomp (my $input = <STDIN>); # read a line 1783 chomp (my $input = <STDIN>); # read a line
1052 warn "read: $input\n"; # output what has been read 1784 warn "read: $input\n"; # output what has been read
1053 $cv->broadcast if $input =~ /^q/i; # quit program if /^q/i 1785 $cv->send if $input =~ /^q/i; # quit program if /^q/i
1054 }, 1786 },
1055 ); 1787 );
1056 1788
1057 my $time_watcher; # can only be used once 1789 my $time_watcher; # can only be used once
1058 1790
1063 }); 1795 });
1064 } 1796 }
1065 1797
1066 new_timer; # create first timer 1798 new_timer; # create first timer
1067 1799
1068 $cv->wait; # wait until user enters /^q/i 1800 $cv->recv; # wait until user enters /^q/i
1069 1801
1070=head1 REAL-WORLD EXAMPLE 1802=head1 REAL-WORLD EXAMPLE
1071 1803
1072Consider the L<Net::FCP> module. It features (among others) the following 1804Consider the L<Net::FCP> module. It features (among others) the following
1073API calls, which are to freenet what HTTP GET requests are to http: 1805API calls, which are to freenet what HTTP GET requests are to http:
1123 syswrite $txn->{fh}, $txn->{request} 1855 syswrite $txn->{fh}, $txn->{request}
1124 or die "connection or write error"; 1856 or die "connection or write error";
1125 $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r }); 1857 $txn->{w} = AnyEvent->io (fh => $txn->{fh}, poll => 'r', cb => sub { $txn->fh_ready_r });
1126 1858
1127Again, C<fh_ready_r> waits till all data has arrived, and then stores the 1859Again, C<fh_ready_r> waits till all data has arrived, and then stores the
1128result and signals any possible waiters that the request ahs finished: 1860result and signals any possible waiters that the request has finished:
1129 1861
1130 sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf}; 1862 sysread $txn->{fh}, $txn->{buf}, length $txn->{$buf};
1131 1863
1132 if (end-of-file or data complete) { 1864 if (end-of-file or data complete) {
1133 $txn->{result} = $txn->{buf}; 1865 $txn->{result} = $txn->{buf};
1134 $txn->{finished}->broadcast; 1866 $txn->{finished}->send;
1135 $txb->{cb}->($txn) of $txn->{cb}; # also call callback 1867 $txb->{cb}->($txn) of $txn->{cb}; # also call callback
1136 } 1868 }
1137 1869
1138The C<result> method, finally, just waits for the finished signal (if the 1870The C<result> method, finally, just waits for the finished signal (if the
1139request was already finished, it doesn't wait, of course, and returns the 1871request was already finished, it doesn't wait, of course, and returns the
1140data: 1872data:
1141 1873
1142 $txn->{finished}->wait; 1874 $txn->{finished}->recv;
1143 return $txn->{result}; 1875 return $txn->{result};
1144 1876
1145The actual code goes further and collects all errors (C<die>s, exceptions) 1877The actual code goes further and collects all errors (C<die>s, exceptions)
1146that occured during request processing. The C<result> method detects 1878that occurred during request processing. The C<result> method detects
1147whether an exception as thrown (it is stored inside the $txn object) 1879whether an exception as thrown (it is stored inside the $txn object)
1148and just throws the exception, which means connection errors and other 1880and just throws the exception, which means connection errors and other
1149problems get reported tot he code that tries to use the result, not in a 1881problems get reported tot he code that tries to use the result, not in a
1150random callback. 1882random callback.
1151 1883
1182 1914
1183 my $quit = AnyEvent->condvar; 1915 my $quit = AnyEvent->condvar;
1184 1916
1185 $fcp->txn_client_get ($url)->cb (sub { 1917 $fcp->txn_client_get ($url)->cb (sub {
1186 ... 1918 ...
1187 $quit->broadcast; 1919 $quit->send;
1188 }); 1920 });
1189 1921
1190 $quit->wait; 1922 $quit->recv;
1191 1923
1192 1924
1193=head1 BENCHMARKS 1925=head1 BENCHMARKS
1194 1926
1195To give you an idea of the performance and overheads that AnyEvent adds 1927To give you an idea of the performance and overheads that AnyEvent adds
1197of various event loops I prepared some benchmarks. 1929of various event loops I prepared some benchmarks.
1198 1930
1199=head2 BENCHMARKING ANYEVENT OVERHEAD 1931=head2 BENCHMARKING ANYEVENT OVERHEAD
1200 1932
1201Here is a benchmark of various supported event models used natively and 1933Here is a benchmark of various supported event models used natively and
1202through anyevent. The benchmark creates a lot of timers (with a zero 1934through AnyEvent. The benchmark creates a lot of timers (with a zero
1203timeout) and I/O watchers (watching STDOUT, a pty, to become writable, 1935timeout) and I/O watchers (watching STDOUT, a pty, to become writable,
1204which it is), lets them fire exactly once and destroys them again. 1936which it is), lets them fire exactly once and destroys them again.
1205 1937
1206Source code for this benchmark is found as F<eg/bench> in the AnyEvent 1938Source code for this benchmark is found as F<eg/bench> in the AnyEvent
1207distribution. 1939distribution.
1224all watchers, to avoid adding memory overhead. That means closure creation 1956all watchers, to avoid adding memory overhead. That means closure creation
1225and memory usage is not included in the figures. 1957and memory usage is not included in the figures.
1226 1958
1227I<invoke> is the time, in microseconds, used to invoke a simple 1959I<invoke> is the time, in microseconds, used to invoke a simple
1228callback. The callback simply counts down a Perl variable and after it was 1960callback. The callback simply counts down a Perl variable and after it was
1229invoked "watcher" times, it would C<< ->broadcast >> a condvar once to 1961invoked "watcher" times, it would C<< ->send >> a condvar once to
1230signal the end of this phase. 1962signal the end of this phase.
1231 1963
1232I<destroy> is the time, in microseconds, that it takes to destroy a single 1964I<destroy> is the time, in microseconds, that it takes to destroy a single
1233watcher. 1965watcher.
1234 1966
1235=head3 Results 1967=head3 Results
1236 1968
1237 name watchers bytes create invoke destroy comment 1969 name watchers bytes create invoke destroy comment
1238 EV/EV 400000 244 0.56 0.46 0.31 EV native interface 1970 EV/EV 400000 224 0.47 0.35 0.27 EV native interface
1239 EV/Any 100000 244 2.50 0.46 0.29 EV + AnyEvent watchers 1971 EV/Any 100000 224 2.88 0.34 0.27 EV + AnyEvent watchers
1240 CoroEV/Any 100000 244 2.49 0.44 0.29 coroutines + Coro::Signal 1972 CoroEV/Any 100000 224 2.85 0.35 0.28 coroutines + Coro::Signal
1241 Perl/Any 100000 513 4.92 0.87 1.12 pure perl implementation 1973 Perl/Any 100000 452 4.13 0.73 0.95 pure perl implementation
1242 Event/Event 16000 516 31.88 31.30 0.85 Event native interface 1974 Event/Event 16000 517 32.20 31.80 0.81 Event native interface
1243 Event/Any 16000 590 35.75 31.42 1.08 Event + AnyEvent watchers 1975 Event/Any 16000 590 35.85 31.55 1.06 Event + AnyEvent watchers
1976 IOAsync/Any 16000 989 38.10 32.77 11.13 via IO::Async::Loop::IO_Poll
1977 IOAsync/Any 16000 990 37.59 29.50 10.61 via IO::Async::Loop::Epoll
1244 Glib/Any 16000 1357 98.22 12.41 54.00 quadratic behaviour 1978 Glib/Any 16000 1357 102.33 12.31 51.00 quadratic behaviour
1245 Tk/Any 2000 1860 26.97 67.98 14.00 SEGV with >> 2000 watchers 1979 Tk/Any 2000 1860 27.20 66.31 14.00 SEGV with >> 2000 watchers
1246 POE/Event 2000 6644 108.64 736.02 14.73 via POE::Loop::Event 1980 POE/Event 2000 6328 109.99 751.67 14.02 via POE::Loop::Event
1247 POE/Select 2000 6343 94.13 809.12 565.96 via POE::Loop::Select 1981 POE/Select 2000 6027 94.54 809.13 579.80 via POE::Loop::Select
1248 1982
1249=head3 Discussion 1983=head3 Discussion
1250 1984
1251The benchmark does I<not> measure scalability of the event loop very 1985The benchmark does I<not> measure scalability of the event loop very
1252well. For example, a select-based event loop (such as the pure perl one) 1986well. For example, a select-based event loop (such as the pure perl one)
1277performance becomes really bad with lots of file descriptors (and few of 2011performance becomes really bad with lots of file descriptors (and few of
1278them active), of course, but this was not subject of this benchmark. 2012them active), of course, but this was not subject of this benchmark.
1279 2013
1280The C<Event> module has a relatively high setup and callback invocation 2014The C<Event> module has a relatively high setup and callback invocation
1281cost, but overall scores in on the third place. 2015cost, but overall scores in on the third place.
2016
2017C<IO::Async> performs admirably well, about on par with C<Event>, even
2018when using its pure perl backend.
1282 2019
1283C<Glib>'s memory usage is quite a bit higher, but it features a 2020C<Glib>'s memory usage is quite a bit higher, but it features a
1284faster callback invocation and overall ends up in the same class as 2021faster callback invocation and overall ends up in the same class as
1285C<Event>. However, Glib scales extremely badly, doubling the number of 2022C<Event>. However, Glib scales extremely badly, doubling the number of
1286watchers increases the processing time by more than a factor of four, 2023watchers increases the processing time by more than a factor of four,
1330 2067
1331=back 2068=back
1332 2069
1333=head2 BENCHMARKING THE LARGE SERVER CASE 2070=head2 BENCHMARKING THE LARGE SERVER CASE
1334 2071
1335This benchmark atcually benchmarks the event loop itself. It works by 2072This benchmark actually benchmarks the event loop itself. It works by
1336creating a number of "servers": each server consists of a socketpair, a 2073creating a number of "servers": each server consists of a socket pair, a
1337timeout watcher that gets reset on activity (but never fires), and an I/O 2074timeout watcher that gets reset on activity (but never fires), and an I/O
1338watcher waiting for input on one side of the socket. Each time the socket 2075watcher waiting for input on one side of the socket. Each time the socket
1339watcher reads a byte it will write that byte to a random other "server". 2076watcher reads a byte it will write that byte to a random other "server".
1340 2077
1341The effect is that there will be a lot of I/O watchers, only part of which 2078The effect is that there will be a lot of I/O watchers, only part of which
1342are active at any one point (so there is a constant number of active 2079are active at any one point (so there is a constant number of active
1343fds for each loop iterstaion, but which fds these are is random). The 2080fds for each loop iteration, but which fds these are is random). The
1344timeout is reset each time something is read because that reflects how 2081timeout is reset each time something is read because that reflects how
1345most timeouts work (and puts extra pressure on the event loops). 2082most timeouts work (and puts extra pressure on the event loops).
1346 2083
1347In this benchmark, we use 10000 socketpairs (20000 sockets), of which 100 2084In this benchmark, we use 10000 socket pairs (20000 sockets), of which 100
1348(1%) are active. This mirrors the activity of large servers with many 2085(1%) are active. This mirrors the activity of large servers with many
1349connections, most of which are idle at any one point in time. 2086connections, most of which are idle at any one point in time.
1350 2087
1351Source code for this benchmark is found as F<eg/bench2> in the AnyEvent 2088Source code for this benchmark is found as F<eg/bench2> in the AnyEvent
1352distribution. 2089distribution.
1354=head3 Explanation of the columns 2091=head3 Explanation of the columns
1355 2092
1356I<sockets> is the number of sockets, and twice the number of "servers" (as 2093I<sockets> is the number of sockets, and twice the number of "servers" (as
1357each server has a read and write socket end). 2094each server has a read and write socket end).
1358 2095
1359I<create> is the time it takes to create a socketpair (which is 2096I<create> is the time it takes to create a socket pair (which is
1360nontrivial) and two watchers: an I/O watcher and a timeout watcher. 2097nontrivial) and two watchers: an I/O watcher and a timeout watcher.
1361 2098
1362I<request>, the most important value, is the time it takes to handle a 2099I<request>, the most important value, is the time it takes to handle a
1363single "request", that is, reading the token from the pipe and forwarding 2100single "request", that is, reading the token from the pipe and forwarding
1364it to another server. This includes deleting the old timeout and creating 2101it to another server. This includes deleting the old timeout and creating
1365a new one that moves the timeout into the future. 2102a new one that moves the timeout into the future.
1366 2103
1367=head3 Results 2104=head3 Results
1368 2105
1369 name sockets create request 2106 name sockets create request
1370 EV 20000 69.01 11.16 2107 EV 20000 69.01 11.16
1371 Perl 20000 73.32 35.87 2108 Perl 20000 73.32 35.87
2109 IOAsync 20000 157.00 98.14 epoll
2110 IOAsync 20000 159.31 616.06 poll
1372 Event 20000 212.62 257.32 2111 Event 20000 212.62 257.32
1373 Glib 20000 651.16 1896.30 2112 Glib 20000 651.16 1896.30
1374 POE 20000 349.67 12317.24 uses POE::Loop::Event 2113 POE 20000 349.67 12317.24 uses POE::Loop::Event
1375 2114
1376=head3 Discussion 2115=head3 Discussion
1377 2116
1378This benchmark I<does> measure scalability and overall performance of the 2117This benchmark I<does> measure scalability and overall performance of the
1379particular event loop. 2118particular event loop.
1381EV is again fastest. Since it is using epoll on my system, the setup time 2120EV is again fastest. Since it is using epoll on my system, the setup time
1382is relatively high, though. 2121is relatively high, though.
1383 2122
1384Perl surprisingly comes second. It is much faster than the C-based event 2123Perl surprisingly comes second. It is much faster than the C-based event
1385loops Event and Glib. 2124loops Event and Glib.
2125
2126IO::Async performs very well when using its epoll backend, and still quite
2127good compared to Glib when using its pure perl backend.
1386 2128
1387Event suffers from high setup time as well (look at its code and you will 2129Event suffers from high setup time as well (look at its code and you will
1388understand why). Callback invocation also has a high overhead compared to 2130understand why). Callback invocation also has a high overhead compared to
1389the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event 2131the C<< $_->() for .. >>-style loop that the Perl event loop uses. Event
1390uses select or poll in basically all documented configurations. 2132uses select or poll in basically all documented configurations.
1437speed most when you have lots of watchers, not when you only have a few of 2179speed most when you have lots of watchers, not when you only have a few of
1438them). 2180them).
1439 2181
1440EV is again fastest. 2182EV is again fastest.
1441 2183
1442Perl again comes second. It is noticably faster than the C-based event 2184Perl again comes second. It is noticeably faster than the C-based event
1443loops Event and Glib, although the difference is too small to really 2185loops Event and Glib, although the difference is too small to really
1444matter. 2186matter.
1445 2187
1446POE also performs much better in this case, but is is still far behind the 2188POE also performs much better in this case, but is is still far behind the
1447others. 2189others.
1450 2192
1451=over 4 2193=over 4
1452 2194
1453=item * C-based event loops perform very well with small number of 2195=item * C-based event loops perform very well with small number of
1454watchers, as the management overhead dominates. 2196watchers, as the management overhead dominates.
2197
2198=back
2199
2200=head2 THE IO::Lambda BENCHMARK
2201
2202Recently I was told about the benchmark in the IO::Lambda manpage, which
2203could be misinterpreted to make AnyEvent look bad. In fact, the benchmark
2204simply compares IO::Lambda with POE, and IO::Lambda looks better (which
2205shouldn't come as a surprise to anybody). As such, the benchmark is
2206fine, and mostly shows that the AnyEvent backend from IO::Lambda isn't
2207very optimal. But how would AnyEvent compare when used without the extra
2208baggage? To explore this, I wrote the equivalent benchmark for AnyEvent.
2209
2210The benchmark itself creates an echo-server, and then, for 500 times,
2211connects to the echo server, sends a line, waits for the reply, and then
2212creates the next connection. This is a rather bad benchmark, as it doesn't
2213test the efficiency of the framework or much non-blocking I/O, but it is a
2214benchmark nevertheless.
2215
2216 name runtime
2217 Lambda/select 0.330 sec
2218 + optimized 0.122 sec
2219 Lambda/AnyEvent 0.327 sec
2220 + optimized 0.138 sec
2221 Raw sockets/select 0.077 sec
2222 POE/select, components 0.662 sec
2223 POE/select, raw sockets 0.226 sec
2224 POE/select, optimized 0.404 sec
2225
2226 AnyEvent/select/nb 0.085 sec
2227 AnyEvent/EV/nb 0.068 sec
2228 +state machine 0.134 sec
2229
2230The benchmark is also a bit unfair (my fault): the IO::Lambda/POE
2231benchmarks actually make blocking connects and use 100% blocking I/O,
2232defeating the purpose of an event-based solution. All of the newly
2233written AnyEvent benchmarks use 100% non-blocking connects (using
2234AnyEvent::Socket::tcp_connect and the asynchronous pure perl DNS
2235resolver), so AnyEvent is at a disadvantage here, as non-blocking connects
2236generally require a lot more bookkeeping and event handling than blocking
2237connects (which involve a single syscall only).
2238
2239The last AnyEvent benchmark additionally uses L<AnyEvent::Handle>, which
2240offers similar expressive power as POE and IO::Lambda, using conventional
2241Perl syntax. This means that both the echo server and the client are 100%
2242non-blocking, further placing it at a disadvantage.
2243
2244As you can see, the AnyEvent + EV combination even beats the
2245hand-optimised "raw sockets benchmark", while AnyEvent + its pure perl
2246backend easily beats IO::Lambda and POE.
2247
2248And even the 100% non-blocking version written using the high-level (and
2249slow :) L<AnyEvent::Handle> abstraction beats both POE and IO::Lambda by a
2250large margin, even though it does all of DNS, tcp-connect and socket I/O
2251in a non-blocking way.
2252
2253The two AnyEvent benchmarks programs can be found as F<eg/ae0.pl> and
2254F<eg/ae2.pl> in the AnyEvent distribution, the remaining benchmarks are
2255part of the IO::lambda distribution and were used without any changes.
2256
2257
2258=head1 SIGNALS
2259
2260AnyEvent currently installs handlers for these signals:
2261
2262=over 4
2263
2264=item SIGCHLD
2265
2266A handler for C<SIGCHLD> is installed by AnyEvent's child watcher
2267emulation for event loops that do not support them natively. Also, some
2268event loops install a similar handler.
2269
2270Additionally, when AnyEvent is loaded and SIGCHLD is set to IGNORE, then
2271AnyEvent will reset it to default, to avoid losing child exit statuses.
2272
2273=item SIGPIPE
2274
2275A no-op handler is installed for C<SIGPIPE> when C<$SIG{PIPE}> is C<undef>
2276when AnyEvent gets loaded.
2277
2278The rationale for this is that AnyEvent users usually do not really depend
2279on SIGPIPE delivery (which is purely an optimisation for shell use, or
2280badly-written programs), but C<SIGPIPE> can cause spurious and rare
2281program exits as a lot of people do not expect C<SIGPIPE> when writing to
2282some random socket.
2283
2284The rationale for installing a no-op handler as opposed to ignoring it is
2285that this way, the handler will be restored to defaults on exec.
2286
2287Feel free to install your own handler, or reset it to defaults.
2288
2289=back
2290
2291=cut
2292
2293undef $SIG{CHLD}
2294 if $SIG{CHLD} eq 'IGNORE';
2295
2296$SIG{PIPE} = sub { }
2297 unless defined $SIG{PIPE};
2298
2299=head1 RECOMMENDED/OPTIONAL MODULES
2300
2301One of AnyEvent's main goals is to be 100% Pure-Perl(tm): only perl (and
2302it's built-in modules) are required to use it.
2303
2304That does not mean that AnyEvent won't take advantage of some additional
2305modules if they are installed.
2306
2307This section epxlains which additional modules will be used, and how they
2308affect AnyEvent's operetion.
2309
2310=over 4
2311
2312=item L<Async::Interrupt>
2313
2314This slightly arcane module is used to implement fast signal handling: To
2315my knowledge, there is no way to do completely race-free and quick
2316signal handling in pure perl. To ensure that signals still get
2317delivered, AnyEvent will start an interval timer to wake up perl (and
2318catch the signals) with some delay (default is 10 seconds, look for
2319C<$AnyEvent::MAX_SIGNAL_LATENCY>).
2320
2321If this module is available, then it will be used to implement signal
2322catching, which means that signals will not be delayed, and the event loop
2323will not be interrupted regularly, which is more efficient (And good for
2324battery life on laptops).
2325
2326This affects not just the pure-perl event loop, but also other event loops
2327that have no signal handling on their own (e.g. Glib, Tk, Qt).
2328
2329Some event loops (POE, Event, Event::Lib) offer signal watchers natively,
2330and either employ their own workarounds (POE) or use AnyEvent's workaround
2331(using C<$AnyEvent::MAX_SIGNAL_LATENCY>). Installing L<Async::Interrupt>
2332does nothing for those backends.
2333
2334=item L<EV>
2335
2336This module isn't really "optional", as it is simply one of the backend
2337event loops that AnyEvent can use. However, it is simply the best event
2338loop available in terms of features, speed and stability: It supports
2339the AnyEvent API optimally, implements all the watcher types in XS, does
2340automatic timer adjustments even when no monotonic clock is available,
2341can take avdantage of advanced kernel interfaces such as C<epoll> and
2342C<kqueue>, and is the fastest backend I<by far>. You can even embed
2343L<Glib>/L<Gtk2> in it (or vice versa, see L<EV::Glib> and L<Glib::EV>).
2344
2345=item L<Guard>
2346
2347The guard module, when used, will be used to implement
2348C<AnyEvent::Util::guard>. This speeds up guards considerably (and uses a
2349lot less memory), but otherwise doesn't affect guard operation much. It is
2350purely used for performance.
2351
2352=item L<JSON> and L<JSON::XS>
2353
2354This module is required when you want to read or write JSON data via
2355L<AnyEvent::Handle>. It is also written in pure-perl, but can take
2356advantage of the ultra-high-speed L<JSON::XS> module when it is installed.
2357
2358In fact, L<AnyEvent::Handle> will use L<JSON::XS> by default if it is
2359installed.
2360
2361=item L<Net::SSLeay>
2362
2363Implementing TLS/SSL in Perl is certainly interesting, but not very
2364worthwhile: If this module is installed, then L<AnyEvent::Handle> (with
2365the help of L<AnyEvent::TLS>), gains the ability to do TLS/SSL.
2366
2367=item L<Time::HiRes>
2368
2369This module is part of perl since release 5.008. It will be used when the
2370chosen event library does not come with a timing source on it's own. The
2371pure-perl event loop (L<AnyEvent::Impl::Perl>) will additionally use it to
2372try to use a monotonic clock for timing stability.
1455 2373
1456=back 2374=back
1457 2375
1458 2376
1459=head1 FORK 2377=head1 FORK
1461Most event libraries are not fork-safe. The ones who are usually are 2379Most event libraries are not fork-safe. The ones who are usually are
1462because they rely on inefficient but fork-safe C<select> or C<poll> 2380because they rely on inefficient but fork-safe C<select> or C<poll>
1463calls. Only L<EV> is fully fork-aware. 2381calls. Only L<EV> is fully fork-aware.
1464 2382
1465If you have to fork, you must either do so I<before> creating your first 2383If you have to fork, you must either do so I<before> creating your first
1466watcher OR you must not use AnyEvent at all in the child. 2384watcher OR you must not use AnyEvent at all in the child OR you must do
2385something completely out of the scope of AnyEvent.
1467 2386
1468 2387
1469=head1 SECURITY CONSIDERATIONS 2388=head1 SECURITY CONSIDERATIONS
1470 2389
1471AnyEvent can be forced to load any event model via 2390AnyEvent can be forced to load any event model via
1476specified in the variable. 2395specified in the variable.
1477 2396
1478You can make AnyEvent completely ignore this variable by deleting it 2397You can make AnyEvent completely ignore this variable by deleting it
1479before the first watcher gets created, e.g. with a C<BEGIN> block: 2398before the first watcher gets created, e.g. with a C<BEGIN> block:
1480 2399
1481 BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} } 2400 BEGIN { delete $ENV{PERL_ANYEVENT_MODEL} }
1482 2401
1483 use AnyEvent; 2402 use AnyEvent;
1484 2403
1485Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can 2404Similar considerations apply to $ENV{PERL_ANYEVENT_VERBOSE}, as that can
1486be used to probe what backend is used and gain other information (which is 2405be used to probe what backend is used and gain other information (which is
1487probably even less useful to an attacker than PERL_ANYEVENT_MODEL). 2406probably even less useful to an attacker than PERL_ANYEVENT_MODEL), and
2407$ENV{PERL_ANYEVENT_STRICT}.
2408
2409Note that AnyEvent will remove I<all> environment variables starting with
2410C<PERL_ANYEVENT_> from C<%ENV> when it is loaded while taint mode is
2411enabled.
2412
2413
2414=head1 BUGS
2415
2416Perl 5.8 has numerous memleaks that sometimes hit this module and are hard
2417to work around. If you suffer from memleaks, first upgrade to Perl 5.10
2418and check wether the leaks still show up. (Perl 5.10.0 has other annoying
2419memleaks, such as leaking on C<map> and C<grep> but it is usually not as
2420pronounced).
1488 2421
1489 2422
1490=head1 SEE ALSO 2423=head1 SEE ALSO
2424
2425Utility functions: L<AnyEvent::Util>.
1491 2426
1492Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>, 2427Event modules: L<EV>, L<EV::Glib>, L<Glib::EV>, L<Event>, L<Glib::Event>,
1493L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>. 2428L<Glib>, L<Tk>, L<Event::Lib>, L<Qt>, L<POE>.
1494 2429
1495Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>, 2430Implementations: L<AnyEvent::Impl::EV>, L<AnyEvent::Impl::Event>,
1496L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>, 2431L<AnyEvent::Impl::Glib>, L<AnyEvent::Impl::Tk>, L<AnyEvent::Impl::Perl>,
1497L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>, 2432L<AnyEvent::Impl::EventLib>, L<AnyEvent::Impl::Qt>,
1498L<AnyEvent::Impl::POE>. 2433L<AnyEvent::Impl::POE>, L<AnyEvent::Impl::IOAsync>, L<Anyevent::Impl::Irssi>.
1499 2434
2435Non-blocking file handles, sockets, TCP clients and
2436servers: L<AnyEvent::Handle>, L<AnyEvent::Socket>, L<AnyEvent::TLS>.
2437
2438Asynchronous DNS: L<AnyEvent::DNS>.
2439
1500Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>, L<Coro::Event>, 2440Coroutine support: L<Coro>, L<Coro::AnyEvent>, L<Coro::EV>,
2441L<Coro::Event>,
1501 2442
1502Nontrivial usage examples: L<Net::FCP>, L<Net::XMPP2>. 2443Nontrivial usage examples: L<AnyEvent::GPSD>, L<AnyEvent::XMPP>,
2444L<AnyEvent::HTTP>.
1503 2445
1504 2446
1505=head1 AUTHOR 2447=head1 AUTHOR
1506 2448
1507 Marc Lehmann <schmorp@schmorp.de> 2449 Marc Lehmann <schmorp@schmorp.de>
1508 http://home.schmorp.de/ 2450 http://home.schmorp.de/
1509 2451
1510=cut 2452=cut
1511 2453
15121 24541
1513 2455

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines