ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libev/ev.pod
(Generate patch)

Comparing libev/ev.pod (file contents):
Revision 1.316 by root, Fri Oct 22 09:34:01 2010 UTC vs.
Revision 1.399 by root, Mon Apr 2 23:14:41 2012 UTC

43 43
44 int 44 int
45 main (void) 45 main (void)
46 { 46 {
47 // use the default event loop unless you have special needs 47 // use the default event loop unless you have special needs
48 struct ev_loop *loop = ev_default_loop (0); 48 struct ev_loop *loop = EV_DEFAULT;
49 49
50 // initialise an io watcher, then start it 50 // initialise an io watcher, then start it
51 // this one will watch for stdin to become readable 51 // this one will watch for stdin to become readable
52 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ); 52 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
53 ev_io_start (loop, &stdin_watcher); 53 ev_io_start (loop, &stdin_watcher);
58 ev_timer_start (loop, &timeout_watcher); 58 ev_timer_start (loop, &timeout_watcher);
59 59
60 // now wait for events to arrive 60 // now wait for events to arrive
61 ev_run (loop, 0); 61 ev_run (loop, 0);
62 62
63 // unloop was called, so exit 63 // break was called, so exit
64 return 0; 64 return 0;
65 } 65 }
66 66
67=head1 ABOUT THIS DOCUMENT 67=head1 ABOUT THIS DOCUMENT
68 68
77on event-based programming, nor will it introduce event-based programming 77on event-based programming, nor will it introduce event-based programming
78with libev. 78with libev.
79 79
80Familiarity with event based programming techniques in general is assumed 80Familiarity with event based programming techniques in general is assumed
81throughout this document. 81throughout this document.
82
83=head1 WHAT TO READ WHEN IN A HURRY
84
85This manual tries to be very detailed, but unfortunately, this also makes
86it very long. If you just want to know the basics of libev, I suggest
87reading L<ANATOMY OF A WATCHER>, then the L<EXAMPLE PROGRAM> above and
88look up the missing functions in L<GLOBAL FUNCTIONS> and the C<ev_io> and
89C<ev_timer> sections in L<WATCHER TYPES>.
82 90
83=head1 ABOUT LIBEV 91=head1 ABOUT LIBEV
84 92
85Libev is an event loop: you register interest in certain events (such as a 93Libev is an event loop: you register interest in certain events (such as a
86file descriptor being readable or a timeout occurring), and it will manage 94file descriptor being readable or a timeout occurring), and it will manage
124this argument. 132this argument.
125 133
126=head2 TIME REPRESENTATION 134=head2 TIME REPRESENTATION
127 135
128Libev represents time as a single floating point number, representing 136Libev represents time as a single floating point number, representing
129the (fractional) number of seconds since the (POSIX) epoch (in practise 137the (fractional) number of seconds since the (POSIX) epoch (in practice
130somewhere near the beginning of 1970, details are complicated, don't 138somewhere near the beginning of 1970, details are complicated, don't
131ask). This type is called C<ev_tstamp>, which is what you should use 139ask). This type is called C<ev_tstamp>, which is what you should use
132too. It usually aliases to the C<double> type in C. When you need to do 140too. It usually aliases to the C<double> type in C. When you need to do
133any calculations on it, you should treat it as some floating point value. 141any calculations on it, you should treat it as some floating point value.
134 142
165 173
166=item ev_tstamp ev_time () 174=item ev_tstamp ev_time ()
167 175
168Returns the current time as libev would use it. Please note that the 176Returns the current time as libev would use it. Please note that the
169C<ev_now> function is usually faster and also often returns the timestamp 177C<ev_now> function is usually faster and also often returns the timestamp
170you actually want to know. 178you actually want to know. Also interesting is the combination of
179C<ev_now_update> and C<ev_now>.
171 180
172=item ev_sleep (ev_tstamp interval) 181=item ev_sleep (ev_tstamp interval)
173 182
174Sleep for the given interval: The current thread will be blocked until 183Sleep for the given interval: The current thread will be blocked
175either it is interrupted or the given time interval has passed. Basically 184until either it is interrupted or the given time interval has
185passed (approximately - it might return a bit earlier even if not
186interrupted). Returns immediately if C<< interval <= 0 >>.
187
176this is a sub-second-resolution C<sleep ()>. 188Basically this is a sub-second-resolution C<sleep ()>.
189
190The range of the C<interval> is limited - libev only guarantees to work
191with sleep times of up to one day (C<< interval <= 86400 >>).
177 192
178=item int ev_version_major () 193=item int ev_version_major ()
179 194
180=item int ev_version_minor () 195=item int ev_version_minor ()
181 196
192as this indicates an incompatible change. Minor versions are usually 207as this indicates an incompatible change. Minor versions are usually
193compatible to older versions, so a larger minor version alone is usually 208compatible to older versions, so a larger minor version alone is usually
194not a problem. 209not a problem.
195 210
196Example: Make sure we haven't accidentally been linked against the wrong 211Example: Make sure we haven't accidentally been linked against the wrong
197version (note, however, that this will not detect ABI mismatches :). 212version (note, however, that this will not detect other ABI mismatches,
213such as LFS or reentrancy).
198 214
199 assert (("libev version mismatch", 215 assert (("libev version mismatch",
200 ev_version_major () == EV_VERSION_MAJOR 216 ev_version_major () == EV_VERSION_MAJOR
201 && ev_version_minor () >= EV_VERSION_MINOR)); 217 && ev_version_minor () >= EV_VERSION_MINOR));
202 218
213 assert (("sorry, no epoll, no sex", 229 assert (("sorry, no epoll, no sex",
214 ev_supported_backends () & EVBACKEND_EPOLL)); 230 ev_supported_backends () & EVBACKEND_EPOLL));
215 231
216=item unsigned int ev_recommended_backends () 232=item unsigned int ev_recommended_backends ()
217 233
218Return the set of all backends compiled into this binary of libev and also 234Return the set of all backends compiled into this binary of libev and
219recommended for this platform. This set is often smaller than the one 235also recommended for this platform, meaning it will work for most file
236descriptor types. This set is often smaller than the one returned by
220returned by C<ev_supported_backends>, as for example kqueue is broken on 237C<ev_supported_backends>, as for example kqueue is broken on most BSDs
221most BSDs and will not be auto-detected unless you explicitly request it 238and will not be auto-detected unless you explicitly request it (assuming
222(assuming you know what you are doing). This is the set of backends that 239you know what you are doing). This is the set of backends that libev will
223libev will probe for if you specify no backends explicitly. 240probe for if you specify no backends explicitly.
224 241
225=item unsigned int ev_embeddable_backends () 242=item unsigned int ev_embeddable_backends ()
226 243
227Returns the set of backends that are embeddable in other event loops. This 244Returns the set of backends that are embeddable in other event loops. This
228is the theoretical, all-platform, value. To find which backends 245value is platform-specific but can include backends not available on the
229might be supported on the current system, you would need to look at 246current system. To find which embeddable backends might be supported on
230C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for 247the current system, you would need to look at C<ev_embeddable_backends ()
231recommended ones. 248& ev_supported_backends ()>, likewise for recommended ones.
232 249
233See the description of C<ev_embed> watchers for more info. 250See the description of C<ev_embed> watchers for more info.
234 251
235=item ev_set_allocator (void *(*cb)(void *ptr, long size)) [NOT REENTRANT] 252=item ev_set_allocator (void *(*cb)(void *ptr, long size))
236 253
237Sets the allocation function to use (the prototype is similar - the 254Sets the allocation function to use (the prototype is similar - the
238semantics are identical to the C<realloc> C89/SuS/POSIX function). It is 255semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
239used to allocate and free memory (no surprises here). If it returns zero 256used to allocate and free memory (no surprises here). If it returns zero
240when memory needs to be allocated (C<size != 0>), the library might abort 257when memory needs to be allocated (C<size != 0>), the library might abort
266 } 283 }
267 284
268 ... 285 ...
269 ev_set_allocator (persistent_realloc); 286 ev_set_allocator (persistent_realloc);
270 287
271=item ev_set_syserr_cb (void (*cb)(const char *msg)); [NOT REENTRANT] 288=item ev_set_syserr_cb (void (*cb)(const char *msg))
272 289
273Set the callback function to call on a retryable system call error (such 290Set the callback function to call on a retryable system call error (such
274as failed select, poll, epoll_wait). The message is a printable string 291as failed select, poll, epoll_wait). The message is a printable string
275indicating the system call or subsystem causing the problem. If this 292indicating the system call or subsystem causing the problem. If this
276callback is set, then libev will expect it to remedy the situation, no 293callback is set, then libev will expect it to remedy the situation, no
288 } 305 }
289 306
290 ... 307 ...
291 ev_set_syserr_cb (fatal_error); 308 ev_set_syserr_cb (fatal_error);
292 309
310=item ev_feed_signal (int signum)
311
312This function can be used to "simulate" a signal receive. It is completely
313safe to call this function at any time, from any context, including signal
314handlers or random threads.
315
316Its main use is to customise signal handling in your process, especially
317in the presence of threads. For example, you could block signals
318by default in all threads (and specifying C<EVFLAG_NOSIGMASK> when
319creating any loops), and in one thread, use C<sigwait> or any other
320mechanism to wait for signals, then "deliver" them to libev by calling
321C<ev_feed_signal>.
322
293=back 323=back
294 324
295=head1 FUNCTIONS CONTROLLING THE EVENT LOOP 325=head1 FUNCTIONS CONTROLLING EVENT LOOPS
296 326
297An event loop is described by a C<struct ev_loop *> (the C<struct> is 327An event loop is described by a C<struct ev_loop *> (the C<struct> is
298I<not> optional in this case unless libev 3 compatibility is disabled, as 328I<not> optional in this case unless libev 3 compatibility is disabled, as
299libev 3 had an C<ev_loop> function colliding with the struct name). 329libev 3 had an C<ev_loop> function colliding with the struct name).
300 330
301The library knows two types of such loops, the I<default> loop, which 331The library knows two types of such loops, the I<default> loop, which
302supports signals and child events, and dynamically created event loops 332supports child process events, and dynamically created event loops which
303which do not. 333do not.
304 334
305=over 4 335=over 4
306 336
307=item struct ev_loop *ev_default_loop (unsigned int flags) 337=item struct ev_loop *ev_default_loop (unsigned int flags)
308 338
309This will initialise the default event loop if it hasn't been initialised 339This returns the "default" event loop object, which is what you should
310yet and return it. If the default loop could not be initialised, returns 340normally use when you just need "the event loop". Event loop objects and
311false. If it already was initialised it simply returns it (and ignores the 341the C<flags> parameter are described in more detail in the entry for
312flags. If that is troubling you, check C<ev_backend ()> afterwards). 342C<ev_loop_new>.
343
344If the default loop is already initialised then this function simply
345returns it (and ignores the flags. If that is troubling you, check
346C<ev_backend ()> afterwards). Otherwise it will create it with the given
347flags, which should almost always be C<0>, unless the caller is also the
348one calling C<ev_run> or otherwise qualifies as "the main program".
313 349
314If you don't know what event loop to use, use the one returned from this 350If you don't know what event loop to use, use the one returned from this
315function. 351function (or via the C<EV_DEFAULT> macro).
316 352
317Note that this function is I<not> thread-safe, so if you want to use it 353Note that this function is I<not> thread-safe, so if you want to use it
318from multiple threads, you have to lock (note also that this is unlikely, 354from multiple threads, you have to employ some kind of mutex (note also
319as loops cannot be shared easily between threads anyway). 355that this case is unlikely, as loops cannot be shared easily between
356threads anyway).
320 357
321The default loop is the only loop that can handle C<ev_signal> and 358The default loop is the only loop that can handle C<ev_child> watchers,
322C<ev_child> watchers, and to do this, it always registers a handler 359and to do this, it always registers a handler for C<SIGCHLD>. If this is
323for C<SIGCHLD>. If this is a problem for your application you can either 360a problem for your application you can either create a dynamic loop with
324create a dynamic loop with C<ev_loop_new> that doesn't do that, or you 361C<ev_loop_new> which doesn't do that, or you can simply overwrite the
325can simply overwrite the C<SIGCHLD> signal handler I<after> calling 362C<SIGCHLD> signal handler I<after> calling C<ev_default_init>.
326C<ev_default_init>. 363
364Example: This is the most typical usage.
365
366 if (!ev_default_loop (0))
367 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
368
369Example: Restrict libev to the select and poll backends, and do not allow
370environment settings to be taken into account:
371
372 ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
373
374=item struct ev_loop *ev_loop_new (unsigned int flags)
375
376This will create and initialise a new event loop object. If the loop
377could not be initialised, returns false.
378
379This function is thread-safe, and one common way to use libev with
380threads is indeed to create one loop per thread, and using the default
381loop in the "main" or "initial" thread.
327 382
328The flags argument can be used to specify special behaviour or specific 383The flags argument can be used to specify special behaviour or specific
329backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>). 384backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
330 385
331The following flags are supported: 386The following flags are supported:
366environment variable. 421environment variable.
367 422
368=item C<EVFLAG_NOINOTIFY> 423=item C<EVFLAG_NOINOTIFY>
369 424
370When this flag is specified, then libev will not attempt to use the 425When this flag is specified, then libev will not attempt to use the
371I<inotify> API for it's C<ev_stat> watchers. Apart from debugging and 426I<inotify> API for its C<ev_stat> watchers. Apart from debugging and
372testing, this flag can be useful to conserve inotify file descriptors, as 427testing, this flag can be useful to conserve inotify file descriptors, as
373otherwise each loop using C<ev_stat> watchers consumes one inotify handle. 428otherwise each loop using C<ev_stat> watchers consumes one inotify handle.
374 429
375=item C<EVFLAG_SIGNALFD> 430=item C<EVFLAG_SIGNALFD>
376 431
377When this flag is specified, then libev will attempt to use the 432When this flag is specified, then libev will attempt to use the
378I<signalfd> API for it's C<ev_signal> (and C<ev_child>) watchers. This API 433I<signalfd> API for its C<ev_signal> (and C<ev_child>) watchers. This API
379delivers signals synchronously, which makes it both faster and might make 434delivers signals synchronously, which makes it both faster and might make
380it possible to get the queued signal data. It can also simplify signal 435it possible to get the queued signal data. It can also simplify signal
381handling with threads, as long as you properly block signals in your 436handling with threads, as long as you properly block signals in your
382threads that are not interested in handling them. 437threads that are not interested in handling them.
383 438
384Signalfd will not be used by default as this changes your signal mask, and 439Signalfd will not be used by default as this changes your signal mask, and
385there are a lot of shoddy libraries and programs (glib's threadpool for 440there are a lot of shoddy libraries and programs (glib's threadpool for
386example) that can't properly initialise their signal masks. 441example) that can't properly initialise their signal masks.
442
443=item C<EVFLAG_NOSIGMASK>
444
445When this flag is specified, then libev will avoid to modify the signal
446mask. Specifically, this means you have to make sure signals are unblocked
447when you want to receive them.
448
449This behaviour is useful when you want to do your own signal handling, or
450want to handle signals only in specific threads and want to avoid libev
451unblocking the signals.
452
453It's also required by POSIX in a threaded program, as libev calls
454C<sigprocmask>, whose behaviour is officially unspecified.
455
456This flag's behaviour will become the default in future versions of libev.
387 457
388=item C<EVBACKEND_SELECT> (value 1, portable select backend) 458=item C<EVBACKEND_SELECT> (value 1, portable select backend)
389 459
390This is your standard select(2) backend. Not I<completely> standard, as 460This is your standard select(2) backend. Not I<completely> standard, as
391libev tries to roll its own fd_set with no limits on the number of fds, 461libev tries to roll its own fd_set with no limits on the number of fds,
419=item C<EVBACKEND_EPOLL> (value 4, Linux) 489=item C<EVBACKEND_EPOLL> (value 4, Linux)
420 490
421Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9 491Use the linux-specific epoll(7) interface (for both pre- and post-2.6.9
422kernels). 492kernels).
423 493
424For few fds, this backend is a bit little slower than poll and select, 494For few fds, this backend is a bit little slower than poll and select, but
425but it scales phenomenally better. While poll and select usually scale 495it scales phenomenally better. While poll and select usually scale like
426like O(total_fds) where n is the total number of fds (or the highest fd), 496O(total_fds) where total_fds is the total number of fds (or the highest
427epoll scales either O(1) or O(active_fds). 497fd), epoll scales either O(1) or O(active_fds).
428 498
429The epoll mechanism deserves honorable mention as the most misdesigned 499The epoll mechanism deserves honorable mention as the most misdesigned
430of the more advanced event mechanisms: mere annoyances include silently 500of the more advanced event mechanisms: mere annoyances include silently
431dropping file descriptors, requiring a system call per change per file 501dropping file descriptors, requiring a system call per change per file
432descriptor (and unnecessary guessing of parameters), problems with dup and 502descriptor (and unnecessary guessing of parameters), problems with dup,
503returning before the timeout value, resulting in additional iterations
504(and only giving 5ms accuracy while select on the same platform gives
433so on. The biggest issue is fork races, however - if a program forks then 5050.1ms) and so on. The biggest issue is fork races, however - if a program
434I<both> parent and child process have to recreate the epoll set, which can 506forks then I<both> parent and child process have to recreate the epoll
435take considerable time (one syscall per file descriptor) and is of course 507set, which can take considerable time (one syscall per file descriptor)
436hard to detect. 508and is of course hard to detect.
437 509
438Epoll is also notoriously buggy - embedding epoll fds I<should> work, but 510Epoll is also notoriously buggy - embedding epoll fds I<should> work,
439of course I<doesn't>, and epoll just loves to report events for totally 511but of course I<doesn't>, and epoll just loves to report events for
440I<different> file descriptors (even already closed ones, so one cannot 512totally I<different> file descriptors (even already closed ones, so
441even remove them from the set) than registered in the set (especially 513one cannot even remove them from the set) than registered in the set
442on SMP systems). Libev tries to counter these spurious notifications by 514(especially on SMP systems). Libev tries to counter these spurious
443employing an additional generation counter and comparing that against the 515notifications by employing an additional generation counter and comparing
444events to filter out spurious ones, recreating the set when required. Last 516that against the events to filter out spurious ones, recreating the set
517when required. Epoll also erroneously rounds down timeouts, but gives you
518no way to know when and by how much, so sometimes you have to busy-wait
519because epoll returns immediately despite a nonzero timeout. And last
445not least, it also refuses to work with some file descriptors which work 520not least, it also refuses to work with some file descriptors which work
446perfectly fine with C<select> (files, many character devices...). 521perfectly fine with C<select> (files, many character devices...).
522
523Epoll is truly the train wreck among event poll mechanisms, a frankenpoll,
524cobbled together in a hurry, no thought to design or interaction with
525others. Oh, the pain, will it ever stop...
447 526
448While stopping, setting and starting an I/O watcher in the same iteration 527While stopping, setting and starting an I/O watcher in the same iteration
449will result in some caching, there is still a system call per such 528will result in some caching, there is still a system call per such
450incident (because the same I<file descriptor> could point to a different 529incident (because the same I<file descriptor> could point to a different
451I<file description> now), so its best to avoid that. Also, C<dup ()>'ed 530I<file description> now), so its best to avoid that. Also, C<dup ()>'ed
488 567
489It scales in the same way as the epoll backend, but the interface to the 568It scales in the same way as the epoll backend, but the interface to the
490kernel is more efficient (which says nothing about its actual speed, of 569kernel is more efficient (which says nothing about its actual speed, of
491course). While stopping, setting and starting an I/O watcher does never 570course). While stopping, setting and starting an I/O watcher does never
492cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to 571cause an extra system call as with C<EVBACKEND_EPOLL>, it still adds up to
493two event changes per incident. Support for C<fork ()> is very bad (but 572two event changes per incident. Support for C<fork ()> is very bad (you
494sane, unlike epoll) and it drops fds silently in similarly hard-to-detect 573might have to leak fd's on fork, but it's more sane than epoll) and it
495cases 574drops fds silently in similarly hard-to-detect cases
496 575
497This backend usually performs well under most conditions. 576This backend usually performs well under most conditions.
498 577
499While nominally embeddable in other event loops, this doesn't work 578While nominally embeddable in other event loops, this doesn't work
500everywhere, so you might need to test for this. And since it is broken 579everywhere, so you might need to test for this. And since it is broken
517=item C<EVBACKEND_PORT> (value 32, Solaris 10) 596=item C<EVBACKEND_PORT> (value 32, Solaris 10)
518 597
519This uses the Solaris 10 event port mechanism. As with everything on Solaris, 598This uses the Solaris 10 event port mechanism. As with everything on Solaris,
520it's really slow, but it still scales very well (O(active_fds)). 599it's really slow, but it still scales very well (O(active_fds)).
521 600
522Please note that Solaris event ports can deliver a lot of spurious
523notifications, so you need to use non-blocking I/O or other means to avoid
524blocking when no data (or space) is available.
525
526While this backend scales well, it requires one system call per active 601While this backend scales well, it requires one system call per active
527file descriptor per loop iteration. For small and medium numbers of file 602file descriptor per loop iteration. For small and medium numbers of file
528descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend 603descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
529might perform better. 604might perform better.
530 605
531On the positive side, with the exception of the spurious readiness 606On the positive side, this backend actually performed fully to
532notifications, this backend actually performed fully to specification
533in all tests and is fully embeddable, which is a rare feat among the 607specification in all tests and is fully embeddable, which is a rare feat
534OS-specific backends (I vastly prefer correctness over speed hacks). 608among the OS-specific backends (I vastly prefer correctness over speed
609hacks).
610
611On the negative side, the interface is I<bizarre> - so bizarre that
612even sun itself gets it wrong in their code examples: The event polling
613function sometimes returns events to the caller even though an error
614occurred, but with no indication whether it has done so or not (yes, it's
615even documented that way) - deadly for edge-triggered interfaces where you
616absolutely have to know whether an event occurred or not because you have
617to re-arm the watcher.
618
619Fortunately libev seems to be able to work around these idiocies.
535 620
536This backend maps C<EV_READ> and C<EV_WRITE> in the same way as 621This backend maps C<EV_READ> and C<EV_WRITE> in the same way as
537C<EVBACKEND_POLL>. 622C<EVBACKEND_POLL>.
538 623
539=item C<EVBACKEND_ALL> 624=item C<EVBACKEND_ALL>
540 625
541Try all backends (even potentially broken ones that wouldn't be tried 626Try all backends (even potentially broken ones that wouldn't be tried
542with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as 627with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
543C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>. 628C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
544 629
545It is definitely not recommended to use this flag. 630It is definitely not recommended to use this flag, use whatever
631C<ev_recommended_backends ()> returns, or simply do not specify a backend
632at all.
633
634=item C<EVBACKEND_MASK>
635
636Not a backend at all, but a mask to select all backend bits from a
637C<flags> value, in case you want to mask out any backends from a flags
638value (e.g. when modifying the C<LIBEV_FLAGS> environment variable).
546 639
547=back 640=back
548 641
549If one or more of the backend flags are or'ed into the flags value, 642If one or more of the backend flags are or'ed into the flags value,
550then only these backends will be tried (in the reverse order as listed 643then only these backends will be tried (in the reverse order as listed
551here). If none are specified, all backends in C<ev_recommended_backends 644here). If none are specified, all backends in C<ev_recommended_backends
552()> will be tried. 645()> will be tried.
553 646
554Example: This is the most typical usage.
555
556 if (!ev_default_loop (0))
557 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
558
559Example: Restrict libev to the select and poll backends, and do not allow
560environment settings to be taken into account:
561
562 ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
563
564Example: Use whatever libev has to offer, but make sure that kqueue is
565used if available (warning, breaks stuff, best use only with your own
566private event loop and only if you know the OS supports your types of
567fds):
568
569 ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
570
571=item struct ev_loop *ev_loop_new (unsigned int flags)
572
573Similar to C<ev_default_loop>, but always creates a new event loop that is
574always distinct from the default loop.
575
576Note that this function I<is> thread-safe, and one common way to use
577libev with threads is indeed to create one loop per thread, and using the
578default loop in the "main" or "initial" thread.
579
580Example: Try to create a event loop that uses epoll and nothing else. 647Example: Try to create a event loop that uses epoll and nothing else.
581 648
582 struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV); 649 struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
583 if (!epoller) 650 if (!epoller)
584 fatal ("no epoll found here, maybe it hides under your chair"); 651 fatal ("no epoll found here, maybe it hides under your chair");
585 652
653Example: Use whatever libev has to offer, but make sure that kqueue is
654used if available.
655
656 struct ev_loop *loop = ev_loop_new (ev_recommended_backends () | EVBACKEND_KQUEUE);
657
586=item ev_default_destroy () 658=item ev_loop_destroy (loop)
587 659
588Destroys the default loop (frees all memory and kernel state etc.). None 660Destroys an event loop object (frees all memory and kernel state
589of the active event watchers will be stopped in the normal sense, so 661etc.). None of the active event watchers will be stopped in the normal
590e.g. C<ev_is_active> might still return true. It is your responsibility to 662sense, so e.g. C<ev_is_active> might still return true. It is your
591either stop all watchers cleanly yourself I<before> calling this function, 663responsibility to either stop all watchers cleanly yourself I<before>
592or cope with the fact afterwards (which is usually the easiest thing, you 664calling this function, or cope with the fact afterwards (which is usually
593can just ignore the watchers and/or C<free ()> them for example). 665the easiest thing, you can just ignore the watchers and/or C<free ()> them
666for example).
594 667
595Note that certain global state, such as signal state (and installed signal 668Note that certain global state, such as signal state (and installed signal
596handlers), will not be freed by this function, and related watchers (such 669handlers), will not be freed by this function, and related watchers (such
597as signal and child watchers) would need to be stopped manually. 670as signal and child watchers) would need to be stopped manually.
598 671
599In general it is not advisable to call this function except in the 672This function is normally used on loop objects allocated by
600rare occasion where you really need to free e.g. the signal handling 673C<ev_loop_new>, but it can also be used on the default loop returned by
674C<ev_default_loop>, in which case it is not thread-safe.
675
676Note that it is not advisable to call this function on the default loop
677except in the rare occasion where you really need to free its resources.
601pipe fds. If you need dynamically allocated loops it is better to use 678If you need dynamically allocated loops it is better to use C<ev_loop_new>
602C<ev_loop_new> and C<ev_loop_destroy>. 679and C<ev_loop_destroy>.
603 680
604=item ev_loop_destroy (loop) 681=item ev_loop_fork (loop)
605 682
606Like C<ev_default_destroy>, but destroys an event loop created by an
607earlier call to C<ev_loop_new>.
608
609=item ev_default_fork ()
610
611This function sets a flag that causes subsequent C<ev_run> iterations 683This function sets a flag that causes subsequent C<ev_run> iterations to
612to reinitialise the kernel state for backends that have one. Despite the 684reinitialise the kernel state for backends that have one. Despite the
613name, you can call it anytime, but it makes most sense after forking, in 685name, you can call it anytime, but it makes most sense after forking, in
614the child process (or both child and parent, but that again makes little 686the child process. You I<must> call it (or use C<EVFLAG_FORKCHECK>) in the
615sense). You I<must> call it in the child before using any of the libev 687child before resuming or calling C<ev_run>.
616functions, and it will only take effect at the next C<ev_run> iteration.
617 688
618Again, you I<have> to call it on I<any> loop that you want to re-use after 689Again, you I<have> to call it on I<any> loop that you want to re-use after
619a fork, I<even if you do not plan to use the loop in the parent>. This is 690a fork, I<even if you do not plan to use the loop in the parent>. This is
620because some kernel interfaces *cough* I<kqueue> *cough* do funny things 691because some kernel interfaces *cough* I<kqueue> *cough* do funny things
621during fork. 692during fork.
626call it at all (in fact, C<epoll> is so badly broken that it makes a 697call it at all (in fact, C<epoll> is so badly broken that it makes a
627difference, but libev will usually detect this case on its own and do a 698difference, but libev will usually detect this case on its own and do a
628costly reset of the backend). 699costly reset of the backend).
629 700
630The function itself is quite fast and it's usually not a problem to call 701The function itself is quite fast and it's usually not a problem to call
631it just in case after a fork. To make this easy, the function will fit in 702it just in case after a fork.
632quite nicely into a call to C<pthread_atfork>:
633 703
704Example: Automate calling C<ev_loop_fork> on the default loop when
705using pthreads.
706
707 static void
708 post_fork_child (void)
709 {
710 ev_loop_fork (EV_DEFAULT);
711 }
712
713 ...
634 pthread_atfork (0, 0, ev_default_fork); 714 pthread_atfork (0, 0, post_fork_child);
635
636=item ev_loop_fork (loop)
637
638Like C<ev_default_fork>, but acts on an event loop created by
639C<ev_loop_new>. Yes, you have to call this on every allocated event loop
640after fork that you want to re-use in the child, and how you keep track of
641them is entirely your own problem.
642 715
643=item int ev_is_default_loop (loop) 716=item int ev_is_default_loop (loop)
644 717
645Returns true when the given loop is, in fact, the default loop, and false 718Returns true when the given loop is, in fact, the default loop, and false
646otherwise. 719otherwise.
657prepare and check phases. 730prepare and check phases.
658 731
659=item unsigned int ev_depth (loop) 732=item unsigned int ev_depth (loop)
660 733
661Returns the number of times C<ev_run> was entered minus the number of 734Returns the number of times C<ev_run> was entered minus the number of
662times C<ev_run> was exited, in other words, the recursion depth. 735times C<ev_run> was exited normally, in other words, the recursion depth.
663 736
664Outside C<ev_run>, this number is zero. In a callback, this number is 737Outside C<ev_run>, this number is zero. In a callback, this number is
665C<1>, unless C<ev_run> was invoked recursively (or from another thread), 738C<1>, unless C<ev_run> was invoked recursively (or from another thread),
666in which case it is higher. 739in which case it is higher.
667 740
668Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread 741Leaving C<ev_run> abnormally (setjmp/longjmp, cancelling the thread,
669etc.), doesn't count as "exit" - consider this as a hint to avoid such 742throwing an exception etc.), doesn't count as "exit" - consider this
670ungentleman-like behaviour unless it's really convenient. 743as a hint to avoid such ungentleman-like behaviour unless it's really
744convenient, in which case it is fully supported.
671 745
672=item unsigned int ev_backend (loop) 746=item unsigned int ev_backend (loop)
673 747
674Returns one of the C<EVBACKEND_*> flags indicating the event backend in 748Returns one of the C<EVBACKEND_*> flags indicating the event backend in
675use. 749use.
718without a previous call to C<ev_suspend>. 792without a previous call to C<ev_suspend>.
719 793
720Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the 794Calling C<ev_suspend>/C<ev_resume> has the side effect of updating the
721event loop time (see C<ev_now_update>). 795event loop time (see C<ev_now_update>).
722 796
723=item ev_run (loop, int flags) 797=item bool ev_run (loop, int flags)
724 798
725Finally, this is it, the event handler. This function usually is called 799Finally, this is it, the event handler. This function usually is called
726after you have initialised all your watchers and you want to start 800after you have initialised all your watchers and you want to start
727handling events. It will ask the operating system for any new events, call 801handling events. It will ask the operating system for any new events, call
728the watcher callbacks, an then repeat the whole process indefinitely: This 802the watcher callbacks, and then repeat the whole process indefinitely: This
729is why event loops are called I<loops>. 803is why event loops are called I<loops>.
730 804
731If the flags argument is specified as C<0>, it will keep handling events 805If the flags argument is specified as C<0>, it will keep handling events
732until either no event watchers are active anymore or C<ev_break> was 806until either no event watchers are active anymore or C<ev_break> was
733called. 807called.
808
809The return value is false if there are no more active watchers (which
810usually means "all jobs done" or "deadlock"), and true in all other cases
811(which usually means " you should call C<ev_run> again").
734 812
735Please note that an explicit C<ev_break> is usually better than 813Please note that an explicit C<ev_break> is usually better than
736relying on all watchers to be stopped when deciding when a program has 814relying on all watchers to be stopped when deciding when a program has
737finished (especially in interactive programs), but having a program 815finished (especially in interactive programs), but having a program
738that automatically loops as long as it has to and no longer by virtue 816that automatically loops as long as it has to and no longer by virtue
739of relying on its watchers stopping correctly, that is truly a thing of 817of relying on its watchers stopping correctly, that is truly a thing of
740beauty. 818beauty.
741 819
820This function is I<mostly> exception-safe - you can break out of a
821C<ev_run> call by calling C<longjmp> in a callback, throwing a C++
822exception and so on. This does not decrement the C<ev_depth> value, nor
823will it clear any outstanding C<EVBREAK_ONE> breaks.
824
742A flags value of C<EVRUN_NOWAIT> will look for new events, will handle 825A flags value of C<EVRUN_NOWAIT> will look for new events, will handle
743those events and any already outstanding ones, but will not wait and 826those events and any already outstanding ones, but will not wait and
744block your process in case there are no events and will return after one 827block your process in case there are no events and will return after one
745iteration of the loop. This is sometimes useful to poll and handle new 828iteration of the loop. This is sometimes useful to poll and handle new
746events while doing lengthy calculations, to keep the program responsive. 829events while doing lengthy calculations, to keep the program responsive.
755This is useful if you are waiting for some external event in conjunction 838This is useful if you are waiting for some external event in conjunction
756with something not expressible using other libev watchers (i.e. "roll your 839with something not expressible using other libev watchers (i.e. "roll your
757own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is 840own C<ev_run>"). However, a pair of C<ev_prepare>/C<ev_check> watchers is
758usually a better approach for this kind of thing. 841usually a better approach for this kind of thing.
759 842
760Here are the gory details of what C<ev_run> does: 843Here are the gory details of what C<ev_run> does (this is for your
844understanding, not a guarantee that things will work exactly like this in
845future versions):
761 846
762 - Increment loop depth. 847 - Increment loop depth.
763 - Reset the ev_break status. 848 - Reset the ev_break status.
764 - Before the first iteration, call any pending watchers. 849 - Before the first iteration, call any pending watchers.
765 LOOP: 850 LOOP:
798anymore. 883anymore.
799 884
800 ... queue jobs here, make sure they register event watchers as long 885 ... queue jobs here, make sure they register event watchers as long
801 ... as they still have work to do (even an idle watcher will do..) 886 ... as they still have work to do (even an idle watcher will do..)
802 ev_run (my_loop, 0); 887 ev_run (my_loop, 0);
803 ... jobs done or somebody called unloop. yeah! 888 ... jobs done or somebody called break. yeah!
804 889
805=item ev_break (loop, how) 890=item ev_break (loop, how)
806 891
807Can be used to make a call to C<ev_run> return early (but only after it 892Can be used to make a call to C<ev_run> return early (but only after it
808has processed all outstanding events). The C<how> argument must be either 893has processed all outstanding events). The C<how> argument must be either
809C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or 894C<EVBREAK_ONE>, which will make the innermost C<ev_run> call return, or
810C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return. 895C<EVBREAK_ALL>, which will make all nested C<ev_run> calls return.
811 896
812This "unloop state" will be cleared when entering C<ev_run> again. 897This "break state" will be cleared on the next call to C<ev_run>.
813 898
814It is safe to call C<ev_break> from outside any C<ev_run> calls. ##TODO## 899It is safe to call C<ev_break> from outside any C<ev_run> calls, too, in
900which case it will have no effect.
815 901
816=item ev_ref (loop) 902=item ev_ref (loop)
817 903
818=item ev_unref (loop) 904=item ev_unref (loop)
819 905
840running when nothing else is active. 926running when nothing else is active.
841 927
842 ev_signal exitsig; 928 ev_signal exitsig;
843 ev_signal_init (&exitsig, sig_cb, SIGINT); 929 ev_signal_init (&exitsig, sig_cb, SIGINT);
844 ev_signal_start (loop, &exitsig); 930 ev_signal_start (loop, &exitsig);
845 evf_unref (loop); 931 ev_unref (loop);
846 932
847Example: For some weird reason, unregister the above signal handler again. 933Example: For some weird reason, unregister the above signal handler again.
848 934
849 ev_ref (loop); 935 ev_ref (loop);
850 ev_signal_stop (loop, &exitsig); 936 ev_signal_stop (loop, &exitsig);
870overhead for the actual polling but can deliver many events at once. 956overhead for the actual polling but can deliver many events at once.
871 957
872By setting a higher I<io collect interval> you allow libev to spend more 958By setting a higher I<io collect interval> you allow libev to spend more
873time collecting I/O events, so you can handle more events per iteration, 959time collecting I/O events, so you can handle more events per iteration,
874at the cost of increasing latency. Timeouts (both C<ev_periodic> and 960at the cost of increasing latency. Timeouts (both C<ev_periodic> and
875C<ev_timer>) will be not affected. Setting this to a non-null value will 961C<ev_timer>) will not be affected. Setting this to a non-null value will
876introduce an additional C<ev_sleep ()> call into most loop iterations. The 962introduce an additional C<ev_sleep ()> call into most loop iterations. The
877sleep time ensures that libev will not poll for I/O events more often then 963sleep time ensures that libev will not poll for I/O events more often then
878once per this interval, on average. 964once per this interval, on average (as long as the host time resolution is
965good enough).
879 966
880Likewise, by setting a higher I<timeout collect interval> you allow libev 967Likewise, by setting a higher I<timeout collect interval> you allow libev
881to spend more time collecting timeouts, at the expense of increased 968to spend more time collecting timeouts, at the expense of increased
882latency/jitter/inexactness (the watcher callback will be called 969latency/jitter/inexactness (the watcher callback will be called
883later). C<ev_io> watchers will not be affected. Setting this to a non-null 970later). C<ev_io> watchers will not be affected. Setting this to a non-null
937can be done relatively simply by putting mutex_lock/unlock calls around 1024can be done relatively simply by putting mutex_lock/unlock calls around
938each call to a libev function. 1025each call to a libev function.
939 1026
940However, C<ev_run> can run an indefinite time, so it is not feasible 1027However, C<ev_run> can run an indefinite time, so it is not feasible
941to wait for it to return. One way around this is to wake up the event 1028to wait for it to return. One way around this is to wake up the event
942loop via C<ev_break> and C<av_async_send>, another way is to set these 1029loop via C<ev_break> and C<ev_async_send>, another way is to set these
943I<release> and I<acquire> callbacks on the loop. 1030I<release> and I<acquire> callbacks on the loop.
944 1031
945When set, then C<release> will be called just before the thread is 1032When set, then C<release> will be called just before the thread is
946suspended waiting for new events, and C<acquire> is called just 1033suspended waiting for new events, and C<acquire> is called just
947afterwards. 1034afterwards.
962See also the locking example in the C<THREADS> section later in this 1049See also the locking example in the C<THREADS> section later in this
963document. 1050document.
964 1051
965=item ev_set_userdata (loop, void *data) 1052=item ev_set_userdata (loop, void *data)
966 1053
967=item ev_userdata (loop) 1054=item void *ev_userdata (loop)
968 1055
969Set and retrieve a single C<void *> associated with a loop. When 1056Set and retrieve a single C<void *> associated with a loop. When
970C<ev_set_userdata> has never been called, then C<ev_userdata> returns 1057C<ev_set_userdata> has never been called, then C<ev_userdata> returns
971C<0.> 1058C<0>.
972 1059
973These two functions can be used to associate arbitrary data with a loop, 1060These two functions can be used to associate arbitrary data with a loop,
974and are intended solely for the C<invoke_pending_cb>, C<release> and 1061and are intended solely for the C<invoke_pending_cb>, C<release> and
975C<acquire> callbacks described above, but of course can be (ab-)used for 1062C<acquire> callbacks described above, but of course can be (ab-)used for
976any other purpose as well. 1063any other purpose as well.
1104=item C<EV_FORK> 1191=item C<EV_FORK>
1105 1192
1106The event loop has been resumed in the child process after fork (see 1193The event loop has been resumed in the child process after fork (see
1107C<ev_fork>). 1194C<ev_fork>).
1108 1195
1196=item C<EV_CLEANUP>
1197
1198The event loop is about to be destroyed (see C<ev_cleanup>).
1199
1109=item C<EV_ASYNC> 1200=item C<EV_ASYNC>
1110 1201
1111The given async watcher has been asynchronously notified (see C<ev_async>). 1202The given async watcher has been asynchronously notified (see C<ev_async>).
1112 1203
1113=item C<EV_CUSTOM> 1204=item C<EV_CUSTOM>
1134programs, though, as the fd could already be closed and reused for another 1225programs, though, as the fd could already be closed and reused for another
1135thing, so beware. 1226thing, so beware.
1136 1227
1137=back 1228=back
1138 1229
1230=head2 GENERIC WATCHER FUNCTIONS
1231
1232=over 4
1233
1234=item C<ev_init> (ev_TYPE *watcher, callback)
1235
1236This macro initialises the generic portion of a watcher. The contents
1237of the watcher object can be arbitrary (so C<malloc> will do). Only
1238the generic parts of the watcher are initialised, you I<need> to call
1239the type-specific C<ev_TYPE_set> macro afterwards to initialise the
1240type-specific parts. For each type there is also a C<ev_TYPE_init> macro
1241which rolls both calls into one.
1242
1243You can reinitialise a watcher at any time as long as it has been stopped
1244(or never started) and there are no pending events outstanding.
1245
1246The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
1247int revents)>.
1248
1249Example: Initialise an C<ev_io> watcher in two steps.
1250
1251 ev_io w;
1252 ev_init (&w, my_cb);
1253 ev_io_set (&w, STDIN_FILENO, EV_READ);
1254
1255=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
1256
1257This macro initialises the type-specific parts of a watcher. You need to
1258call C<ev_init> at least once before you call this macro, but you can
1259call C<ev_TYPE_set> any number of times. You must not, however, call this
1260macro on a watcher that is active (it can be pending, however, which is a
1261difference to the C<ev_init> macro).
1262
1263Although some watcher types do not have type-specific arguments
1264(e.g. C<ev_prepare>) you still need to call its C<set> macro.
1265
1266See C<ev_init>, above, for an example.
1267
1268=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
1269
1270This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
1271calls into a single call. This is the most convenient method to initialise
1272a watcher. The same limitations apply, of course.
1273
1274Example: Initialise and set an C<ev_io> watcher in one step.
1275
1276 ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
1277
1278=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
1279
1280Starts (activates) the given watcher. Only active watchers will receive
1281events. If the watcher is already active nothing will happen.
1282
1283Example: Start the C<ev_io> watcher that is being abused as example in this
1284whole section.
1285
1286 ev_io_start (EV_DEFAULT_UC, &w);
1287
1288=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
1289
1290Stops the given watcher if active, and clears the pending status (whether
1291the watcher was active or not).
1292
1293It is possible that stopped watchers are pending - for example,
1294non-repeating timers are being stopped when they become pending - but
1295calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
1296pending. If you want to free or reuse the memory used by the watcher it is
1297therefore a good idea to always call its C<ev_TYPE_stop> function.
1298
1299=item bool ev_is_active (ev_TYPE *watcher)
1300
1301Returns a true value iff the watcher is active (i.e. it has been started
1302and not yet been stopped). As long as a watcher is active you must not modify
1303it.
1304
1305=item bool ev_is_pending (ev_TYPE *watcher)
1306
1307Returns a true value iff the watcher is pending, (i.e. it has outstanding
1308events but its callback has not yet been invoked). As long as a watcher
1309is pending (but not active) you must not call an init function on it (but
1310C<ev_TYPE_set> is safe), you must not change its priority, and you must
1311make sure the watcher is available to libev (e.g. you cannot C<free ()>
1312it).
1313
1314=item callback ev_cb (ev_TYPE *watcher)
1315
1316Returns the callback currently set on the watcher.
1317
1318=item ev_cb_set (ev_TYPE *watcher, callback)
1319
1320Change the callback. You can change the callback at virtually any time
1321(modulo threads).
1322
1323=item ev_set_priority (ev_TYPE *watcher, int priority)
1324
1325=item int ev_priority (ev_TYPE *watcher)
1326
1327Set and query the priority of the watcher. The priority is a small
1328integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
1329(default: C<-2>). Pending watchers with higher priority will be invoked
1330before watchers with lower priority, but priority will not keep watchers
1331from being executed (except for C<ev_idle> watchers).
1332
1333If you need to suppress invocation when higher priority events are pending
1334you need to look at C<ev_idle> watchers, which provide this functionality.
1335
1336You I<must not> change the priority of a watcher as long as it is active or
1337pending.
1338
1339Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
1340fine, as long as you do not mind that the priority value you query might
1341or might not have been clamped to the valid range.
1342
1343The default priority used by watchers when no priority has been set is
1344always C<0>, which is supposed to not be too high and not be too low :).
1345
1346See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
1347priorities.
1348
1349=item ev_invoke (loop, ev_TYPE *watcher, int revents)
1350
1351Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
1352C<loop> nor C<revents> need to be valid as long as the watcher callback
1353can deal with that fact, as both are simply passed through to the
1354callback.
1355
1356=item int ev_clear_pending (loop, ev_TYPE *watcher)
1357
1358If the watcher is pending, this function clears its pending status and
1359returns its C<revents> bitset (as if its callback was invoked). If the
1360watcher isn't pending it does nothing and returns C<0>.
1361
1362Sometimes it can be useful to "poll" a watcher instead of waiting for its
1363callback to be invoked, which can be accomplished with this function.
1364
1365=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
1366
1367Feeds the given event set into the event loop, as if the specified event
1368had happened for the specified watcher (which must be a pointer to an
1369initialised but not necessarily started event watcher). Obviously you must
1370not free the watcher as long as it has pending events.
1371
1372Stopping the watcher, letting libev invoke it, or calling
1373C<ev_clear_pending> will clear the pending event, even if the watcher was
1374not started in the first place.
1375
1376See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
1377functions that do not need a watcher.
1378
1379=back
1380
1381See also the L<ASSOCIATING CUSTOM DATA WITH A WATCHER> and L<BUILDING YOUR
1382OWN COMPOSITE WATCHERS> idioms.
1383
1139=head2 WATCHER STATES 1384=head2 WATCHER STATES
1140 1385
1141There are various watcher states mentioned throughout this manual - 1386There are various watcher states mentioned throughout this manual -
1142active, pending and so on. In this section these states and the rules to 1387active, pending and so on. In this section these states and the rules to
1143transition between them will be described in more detail - and while these 1388transition between them will be described in more detail - and while these
1145 1390
1146=over 4 1391=over 4
1147 1392
1148=item initialiased 1393=item initialiased
1149 1394
1150Before a watcher can be registered with the event looop it has to be 1395Before a watcher can be registered with the event loop it has to be
1151initialised. This can be done with a call to C<ev_TYPE_init>, or calls to 1396initialised. This can be done with a call to C<ev_TYPE_init>, or calls to
1152C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function. 1397C<ev_init> followed by the watcher-specific C<ev_TYPE_set> function.
1153 1398
1154In this state it is simply some block of memory that is suitable for use 1399In this state it is simply some block of memory that is suitable for
1155in an event loop. It can be moved around, freed, reused etc. at will. 1400use in an event loop. It can be moved around, freed, reused etc. at
1401will - as long as you either keep the memory contents intact, or call
1402C<ev_TYPE_init> again.
1156 1403
1157=item started/running/active 1404=item started/running/active
1158 1405
1159Once a watcher has been started with a call to C<ev_TYPE_start> it becomes 1406Once a watcher has been started with a call to C<ev_TYPE_start> it becomes
1160property of the event loop, and is actively waiting for events. While in 1407property of the event loop, and is actively waiting for events. While in
1188latter will clear any pending state the watcher might be in, regardless 1435latter will clear any pending state the watcher might be in, regardless
1189of whether it was active or not, so stopping a watcher explicitly before 1436of whether it was active or not, so stopping a watcher explicitly before
1190freeing it is often a good idea. 1437freeing it is often a good idea.
1191 1438
1192While stopped (and not pending) the watcher is essentially in the 1439While stopped (and not pending) the watcher is essentially in the
1193initialised state, that is it can be reused, moved, modified in any way 1440initialised state, that is, it can be reused, moved, modified in any way
1194you wish. 1441you wish (but when you trash the memory block, you need to C<ev_TYPE_init>
1442it again).
1195 1443
1196=back 1444=back
1197
1198=head2 GENERIC WATCHER FUNCTIONS
1199
1200=over 4
1201
1202=item C<ev_init> (ev_TYPE *watcher, callback)
1203
1204This macro initialises the generic portion of a watcher. The contents
1205of the watcher object can be arbitrary (so C<malloc> will do). Only
1206the generic parts of the watcher are initialised, you I<need> to call
1207the type-specific C<ev_TYPE_set> macro afterwards to initialise the
1208type-specific parts. For each type there is also a C<ev_TYPE_init> macro
1209which rolls both calls into one.
1210
1211You can reinitialise a watcher at any time as long as it has been stopped
1212(or never started) and there are no pending events outstanding.
1213
1214The callback is always of type C<void (*)(struct ev_loop *loop, ev_TYPE *watcher,
1215int revents)>.
1216
1217Example: Initialise an C<ev_io> watcher in two steps.
1218
1219 ev_io w;
1220 ev_init (&w, my_cb);
1221 ev_io_set (&w, STDIN_FILENO, EV_READ);
1222
1223=item C<ev_TYPE_set> (ev_TYPE *watcher, [args])
1224
1225This macro initialises the type-specific parts of a watcher. You need to
1226call C<ev_init> at least once before you call this macro, but you can
1227call C<ev_TYPE_set> any number of times. You must not, however, call this
1228macro on a watcher that is active (it can be pending, however, which is a
1229difference to the C<ev_init> macro).
1230
1231Although some watcher types do not have type-specific arguments
1232(e.g. C<ev_prepare>) you still need to call its C<set> macro.
1233
1234See C<ev_init>, above, for an example.
1235
1236=item C<ev_TYPE_init> (ev_TYPE *watcher, callback, [args])
1237
1238This convenience macro rolls both C<ev_init> and C<ev_TYPE_set> macro
1239calls into a single call. This is the most convenient method to initialise
1240a watcher. The same limitations apply, of course.
1241
1242Example: Initialise and set an C<ev_io> watcher in one step.
1243
1244 ev_io_init (&w, my_cb, STDIN_FILENO, EV_READ);
1245
1246=item C<ev_TYPE_start> (loop, ev_TYPE *watcher)
1247
1248Starts (activates) the given watcher. Only active watchers will receive
1249events. If the watcher is already active nothing will happen.
1250
1251Example: Start the C<ev_io> watcher that is being abused as example in this
1252whole section.
1253
1254 ev_io_start (EV_DEFAULT_UC, &w);
1255
1256=item C<ev_TYPE_stop> (loop, ev_TYPE *watcher)
1257
1258Stops the given watcher if active, and clears the pending status (whether
1259the watcher was active or not).
1260
1261It is possible that stopped watchers are pending - for example,
1262non-repeating timers are being stopped when they become pending - but
1263calling C<ev_TYPE_stop> ensures that the watcher is neither active nor
1264pending. If you want to free or reuse the memory used by the watcher it is
1265therefore a good idea to always call its C<ev_TYPE_stop> function.
1266
1267=item bool ev_is_active (ev_TYPE *watcher)
1268
1269Returns a true value iff the watcher is active (i.e. it has been started
1270and not yet been stopped). As long as a watcher is active you must not modify
1271it.
1272
1273=item bool ev_is_pending (ev_TYPE *watcher)
1274
1275Returns a true value iff the watcher is pending, (i.e. it has outstanding
1276events but its callback has not yet been invoked). As long as a watcher
1277is pending (but not active) you must not call an init function on it (but
1278C<ev_TYPE_set> is safe), you must not change its priority, and you must
1279make sure the watcher is available to libev (e.g. you cannot C<free ()>
1280it).
1281
1282=item callback ev_cb (ev_TYPE *watcher)
1283
1284Returns the callback currently set on the watcher.
1285
1286=item ev_cb_set (ev_TYPE *watcher, callback)
1287
1288Change the callback. You can change the callback at virtually any time
1289(modulo threads).
1290
1291=item ev_set_priority (ev_TYPE *watcher, int priority)
1292
1293=item int ev_priority (ev_TYPE *watcher)
1294
1295Set and query the priority of the watcher. The priority is a small
1296integer between C<EV_MAXPRI> (default: C<2>) and C<EV_MINPRI>
1297(default: C<-2>). Pending watchers with higher priority will be invoked
1298before watchers with lower priority, but priority will not keep watchers
1299from being executed (except for C<ev_idle> watchers).
1300
1301If you need to suppress invocation when higher priority events are pending
1302you need to look at C<ev_idle> watchers, which provide this functionality.
1303
1304You I<must not> change the priority of a watcher as long as it is active or
1305pending.
1306
1307Setting a priority outside the range of C<EV_MINPRI> to C<EV_MAXPRI> is
1308fine, as long as you do not mind that the priority value you query might
1309or might not have been clamped to the valid range.
1310
1311The default priority used by watchers when no priority has been set is
1312always C<0>, which is supposed to not be too high and not be too low :).
1313
1314See L<WATCHER PRIORITY MODELS>, below, for a more thorough treatment of
1315priorities.
1316
1317=item ev_invoke (loop, ev_TYPE *watcher, int revents)
1318
1319Invoke the C<watcher> with the given C<loop> and C<revents>. Neither
1320C<loop> nor C<revents> need to be valid as long as the watcher callback
1321can deal with that fact, as both are simply passed through to the
1322callback.
1323
1324=item int ev_clear_pending (loop, ev_TYPE *watcher)
1325
1326If the watcher is pending, this function clears its pending status and
1327returns its C<revents> bitset (as if its callback was invoked). If the
1328watcher isn't pending it does nothing and returns C<0>.
1329
1330Sometimes it can be useful to "poll" a watcher instead of waiting for its
1331callback to be invoked, which can be accomplished with this function.
1332
1333=item ev_feed_event (loop, ev_TYPE *watcher, int revents)
1334
1335Feeds the given event set into the event loop, as if the specified event
1336had happened for the specified watcher (which must be a pointer to an
1337initialised but not necessarily started event watcher). Obviously you must
1338not free the watcher as long as it has pending events.
1339
1340Stopping the watcher, letting libev invoke it, or calling
1341C<ev_clear_pending> will clear the pending event, even if the watcher was
1342not started in the first place.
1343
1344See also C<ev_feed_fd_event> and C<ev_feed_signal_event> for related
1345functions that do not need a watcher.
1346
1347=back
1348
1349
1350=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
1351
1352Each watcher has, by default, a member C<void *data> that you can change
1353and read at any time: libev will completely ignore it. This can be used
1354to associate arbitrary data with your watcher. If you need more data and
1355don't want to allocate memory and store a pointer to it in that data
1356member, you can also "subclass" the watcher type and provide your own
1357data:
1358
1359 struct my_io
1360 {
1361 ev_io io;
1362 int otherfd;
1363 void *somedata;
1364 struct whatever *mostinteresting;
1365 };
1366
1367 ...
1368 struct my_io w;
1369 ev_io_init (&w.io, my_cb, fd, EV_READ);
1370
1371And since your callback will be called with a pointer to the watcher, you
1372can cast it back to your own type:
1373
1374 static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
1375 {
1376 struct my_io *w = (struct my_io *)w_;
1377 ...
1378 }
1379
1380More interesting and less C-conformant ways of casting your callback type
1381instead have been omitted.
1382
1383Another common scenario is to use some data structure with multiple
1384embedded watchers:
1385
1386 struct my_biggy
1387 {
1388 int some_data;
1389 ev_timer t1;
1390 ev_timer t2;
1391 }
1392
1393In this case getting the pointer to C<my_biggy> is a bit more
1394complicated: Either you store the address of your C<my_biggy> struct
1395in the C<data> member of the watcher (for woozies), or you need to use
1396some pointer arithmetic using C<offsetof> inside your watchers (for real
1397programmers):
1398
1399 #include <stddef.h>
1400
1401 static void
1402 t1_cb (EV_P_ ev_timer *w, int revents)
1403 {
1404 struct my_biggy big = (struct my_biggy *)
1405 (((char *)w) - offsetof (struct my_biggy, t1));
1406 }
1407
1408 static void
1409 t2_cb (EV_P_ ev_timer *w, int revents)
1410 {
1411 struct my_biggy big = (struct my_biggy *)
1412 (((char *)w) - offsetof (struct my_biggy, t2));
1413 }
1414 1445
1415=head2 WATCHER PRIORITY MODELS 1446=head2 WATCHER PRIORITY MODELS
1416 1447
1417Many event loops support I<watcher priorities>, which are usually small 1448Many event loops support I<watcher priorities>, which are usually small
1418integers that influence the ordering of event callback invocation 1449integers that influence the ordering of event callback invocation
1545In general you can register as many read and/or write event watchers per 1576In general you can register as many read and/or write event watchers per
1546fd as you want (as long as you don't confuse yourself). Setting all file 1577fd as you want (as long as you don't confuse yourself). Setting all file
1547descriptors to non-blocking mode is also usually a good idea (but not 1578descriptors to non-blocking mode is also usually a good idea (but not
1548required if you know what you are doing). 1579required if you know what you are doing).
1549 1580
1550If you cannot use non-blocking mode, then force the use of a
1551known-to-be-good backend (at the time of this writing, this includes only
1552C<EVBACKEND_SELECT> and C<EVBACKEND_POLL>). The same applies to file
1553descriptors for which non-blocking operation makes no sense (such as
1554files) - libev doesn't guarantee any specific behaviour in that case.
1555
1556Another thing you have to watch out for is that it is quite easy to 1581Another thing you have to watch out for is that it is quite easy to
1557receive "spurious" readiness notifications, that is your callback might 1582receive "spurious" readiness notifications, that is, your callback might
1558be called with C<EV_READ> but a subsequent C<read>(2) will actually block 1583be called with C<EV_READ> but a subsequent C<read>(2) will actually block
1559because there is no data. Not only are some backends known to create a 1584because there is no data. It is very easy to get into this situation even
1560lot of those (for example Solaris ports), it is very easy to get into 1585with a relatively standard program structure. Thus it is best to always
1561this situation even with a relatively standard program structure. Thus 1586use non-blocking I/O: An extra C<read>(2) returning C<EAGAIN> is far
1562it is best to always use non-blocking I/O: An extra C<read>(2) returning
1563C<EAGAIN> is far preferable to a program hanging until some data arrives. 1587preferable to a program hanging until some data arrives.
1564 1588
1565If you cannot run the fd in non-blocking mode (for example you should 1589If you cannot run the fd in non-blocking mode (for example you should
1566not play around with an Xlib connection), then you have to separately 1590not play around with an Xlib connection), then you have to separately
1567re-test whether a file descriptor is really ready with a known-to-be good 1591re-test whether a file descriptor is really ready with a known-to-be good
1568interface such as poll (fortunately in our Xlib example, Xlib already 1592interface such as poll (fortunately in the case of Xlib, it already does
1569does this on its own, so its quite safe to use). Some people additionally 1593this on its own, so its quite safe to use). Some people additionally
1570use C<SIGALRM> and an interval timer, just to be sure you won't block 1594use C<SIGALRM> and an interval timer, just to be sure you won't block
1571indefinitely. 1595indefinitely.
1572 1596
1573But really, best use non-blocking mode. 1597But really, best use non-blocking mode.
1574 1598
1602 1626
1603There is no workaround possible except not registering events 1627There is no workaround possible except not registering events
1604for potentially C<dup ()>'ed file descriptors, or to resort to 1628for potentially C<dup ()>'ed file descriptors, or to resort to
1605C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>. 1629C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1606 1630
1631=head3 The special problem of files
1632
1633Many people try to use C<select> (or libev) on file descriptors
1634representing files, and expect it to become ready when their program
1635doesn't block on disk accesses (which can take a long time on their own).
1636
1637However, this cannot ever work in the "expected" way - you get a readiness
1638notification as soon as the kernel knows whether and how much data is
1639there, and in the case of open files, that's always the case, so you
1640always get a readiness notification instantly, and your read (or possibly
1641write) will still block on the disk I/O.
1642
1643Another way to view it is that in the case of sockets, pipes, character
1644devices and so on, there is another party (the sender) that delivers data
1645on its own, but in the case of files, there is no such thing: the disk
1646will not send data on its own, simply because it doesn't know what you
1647wish to read - you would first have to request some data.
1648
1649Since files are typically not-so-well supported by advanced notification
1650mechanism, libev tries hard to emulate POSIX behaviour with respect
1651to files, even though you should not use it. The reason for this is
1652convenience: sometimes you want to watch STDIN or STDOUT, which is
1653usually a tty, often a pipe, but also sometimes files or special devices
1654(for example, C<epoll> on Linux works with F</dev/random> but not with
1655F</dev/urandom>), and even though the file might better be served with
1656asynchronous I/O instead of with non-blocking I/O, it is still useful when
1657it "just works" instead of freezing.
1658
1659So avoid file descriptors pointing to files when you know it (e.g. use
1660libeio), but use them when it is convenient, e.g. for STDIN/STDOUT, or
1661when you rarely read from a file instead of from a socket, and want to
1662reuse the same code path.
1663
1607=head3 The special problem of fork 1664=head3 The special problem of fork
1608 1665
1609Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit 1666Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
1610useless behaviour. Libev fully supports fork, but needs to be told about 1667useless behaviour. Libev fully supports fork, but needs to be told about
1611it in the child. 1668it in the child if you want to continue to use it in the child.
1612 1669
1613To support fork in your programs, you either have to call 1670To support fork in your child processes, you have to call C<ev_loop_fork
1614C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child, 1671()> after a fork in the child, enable C<EVFLAG_FORKCHECK>, or resort to
1615enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or 1672C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1616C<EVBACKEND_POLL>.
1617 1673
1618=head3 The special problem of SIGPIPE 1674=head3 The special problem of SIGPIPE
1619 1675
1620While not really specific to libev, it is easy to forget about C<SIGPIPE>: 1676While not really specific to libev, it is easy to forget about C<SIGPIPE>:
1621when writing to a pipe whose other end has been closed, your program gets 1677when writing to a pipe whose other end has been closed, your program gets
1719detecting time jumps is hard, and some inaccuracies are unavoidable (the 1775detecting time jumps is hard, and some inaccuracies are unavoidable (the
1720monotonic clock option helps a lot here). 1776monotonic clock option helps a lot here).
1721 1777
1722The callback is guaranteed to be invoked only I<after> its timeout has 1778The callback is guaranteed to be invoked only I<after> its timeout has
1723passed (not I<at>, so on systems with very low-resolution clocks this 1779passed (not I<at>, so on systems with very low-resolution clocks this
1724might introduce a small delay). If multiple timers become ready during the 1780might introduce a small delay, see "the special problem of being too
1781early", below). If multiple timers become ready during the same loop
1725same loop iteration then the ones with earlier time-out values are invoked 1782iteration then the ones with earlier time-out values are invoked before
1726before ones of the same priority with later time-out values (but this is 1783ones of the same priority with later time-out values (but this is no
1727no longer true when a callback calls C<ev_run> recursively). 1784longer true when a callback calls C<ev_run> recursively).
1728 1785
1729=head3 Be smart about timeouts 1786=head3 Be smart about timeouts
1730 1787
1731Many real-world problems involve some kind of timeout, usually for error 1788Many real-world problems involve some kind of timeout, usually for error
1732recovery. A typical example is an HTTP request - if the other side hangs, 1789recovery. A typical example is an HTTP request - if the other side hangs,
1807 1864
1808In this case, it would be more efficient to leave the C<ev_timer> alone, 1865In this case, it would be more efficient to leave the C<ev_timer> alone,
1809but remember the time of last activity, and check for a real timeout only 1866but remember the time of last activity, and check for a real timeout only
1810within the callback: 1867within the callback:
1811 1868
1869 ev_tstamp timeout = 60.;
1812 ev_tstamp last_activity; // time of last activity 1870 ev_tstamp last_activity; // time of last activity
1871 ev_timer timer;
1813 1872
1814 static void 1873 static void
1815 callback (EV_P_ ev_timer *w, int revents) 1874 callback (EV_P_ ev_timer *w, int revents)
1816 { 1875 {
1817 ev_tstamp now = ev_now (EV_A); 1876 // calculate when the timeout would happen
1818 ev_tstamp timeout = last_activity + 60.; 1877 ev_tstamp after = last_activity - ev_now (EV_A) + timeout;
1819 1878
1820 // if last_activity + 60. is older than now, we did time out 1879 // if negative, it means we the timeout already occured
1821 if (timeout < now) 1880 if (after < 0.)
1822 { 1881 {
1823 // timeout occurred, take action 1882 // timeout occurred, take action
1824 } 1883 }
1825 else 1884 else
1826 { 1885 {
1827 // callback was invoked, but there was some activity, re-arm 1886 // callback was invoked, but there was some recent
1828 // the watcher to fire in last_activity + 60, which is 1887 // activity. simply restart the timer to time out
1829 // guaranteed to be in the future, so "again" is positive: 1888 // after "after" seconds, which is the earliest time
1830 w->repeat = timeout - now; 1889 // the timeout can occur.
1890 ev_timer_set (w, after, 0.);
1831 ev_timer_again (EV_A_ w); 1891 ev_timer_start (EV_A_ w);
1832 } 1892 }
1833 } 1893 }
1834 1894
1835To summarise the callback: first calculate the real timeout (defined 1895To summarise the callback: first calculate in how many seconds the
1836as "60 seconds after the last activity"), then check if that time has 1896timeout will occur (by calculating the absolute time when it would occur,
1837been reached, which means something I<did>, in fact, time out. Otherwise 1897C<last_activity + timeout>, and subtracting the current time, C<ev_now
1838the callback was invoked too early (C<timeout> is in the future), so 1898(EV_A)> from that).
1839re-schedule the timer to fire at that future time, to see if maybe we have
1840a timeout then.
1841 1899
1842Note how C<ev_timer_again> is used, taking advantage of the 1900If this value is negative, then we are already past the timeout, i.e. we
1843C<ev_timer_again> optimisation when the timer is already running. 1901timed out, and need to do whatever is needed in this case.
1902
1903Otherwise, we now the earliest time at which the timeout would trigger,
1904and simply start the timer with this timeout value.
1905
1906In other words, each time the callback is invoked it will check whether
1907the timeout cocured. If not, it will simply reschedule itself to check
1908again at the earliest time it could time out. Rinse. Repeat.
1844 1909
1845This scheme causes more callback invocations (about one every 60 seconds 1910This scheme causes more callback invocations (about one every 60 seconds
1846minus half the average time between activity), but virtually no calls to 1911minus half the average time between activity), but virtually no calls to
1847libev to change the timeout. 1912libev to change the timeout.
1848 1913
1849To start the timer, simply initialise the watcher and set C<last_activity> 1914To start the machinery, simply initialise the watcher and set
1850to the current time (meaning we just have some activity :), then call the 1915C<last_activity> to the current time (meaning there was some activity just
1851callback, which will "do the right thing" and start the timer: 1916now), then call the callback, which will "do the right thing" and start
1917the timer:
1852 1918
1919 last_activity = ev_now (EV_A);
1853 ev_init (timer, callback); 1920 ev_init (&timer, callback);
1854 last_activity = ev_now (loop); 1921 callback (EV_A_ &timer, 0);
1855 callback (loop, timer, EV_TIMER);
1856 1922
1857And when there is some activity, simply store the current time in 1923When there is some activity, simply store the current time in
1858C<last_activity>, no libev calls at all: 1924C<last_activity>, no libev calls at all:
1859 1925
1926 if (activity detected)
1860 last_activity = ev_now (loop); 1927 last_activity = ev_now (EV_A);
1928
1929When your timeout value changes, then the timeout can be changed by simply
1930providing a new value, stopping the timer and calling the callback, which
1931will agaion do the right thing (for example, time out immediately :).
1932
1933 timeout = new_value;
1934 ev_timer_stop (EV_A_ &timer);
1935 callback (EV_A_ &timer, 0);
1861 1936
1862This technique is slightly more complex, but in most cases where the 1937This technique is slightly more complex, but in most cases where the
1863time-out is unlikely to be triggered, much more efficient. 1938time-out is unlikely to be triggered, much more efficient.
1864
1865Changing the timeout is trivial as well (if it isn't hard-coded in the
1866callback :) - just change the timeout and invoke the callback, which will
1867fix things for you.
1868 1939
1869=item 4. Wee, just use a double-linked list for your timeouts. 1940=item 4. Wee, just use a double-linked list for your timeouts.
1870 1941
1871If there is not one request, but many thousands (millions...), all 1942If there is not one request, but many thousands (millions...), all
1872employing some kind of timeout with the same timeout value, then one can 1943employing some kind of timeout with the same timeout value, then one can
1899Method #1 is almost always a bad idea, and buys you nothing. Method #4 is 1970Method #1 is almost always a bad idea, and buys you nothing. Method #4 is
1900rather complicated, but extremely efficient, something that really pays 1971rather complicated, but extremely efficient, something that really pays
1901off after the first million or so of active timers, i.e. it's usually 1972off after the first million or so of active timers, i.e. it's usually
1902overkill :) 1973overkill :)
1903 1974
1975=head3 The special problem of being too early
1976
1977If you ask a timer to call your callback after three seconds, then
1978you expect it to be invoked after three seconds - but of course, this
1979cannot be guaranteed to infinite precision. Less obviously, it cannot be
1980guaranteed to any precision by libev - imagine somebody suspending the
1981process with a STOP signal for a few hours for example.
1982
1983So, libev tries to invoke your callback as soon as possible I<after> the
1984delay has occurred, but cannot guarantee this.
1985
1986A less obvious failure mode is calling your callback too early: many event
1987loops compare timestamps with a "elapsed delay >= requested delay", but
1988this can cause your callback to be invoked much earlier than you would
1989expect.
1990
1991To see why, imagine a system with a clock that only offers full second
1992resolution (think windows if you can't come up with a broken enough OS
1993yourself). If you schedule a one-second timer at the time 500.9, then the
1994event loop will schedule your timeout to elapse at a system time of 500
1995(500.9 truncated to the resolution) + 1, or 501.
1996
1997If an event library looks at the timeout 0.1s later, it will see "501 >=
1998501" and invoke the callback 0.1s after it was started, even though a
1999one-second delay was requested - this is being "too early", despite best
2000intentions.
2001
2002This is the reason why libev will never invoke the callback if the elapsed
2003delay equals the requested delay, but only when the elapsed delay is
2004larger than the requested delay. In the example above, libev would only invoke
2005the callback at system time 502, or 1.1s after the timer was started.
2006
2007So, while libev cannot guarantee that your callback will be invoked
2008exactly when requested, it I<can> and I<does> guarantee that the requested
2009delay has actually elapsed, or in other words, it always errs on the "too
2010late" side of things.
2011
1904=head3 The special problem of time updates 2012=head3 The special problem of time updates
1905 2013
1906Establishing the current time is a costly operation (it usually takes at 2014Establishing the current time is a costly operation (it usually takes
1907least two system calls): EV therefore updates its idea of the current 2015at least one system call): EV therefore updates its idea of the current
1908time only before and after C<ev_run> collects new events, which causes a 2016time only before and after C<ev_run> collects new events, which causes a
1909growing difference between C<ev_now ()> and C<ev_time ()> when handling 2017growing difference between C<ev_now ()> and C<ev_time ()> when handling
1910lots of events in one iteration. 2018lots of events in one iteration.
1911 2019
1912The relative timeouts are calculated relative to the C<ev_now ()> 2020The relative timeouts are calculated relative to the C<ev_now ()>
1918 ev_timer_set (&timer, after + ev_now () - ev_time (), 0.); 2026 ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
1919 2027
1920If the event loop is suspended for a long time, you can also force an 2028If the event loop is suspended for a long time, you can also force an
1921update of the time returned by C<ev_now ()> by calling C<ev_now_update 2029update of the time returned by C<ev_now ()> by calling C<ev_now_update
1922()>. 2030()>.
2031
2032=head3 The special problem of unsynchronised clocks
2033
2034Modern systems have a variety of clocks - libev itself uses the normal
2035"wall clock" clock and, if available, the monotonic clock (to avoid time
2036jumps).
2037
2038Neither of these clocks is synchronised with each other or any other clock
2039on the system, so C<ev_time ()> might return a considerably different time
2040than C<gettimeofday ()> or C<time ()>. On a GNU/Linux system, for example,
2041a call to C<gettimeofday> might return a second count that is one higher
2042than a directly following call to C<time>.
2043
2044The moral of this is to only compare libev-related timestamps with
2045C<ev_time ()> and C<ev_now ()>, at least if you want better precision than
2046a second or so.
2047
2048One more problem arises due to this lack of synchronisation: if libev uses
2049the system monotonic clock and you compare timestamps from C<ev_time>
2050or C<ev_now> from when you started your timer and when your callback is
2051invoked, you will find that sometimes the callback is a bit "early".
2052
2053This is because C<ev_timer>s work in real time, not wall clock time, so
2054libev makes sure your callback is not invoked before the delay happened,
2055I<measured according to the real time>, not the system clock.
2056
2057If your timeouts are based on a physical timescale (e.g. "time out this
2058connection after 100 seconds") then this shouldn't bother you as it is
2059exactly the right behaviour.
2060
2061If you want to compare wall clock/system timestamps to your timers, then
2062you need to use C<ev_periodic>s, as these are based on the wall clock
2063time, where your comparisons will always generate correct results.
1923 2064
1924=head3 The special problems of suspended animation 2065=head3 The special problems of suspended animation
1925 2066
1926When you leave the server world it is quite customary to hit machines that 2067When you leave the server world it is quite customary to hit machines that
1927can suspend/hibernate - what happens to the clocks during such a suspend? 2068can suspend/hibernate - what happens to the clocks during such a suspend?
1971keep up with the timer (because it takes longer than those 10 seconds to 2112keep up with the timer (because it takes longer than those 10 seconds to
1972do stuff) the timer will not fire more than once per event loop iteration. 2113do stuff) the timer will not fire more than once per event loop iteration.
1973 2114
1974=item ev_timer_again (loop, ev_timer *) 2115=item ev_timer_again (loop, ev_timer *)
1975 2116
1976This will act as if the timer timed out and restart it again if it is 2117This will act as if the timer timed out, and restarts it again if it is
1977repeating. The exact semantics are: 2118repeating. It basically works like calling C<ev_timer_stop>, updating the
2119timeout to the C<repeat> value and calling C<ev_timer_start>.
1978 2120
2121The exact semantics are as in the following rules, all of which will be
2122applied to the watcher:
2123
2124=over 4
2125
1979If the timer is pending, its pending status is cleared. 2126=item If the timer is pending, the pending status is always cleared.
1980 2127
1981If the timer is started but non-repeating, stop it (as if it timed out). 2128=item If the timer is started but non-repeating, stop it (as if it timed
2129out, without invoking it).
1982 2130
1983If the timer is repeating, either start it if necessary (with the 2131=item If the timer is repeating, make the C<repeat> value the new timeout
1984C<repeat> value), or reset the running timer to the C<repeat> value. 2132and start the timer, if necessary.
2133
2134=back
1985 2135
1986This sounds a bit complicated, see L<Be smart about timeouts>, above, for a 2136This sounds a bit complicated, see L<Be smart about timeouts>, above, for a
1987usage example. 2137usage example.
1988 2138
1989=item ev_tstamp ev_timer_remaining (loop, ev_timer *) 2139=item ev_tstamp ev_timer_remaining (loop, ev_timer *)
2111 2261
2112Another way to think about it (for the mathematically inclined) is that 2262Another way to think about it (for the mathematically inclined) is that
2113C<ev_periodic> will try to run the callback in this mode at the next possible 2263C<ev_periodic> will try to run the callback in this mode at the next possible
2114time where C<time = offset (mod interval)>, regardless of any time jumps. 2264time where C<time = offset (mod interval)>, regardless of any time jumps.
2115 2265
2116For numerical stability it is preferable that the C<offset> value is near 2266The C<interval> I<MUST> be positive, and for numerical stability, the
2117C<ev_now ()> (the current time), but there is no range requirement for 2267interval value should be higher than C<1/8192> (which is around 100
2118this value, and in fact is often specified as zero. 2268microseconds) and C<offset> should be higher than C<0> and should have
2269at most a similar magnitude as the current time (say, within a factor of
2270ten). Typical values for offset are, in fact, C<0> or something between
2271C<0> and C<interval>, which is also the recommended range.
2119 2272
2120Note also that there is an upper limit to how often a timer can fire (CPU 2273Note also that there is an upper limit to how often a timer can fire (CPU
2121speed for example), so if C<interval> is very small then timing stability 2274speed for example), so if C<interval> is very small then timing stability
2122will of course deteriorate. Libev itself tries to be exact to be about one 2275will of course deteriorate. Libev itself tries to be exact to be about one
2123millisecond (if the OS supports it and the machine is fast enough). 2276millisecond (if the OS supports it and the machine is fast enough).
2237 2390
2238=head2 C<ev_signal> - signal me when a signal gets signalled! 2391=head2 C<ev_signal> - signal me when a signal gets signalled!
2239 2392
2240Signal watchers will trigger an event when the process receives a specific 2393Signal watchers will trigger an event when the process receives a specific
2241signal one or more times. Even though signals are very asynchronous, libev 2394signal one or more times. Even though signals are very asynchronous, libev
2242will try it's best to deliver signals synchronously, i.e. as part of the 2395will try its best to deliver signals synchronously, i.e. as part of the
2243normal event processing, like any other event. 2396normal event processing, like any other event.
2244 2397
2245If you want signals to be delivered truly asynchronously, just use 2398If you want signals to be delivered truly asynchronously, just use
2246C<sigaction> as you would do without libev and forget about sharing 2399C<sigaction> as you would do without libev and forget about sharing
2247the signal. You can even use C<ev_async> from a signal handler to 2400the signal. You can even use C<ev_async> from a signal handler to
2266=head3 The special problem of inheritance over fork/execve/pthread_create 2419=head3 The special problem of inheritance over fork/execve/pthread_create
2267 2420
2268Both the signal mask (C<sigprocmask>) and the signal disposition 2421Both the signal mask (C<sigprocmask>) and the signal disposition
2269(C<sigaction>) are unspecified after starting a signal watcher (and after 2422(C<sigaction>) are unspecified after starting a signal watcher (and after
2270stopping it again), that is, libev might or might not block the signal, 2423stopping it again), that is, libev might or might not block the signal,
2271and might or might not set or restore the installed signal handler. 2424and might or might not set or restore the installed signal handler (but
2425see C<EVFLAG_NOSIGMASK>).
2272 2426
2273While this does not matter for the signal disposition (libev never 2427While this does not matter for the signal disposition (libev never
2274sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on 2428sets signals to C<SIG_IGN>, so handlers will be reset to C<SIG_DFL> on
2275C<execve>), this matters for the signal mask: many programs do not expect 2429C<execve>), this matters for the signal mask: many programs do not expect
2276certain signals to be blocked. 2430certain signals to be blocked.
2289I<has> to modify the signal mask, at least temporarily. 2443I<has> to modify the signal mask, at least temporarily.
2290 2444
2291So I can't stress this enough: I<If you do not reset your signal mask when 2445So I can't stress this enough: I<If you do not reset your signal mask when
2292you expect it to be empty, you have a race condition in your code>. This 2446you expect it to be empty, you have a race condition in your code>. This
2293is not a libev-specific thing, this is true for most event libraries. 2447is not a libev-specific thing, this is true for most event libraries.
2448
2449=head3 The special problem of threads signal handling
2450
2451POSIX threads has problematic signal handling semantics, specifically,
2452a lot of functionality (sigfd, sigwait etc.) only really works if all
2453threads in a process block signals, which is hard to achieve.
2454
2455When you want to use sigwait (or mix libev signal handling with your own
2456for the same signals), you can tackle this problem by globally blocking
2457all signals before creating any threads (or creating them with a fully set
2458sigprocmask) and also specifying the C<EVFLAG_NOSIGMASK> when creating
2459loops. Then designate one thread as "signal receiver thread" which handles
2460these signals. You can pass on any signals that libev might be interested
2461in by calling C<ev_feed_signal>.
2294 2462
2295=head3 Watcher-Specific Functions and Data Members 2463=head3 Watcher-Specific Functions and Data Members
2296 2464
2297=over 4 2465=over 4
2298 2466
3072disadvantage of having to use multiple event loops (which do not support 3240disadvantage of having to use multiple event loops (which do not support
3073signal watchers). 3241signal watchers).
3074 3242
3075When this is not possible, or you want to use the default loop for 3243When this is not possible, or you want to use the default loop for
3076other reasons, then in the process that wants to start "fresh", call 3244other reasons, then in the process that wants to start "fresh", call
3077C<ev_default_destroy ()> followed by C<ev_default_loop (...)>. Destroying 3245C<ev_loop_destroy (EV_DEFAULT)> followed by C<ev_default_loop (...)>.
3078the default loop will "orphan" (not stop) all registered watchers, so you 3246Destroying the default loop will "orphan" (not stop) all registered
3079have to be careful not to execute code that modifies those watchers. Note 3247watchers, so you have to be careful not to execute code that modifies
3080also that in that case, you have to re-register any signal watchers. 3248those watchers. Note also that in that case, you have to re-register any
3249signal watchers.
3081 3250
3082=head3 Watcher-Specific Functions and Data Members 3251=head3 Watcher-Specific Functions and Data Members
3083 3252
3084=over 4 3253=over 4
3085 3254
3086=item ev_fork_init (ev_signal *, callback) 3255=item ev_fork_init (ev_fork *, callback)
3087 3256
3088Initialises and configures the fork watcher - it has no parameters of any 3257Initialises and configures the fork watcher - it has no parameters of any
3089kind. There is a C<ev_fork_set> macro, but using it is utterly pointless, 3258kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
3090believe me. 3259really.
3091 3260
3092=back 3261=back
3093 3262
3094 3263
3264=head2 C<ev_cleanup> - even the best things end
3265
3266Cleanup watchers are called just before the event loop is being destroyed
3267by a call to C<ev_loop_destroy>.
3268
3269While there is no guarantee that the event loop gets destroyed, cleanup
3270watchers provide a convenient method to install cleanup hooks for your
3271program, worker threads and so on - you just to make sure to destroy the
3272loop when you want them to be invoked.
3273
3274Cleanup watchers are invoked in the same way as any other watcher. Unlike
3275all other watchers, they do not keep a reference to the event loop (which
3276makes a lot of sense if you think about it). Like all other watchers, you
3277can call libev functions in the callback, except C<ev_cleanup_start>.
3278
3279=head3 Watcher-Specific Functions and Data Members
3280
3281=over 4
3282
3283=item ev_cleanup_init (ev_cleanup *, callback)
3284
3285Initialises and configures the cleanup watcher - it has no parameters of
3286any kind. There is a C<ev_cleanup_set> macro, but using it is utterly
3287pointless, I assure you.
3288
3289=back
3290
3291Example: Register an atexit handler to destroy the default loop, so any
3292cleanup functions are called.
3293
3294 static void
3295 program_exits (void)
3296 {
3297 ev_loop_destroy (EV_DEFAULT_UC);
3298 }
3299
3300 ...
3301 atexit (program_exits);
3302
3303
3095=head2 C<ev_async> - how to wake up an event loop 3304=head2 C<ev_async> - how to wake up an event loop
3096 3305
3097In general, you cannot use an C<ev_run> from multiple threads or other 3306In general, you cannot use an C<ev_loop> from multiple threads or other
3098asynchronous sources such as signal handlers (as opposed to multiple event 3307asynchronous sources such as signal handlers (as opposed to multiple event
3099loops - those are of course safe to use in different threads). 3308loops - those are of course safe to use in different threads).
3100 3309
3101Sometimes, however, you need to wake up an event loop you do not control, 3310Sometimes, however, you need to wake up an event loop you do not control,
3102for example because it belongs to another thread. This is what C<ev_async> 3311for example because it belongs to another thread. This is what C<ev_async>
3104it by calling C<ev_async_send>, which is thread- and signal safe. 3313it by calling C<ev_async_send>, which is thread- and signal safe.
3105 3314
3106This functionality is very similar to C<ev_signal> watchers, as signals, 3315This functionality is very similar to C<ev_signal> watchers, as signals,
3107too, are asynchronous in nature, and signals, too, will be compressed 3316too, are asynchronous in nature, and signals, too, will be compressed
3108(i.e. the number of callback invocations may be less than the number of 3317(i.e. the number of callback invocations may be less than the number of
3109C<ev_async_sent> calls). 3318C<ev_async_sent> calls). In fact, you could use signal watchers as a kind
3110 3319of "global async watchers" by using a watcher on an otherwise unused
3111Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not 3320signal, and C<ev_feed_signal> to signal this watcher from another thread,
3112just the default loop. 3321even without knowing which loop owns the signal.
3113 3322
3114=head3 Queueing 3323=head3 Queueing
3115 3324
3116C<ev_async> does not support queueing of data in any way. The reason 3325C<ev_async> does not support queueing of data in any way. The reason
3117is that the author does not know of a simple (or any) algorithm for a 3326is that the author does not know of a simple (or any) algorithm for a
3209trust me. 3418trust me.
3210 3419
3211=item ev_async_send (loop, ev_async *) 3420=item ev_async_send (loop, ev_async *)
3212 3421
3213Sends/signals/activates the given C<ev_async> watcher, that is, feeds 3422Sends/signals/activates the given C<ev_async> watcher, that is, feeds
3214an C<EV_ASYNC> event on the watcher into the event loop. Unlike 3423an C<EV_ASYNC> event on the watcher into the event loop, and instantly
3424returns.
3425
3215C<ev_feed_event>, this call is safe to do from other threads, signal or 3426Unlike C<ev_feed_event>, this call is safe to do from other threads,
3216similar contexts (see the discussion of C<EV_ATOMIC_T> in the embedding 3427signal or similar contexts (see the discussion of C<EV_ATOMIC_T> in the
3217section below on what exactly this means). 3428embedding section below on what exactly this means).
3218 3429
3219Note that, as with other watchers in libev, multiple events might get 3430Note that, as with other watchers in libev, multiple events might get
3220compressed into a single callback invocation (another way to look at this 3431compressed into a single callback invocation (another way to look at
3221is that C<ev_async> watchers are level-triggered, set on C<ev_async_send>, 3432this is that C<ev_async> watchers are level-triggered: they are set on
3222reset when the event loop detects that). 3433C<ev_async_send>, reset when the event loop detects that).
3223 3434
3224This call incurs the overhead of a system call only once per event loop 3435This call incurs the overhead of at most one extra system call per event
3225iteration, so while the overhead might be noticeable, it doesn't apply to 3436loop iteration, if the event loop is blocked, and no syscall at all if
3226repeated calls to C<ev_async_send> for the same event loop. 3437the event loop (or your program) is processing events. That means that
3438repeated calls are basically free (there is no need to avoid calls for
3439performance reasons) and that the overhead becomes smaller (typically
3440zero) under load.
3227 3441
3228=item bool = ev_async_pending (ev_async *) 3442=item bool = ev_async_pending (ev_async *)
3229 3443
3230Returns a non-zero value when C<ev_async_send> has been called on the 3444Returns a non-zero value when C<ev_async_send> has been called on the
3231watcher but the event has not yet been processed (or even noted) by the 3445watcher but the event has not yet been processed (or even noted) by the
3286 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0); 3500 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
3287 3501
3288=item ev_feed_fd_event (loop, int fd, int revents) 3502=item ev_feed_fd_event (loop, int fd, int revents)
3289 3503
3290Feed an event on the given fd, as if a file descriptor backend detected 3504Feed an event on the given fd, as if a file descriptor backend detected
3291the given events it. 3505the given events.
3292 3506
3293=item ev_feed_signal_event (loop, int signum) 3507=item ev_feed_signal_event (loop, int signum)
3294 3508
3295Feed an event as if the given signal occurred (C<loop> must be the default 3509Feed an event as if the given signal occurred. See also C<ev_feed_signal>,
3296loop!). 3510which is async-safe.
3297 3511
3298=back 3512=back
3513
3514
3515=head1 COMMON OR USEFUL IDIOMS (OR BOTH)
3516
3517This section explains some common idioms that are not immediately
3518obvious. Note that examples are sprinkled over the whole manual, and this
3519section only contains stuff that wouldn't fit anywhere else.
3520
3521=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
3522
3523Each watcher has, by default, a C<void *data> member that you can read
3524or modify at any time: libev will completely ignore it. This can be used
3525to associate arbitrary data with your watcher. If you need more data and
3526don't want to allocate memory separately and store a pointer to it in that
3527data member, you can also "subclass" the watcher type and provide your own
3528data:
3529
3530 struct my_io
3531 {
3532 ev_io io;
3533 int otherfd;
3534 void *somedata;
3535 struct whatever *mostinteresting;
3536 };
3537
3538 ...
3539 struct my_io w;
3540 ev_io_init (&w.io, my_cb, fd, EV_READ);
3541
3542And since your callback will be called with a pointer to the watcher, you
3543can cast it back to your own type:
3544
3545 static void my_cb (struct ev_loop *loop, ev_io *w_, int revents)
3546 {
3547 struct my_io *w = (struct my_io *)w_;
3548 ...
3549 }
3550
3551More interesting and less C-conformant ways of casting your callback
3552function type instead have been omitted.
3553
3554=head2 BUILDING YOUR OWN COMPOSITE WATCHERS
3555
3556Another common scenario is to use some data structure with multiple
3557embedded watchers, in effect creating your own watcher that combines
3558multiple libev event sources into one "super-watcher":
3559
3560 struct my_biggy
3561 {
3562 int some_data;
3563 ev_timer t1;
3564 ev_timer t2;
3565 }
3566
3567In this case getting the pointer to C<my_biggy> is a bit more
3568complicated: Either you store the address of your C<my_biggy> struct in
3569the C<data> member of the watcher (for woozies or C++ coders), or you need
3570to use some pointer arithmetic using C<offsetof> inside your watchers (for
3571real programmers):
3572
3573 #include <stddef.h>
3574
3575 static void
3576 t1_cb (EV_P_ ev_timer *w, int revents)
3577 {
3578 struct my_biggy big = (struct my_biggy *)
3579 (((char *)w) - offsetof (struct my_biggy, t1));
3580 }
3581
3582 static void
3583 t2_cb (EV_P_ ev_timer *w, int revents)
3584 {
3585 struct my_biggy big = (struct my_biggy *)
3586 (((char *)w) - offsetof (struct my_biggy, t2));
3587 }
3588
3589=head2 AVOIDING FINISHING BEFORE RETURNING
3590
3591Often you have structures like this in event-based programs:
3592
3593 callback ()
3594 {
3595 free (request);
3596 }
3597
3598 request = start_new_request (..., callback);
3599
3600The intent is to start some "lengthy" operation. The C<request> could be
3601used to cancel the operation, or do other things with it.
3602
3603It's not uncommon to have code paths in C<start_new_request> that
3604immediately invoke the callback, for example, to report errors. Or you add
3605some caching layer that finds that it can skip the lengthy aspects of the
3606operation and simply invoke the callback with the result.
3607
3608The problem here is that this will happen I<before> C<start_new_request>
3609has returned, so C<request> is not set.
3610
3611Even if you pass the request by some safer means to the callback, you
3612might want to do something to the request after starting it, such as
3613canceling it, which probably isn't working so well when the callback has
3614already been invoked.
3615
3616A common way around all these issues is to make sure that
3617C<start_new_request> I<always> returns before the callback is invoked. If
3618C<start_new_request> immediately knows the result, it can artificially
3619delay invoking the callback by e.g. using a C<prepare> or C<idle> watcher
3620for example, or more sneakily, by reusing an existing (stopped) watcher
3621and pushing it into the pending queue:
3622
3623 ev_set_cb (watcher, callback);
3624 ev_feed_event (EV_A_ watcher, 0);
3625
3626This way, C<start_new_request> can safely return before the callback is
3627invoked, while not delaying callback invocation too much.
3628
3629=head2 MODEL/NESTED EVENT LOOP INVOCATIONS AND EXIT CONDITIONS
3630
3631Often (especially in GUI toolkits) there are places where you have
3632I<modal> interaction, which is most easily implemented by recursively
3633invoking C<ev_run>.
3634
3635This brings the problem of exiting - a callback might want to finish the
3636main C<ev_run> call, but not the nested one (e.g. user clicked "Quit", but
3637a modal "Are you sure?" dialog is still waiting), or just the nested one
3638and not the main one (e.g. user clocked "Ok" in a modal dialog), or some
3639other combination: In these cases, C<ev_break> will not work alone.
3640
3641The solution is to maintain "break this loop" variable for each C<ev_run>
3642invocation, and use a loop around C<ev_run> until the condition is
3643triggered, using C<EVRUN_ONCE>:
3644
3645 // main loop
3646 int exit_main_loop = 0;
3647
3648 while (!exit_main_loop)
3649 ev_run (EV_DEFAULT_ EVRUN_ONCE);
3650
3651 // in a modal watcher
3652 int exit_nested_loop = 0;
3653
3654 while (!exit_nested_loop)
3655 ev_run (EV_A_ EVRUN_ONCE);
3656
3657To exit from any of these loops, just set the corresponding exit variable:
3658
3659 // exit modal loop
3660 exit_nested_loop = 1;
3661
3662 // exit main program, after modal loop is finished
3663 exit_main_loop = 1;
3664
3665 // exit both
3666 exit_main_loop = exit_nested_loop = 1;
3667
3668=head2 THREAD LOCKING EXAMPLE
3669
3670Here is a fictitious example of how to run an event loop in a different
3671thread from where callbacks are being invoked and watchers are
3672created/added/removed.
3673
3674For a real-world example, see the C<EV::Loop::Async> perl module,
3675which uses exactly this technique (which is suited for many high-level
3676languages).
3677
3678The example uses a pthread mutex to protect the loop data, a condition
3679variable to wait for callback invocations, an async watcher to notify the
3680event loop thread and an unspecified mechanism to wake up the main thread.
3681
3682First, you need to associate some data with the event loop:
3683
3684 typedef struct {
3685 mutex_t lock; /* global loop lock */
3686 ev_async async_w;
3687 thread_t tid;
3688 cond_t invoke_cv;
3689 } userdata;
3690
3691 void prepare_loop (EV_P)
3692 {
3693 // for simplicity, we use a static userdata struct.
3694 static userdata u;
3695
3696 ev_async_init (&u->async_w, async_cb);
3697 ev_async_start (EV_A_ &u->async_w);
3698
3699 pthread_mutex_init (&u->lock, 0);
3700 pthread_cond_init (&u->invoke_cv, 0);
3701
3702 // now associate this with the loop
3703 ev_set_userdata (EV_A_ u);
3704 ev_set_invoke_pending_cb (EV_A_ l_invoke);
3705 ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
3706
3707 // then create the thread running ev_run
3708 pthread_create (&u->tid, 0, l_run, EV_A);
3709 }
3710
3711The callback for the C<ev_async> watcher does nothing: the watcher is used
3712solely to wake up the event loop so it takes notice of any new watchers
3713that might have been added:
3714
3715 static void
3716 async_cb (EV_P_ ev_async *w, int revents)
3717 {
3718 // just used for the side effects
3719 }
3720
3721The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
3722protecting the loop data, respectively.
3723
3724 static void
3725 l_release (EV_P)
3726 {
3727 userdata *u = ev_userdata (EV_A);
3728 pthread_mutex_unlock (&u->lock);
3729 }
3730
3731 static void
3732 l_acquire (EV_P)
3733 {
3734 userdata *u = ev_userdata (EV_A);
3735 pthread_mutex_lock (&u->lock);
3736 }
3737
3738The event loop thread first acquires the mutex, and then jumps straight
3739into C<ev_run>:
3740
3741 void *
3742 l_run (void *thr_arg)
3743 {
3744 struct ev_loop *loop = (struct ev_loop *)thr_arg;
3745
3746 l_acquire (EV_A);
3747 pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
3748 ev_run (EV_A_ 0);
3749 l_release (EV_A);
3750
3751 return 0;
3752 }
3753
3754Instead of invoking all pending watchers, the C<l_invoke> callback will
3755signal the main thread via some unspecified mechanism (signals? pipe
3756writes? C<Async::Interrupt>?) and then waits until all pending watchers
3757have been called (in a while loop because a) spurious wakeups are possible
3758and b) skipping inter-thread-communication when there are no pending
3759watchers is very beneficial):
3760
3761 static void
3762 l_invoke (EV_P)
3763 {
3764 userdata *u = ev_userdata (EV_A);
3765
3766 while (ev_pending_count (EV_A))
3767 {
3768 wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
3769 pthread_cond_wait (&u->invoke_cv, &u->lock);
3770 }
3771 }
3772
3773Now, whenever the main thread gets told to invoke pending watchers, it
3774will grab the lock, call C<ev_invoke_pending> and then signal the loop
3775thread to continue:
3776
3777 static void
3778 real_invoke_pending (EV_P)
3779 {
3780 userdata *u = ev_userdata (EV_A);
3781
3782 pthread_mutex_lock (&u->lock);
3783 ev_invoke_pending (EV_A);
3784 pthread_cond_signal (&u->invoke_cv);
3785 pthread_mutex_unlock (&u->lock);
3786 }
3787
3788Whenever you want to start/stop a watcher or do other modifications to an
3789event loop, you will now have to lock:
3790
3791 ev_timer timeout_watcher;
3792 userdata *u = ev_userdata (EV_A);
3793
3794 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
3795
3796 pthread_mutex_lock (&u->lock);
3797 ev_timer_start (EV_A_ &timeout_watcher);
3798 ev_async_send (EV_A_ &u->async_w);
3799 pthread_mutex_unlock (&u->lock);
3800
3801Note that sending the C<ev_async> watcher is required because otherwise
3802an event loop currently blocking in the kernel will have no knowledge
3803about the newly added timer. By waking up the loop it will pick up any new
3804watchers in the next event loop iteration.
3805
3806=head2 THREADS, COROUTINES, CONTINUATIONS, QUEUES... INSTEAD OF CALLBACKS
3807
3808While the overhead of a callback that e.g. schedules a thread is small, it
3809is still an overhead. If you embed libev, and your main usage is with some
3810kind of threads or coroutines, you might want to customise libev so that
3811doesn't need callbacks anymore.
3812
3813Imagine you have coroutines that you can switch to using a function
3814C<switch_to (coro)>, that libev runs in a coroutine called C<libev_coro>
3815and that due to some magic, the currently active coroutine is stored in a
3816global called C<current_coro>. Then you can build your own "wait for libev
3817event" primitive by changing C<EV_CB_DECLARE> and C<EV_CB_INVOKE> (note
3818the differing C<;> conventions):
3819
3820 #define EV_CB_DECLARE(type) struct my_coro *cb;
3821 #define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb)
3822
3823That means instead of having a C callback function, you store the
3824coroutine to switch to in each watcher, and instead of having libev call
3825your callback, you instead have it switch to that coroutine.
3826
3827A coroutine might now wait for an event with a function called
3828C<wait_for_event>. (the watcher needs to be started, as always, but it doesn't
3829matter when, or whether the watcher is active or not when this function is
3830called):
3831
3832 void
3833 wait_for_event (ev_watcher *w)
3834 {
3835 ev_cb_set (w) = current_coro;
3836 switch_to (libev_coro);
3837 }
3838
3839That basically suspends the coroutine inside C<wait_for_event> and
3840continues the libev coroutine, which, when appropriate, switches back to
3841this or any other coroutine.
3842
3843You can do similar tricks if you have, say, threads with an event queue -
3844instead of storing a coroutine, you store the queue object and instead of
3845switching to a coroutine, you push the watcher onto the queue and notify
3846any waiters.
3847
3848To embed libev, see L<EMBEDDING>, but in short, it's easiest to create two
3849files, F<my_ev.h> and F<my_ev.c> that include the respective libev files:
3850
3851 // my_ev.h
3852 #define EV_CB_DECLARE(type) struct my_coro *cb;
3853 #define EV_CB_INVOKE(watcher) switch_to ((watcher)->cb);
3854 #include "../libev/ev.h"
3855
3856 // my_ev.c
3857 #define EV_H "my_ev.h"
3858 #include "../libev/ev.c"
3859
3860And then use F<my_ev.h> when you would normally use F<ev.h>, and compile
3861F<my_ev.c> into your project. When properly specifying include paths, you
3862can even use F<ev.h> as header file name directly.
3299 3863
3300 3864
3301=head1 LIBEVENT EMULATION 3865=head1 LIBEVENT EMULATION
3302 3866
3303Libev offers a compatibility emulation layer for libevent. It cannot 3867Libev offers a compatibility emulation layer for libevent. It cannot
3304emulate the internals of libevent, so here are some usage hints: 3868emulate the internals of libevent, so here are some usage hints:
3305 3869
3306=over 4 3870=over 4
3871
3872=item * Only the libevent-1.4.1-beta API is being emulated.
3873
3874This was the newest libevent version available when libev was implemented,
3875and is still mostly unchanged in 2010.
3307 3876
3308=item * Use it by including <event.h>, as usual. 3877=item * Use it by including <event.h>, as usual.
3309 3878
3310=item * The following members are fully supported: ev_base, ev_callback, 3879=item * The following members are fully supported: ev_base, ev_callback,
3311ev_arg, ev_fd, ev_res, ev_events. 3880ev_arg, ev_fd, ev_res, ev_events.
3317=item * Priorities are not currently supported. Initialising priorities 3886=item * Priorities are not currently supported. Initialising priorities
3318will fail and all watchers will have the same priority, even though there 3887will fail and all watchers will have the same priority, even though there
3319is an ev_pri field. 3888is an ev_pri field.
3320 3889
3321=item * In libevent, the last base created gets the signals, in libev, the 3890=item * In libevent, the last base created gets the signals, in libev, the
3322first base created (== the default loop) gets the signals. 3891base that registered the signal gets the signals.
3323 3892
3324=item * Other members are not supported. 3893=item * Other members are not supported.
3325 3894
3326=item * The libev emulation is I<not> ABI compatible to libevent, you need 3895=item * The libev emulation is I<not> ABI compatible to libevent, you need
3327to use the libev header file and library. 3896to use the libev header file and library.
3346Care has been taken to keep the overhead low. The only data member the C++ 3915Care has been taken to keep the overhead low. The only data member the C++
3347classes add (compared to plain C-style watchers) is the event loop pointer 3916classes add (compared to plain C-style watchers) is the event loop pointer
3348that the watcher is associated with (or no additional members at all if 3917that the watcher is associated with (or no additional members at all if
3349you disable C<EV_MULTIPLICITY> when embedding libev). 3918you disable C<EV_MULTIPLICITY> when embedding libev).
3350 3919
3351Currently, functions, and static and non-static member functions can be 3920Currently, functions, static and non-static member functions and classes
3352used as callbacks. Other types should be easy to add as long as they only 3921with C<operator ()> can be used as callbacks. Other types should be easy
3353need one additional pointer for context. If you need support for other 3922to add as long as they only need one additional pointer for context. If
3354types of functors please contact the author (preferably after implementing 3923you need support for other types of functors please contact the author
3355it). 3924(preferably after implementing it).
3925
3926For all this to work, your C++ compiler either has to use the same calling
3927conventions as your C compiler (for static member functions), or you have
3928to embed libev and compile libev itself as C++.
3356 3929
3357Here is a list of things available in the C<ev> namespace: 3930Here is a list of things available in the C<ev> namespace:
3358 3931
3359=over 4 3932=over 4
3360 3933
3370=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc. 3943=item C<ev::io>, C<ev::timer>, C<ev::periodic>, C<ev::idle>, C<ev::sig> etc.
3371 3944
3372For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of 3945For each C<ev_TYPE> watcher in F<ev.h> there is a corresponding class of
3373the same name in the C<ev> namespace, with the exception of C<ev_signal> 3946the same name in the C<ev> namespace, with the exception of C<ev_signal>
3374which is called C<ev::sig> to avoid clashes with the C<signal> macro 3947which is called C<ev::sig> to avoid clashes with the C<signal> macro
3375defines by many implementations. 3948defined by many implementations.
3376 3949
3377All of those classes have these methods: 3950All of those classes have these methods:
3378 3951
3379=over 4 3952=over 4
3380 3953
3513watchers in the constructor. 4086watchers in the constructor.
3514 4087
3515 class myclass 4088 class myclass
3516 { 4089 {
3517 ev::io io ; void io_cb (ev::io &w, int revents); 4090 ev::io io ; void io_cb (ev::io &w, int revents);
3518 ev::io2 io2 ; void io2_cb (ev::io &w, int revents); 4091 ev::io io2 ; void io2_cb (ev::io &w, int revents);
3519 ev::idle idle; void idle_cb (ev::idle &w, int revents); 4092 ev::idle idle; void idle_cb (ev::idle &w, int revents);
3520 4093
3521 myclass (int fd) 4094 myclass (int fd)
3522 { 4095 {
3523 io .set <myclass, &myclass::io_cb > (this); 4096 io .set <myclass, &myclass::io_cb > (this);
3574L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>. 4147L<http://hackage.haskell.org/cgi-bin/hackage-scripts/package/hlibev>.
3575 4148
3576=item D 4149=item D
3577 4150
3578Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to 4151Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
3579be found at L<http://proj.llucax.com.ar/wiki/evd>. 4152be found at L<http://www.llucax.com.ar/proj/ev.d/index.html>.
3580 4153
3581=item Ocaml 4154=item Ocaml
3582 4155
3583Erkki Seppala has written Ocaml bindings for libev, to be found at 4156Erkki Seppala has written Ocaml bindings for libev, to be found at
3584L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>. 4157L<http://modeemi.cs.tut.fi/~flux/software/ocaml-ev/>.
3632suitable for use with C<EV_A>. 4205suitable for use with C<EV_A>.
3633 4206
3634=item C<EV_DEFAULT>, C<EV_DEFAULT_> 4207=item C<EV_DEFAULT>, C<EV_DEFAULT_>
3635 4208
3636Similar to the other two macros, this gives you the value of the default 4209Similar to the other two macros, this gives you the value of the default
3637loop, if multiple loops are supported ("ev loop default"). 4210loop, if multiple loops are supported ("ev loop default"). The default loop
4211will be initialised if it isn't already initialised.
4212
4213For non-multiplicity builds, these macros do nothing, so you always have
4214to initialise the loop somewhere.
3638 4215
3639=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_> 4216=item C<EV_DEFAULT_UC>, C<EV_DEFAULT_UC_>
3640 4217
3641Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the 4218Usage identical to C<EV_DEFAULT> and C<EV_DEFAULT_>, but requires that the
3642default loop has been initialised (C<UC> == unchecked). Their behaviour 4219default loop has been initialised (C<UC> == unchecked). Their behaviour
3787supported). It will also not define any of the structs usually found in 4364supported). It will also not define any of the structs usually found in
3788F<event.h> that are not directly supported by the libev core alone. 4365F<event.h> that are not directly supported by the libev core alone.
3789 4366
3790In standalone mode, libev will still try to automatically deduce the 4367In standalone mode, libev will still try to automatically deduce the
3791configuration, but has to be more conservative. 4368configuration, but has to be more conservative.
4369
4370=item EV_USE_FLOOR
4371
4372If defined to be C<1>, libev will use the C<floor ()> function for its
4373periodic reschedule calculations, otherwise libev will fall back on a
4374portable (slower) implementation. If you enable this, you usually have to
4375link against libm or something equivalent. Enabling this when the C<floor>
4376function is not available will fail, so the safe default is to not enable
4377this.
3792 4378
3793=item EV_USE_MONOTONIC 4379=item EV_USE_MONOTONIC
3794 4380
3795If defined to be C<1>, libev will try to detect the availability of the 4381If defined to be C<1>, libev will try to detect the availability of the
3796monotonic clock option at both compile time and runtime. Otherwise no 4382monotonic clock option at both compile time and runtime. Otherwise no
3926If defined to be C<1>, libev will compile in support for the Linux inotify 4512If defined to be C<1>, libev will compile in support for the Linux inotify
3927interface to speed up C<ev_stat> watchers. Its actual availability will 4513interface to speed up C<ev_stat> watchers. Its actual availability will
3928be detected at runtime. If undefined, it will be enabled if the headers 4514be detected at runtime. If undefined, it will be enabled if the headers
3929indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled. 4515indicate GNU/Linux + Glibc 2.4 or newer, otherwise disabled.
3930 4516
4517=item EV_NO_SMP
4518
4519If defined to be C<1>, libev will assume that memory is always coherent
4520between threads, that is, threads can be used, but threads never run on
4521different cpus (or different cpu cores). This reduces dependencies
4522and makes libev faster.
4523
4524=item EV_NO_THREADS
4525
4526If defined to be C<1>, libev will assume that it will never be called
4527from different threads, which is a stronger assumption than C<EV_NO_SMP>,
4528above. This reduces dependencies and makes libev faster.
4529
3931=item EV_ATOMIC_T 4530=item EV_ATOMIC_T
3932 4531
3933Libev requires an integer type (suitable for storing C<0> or C<1>) whose 4532Libev requires an integer type (suitable for storing C<0> or C<1>) whose
3934access is atomic with respect to other threads or signal contexts. No such 4533access is atomic and serialised with respect to other threads or signal
3935type is easily found in the C language, so you can provide your own type 4534contexts. No such type is easily found in the C language, so you can
3936that you know is safe for your purposes. It is used both for signal handler "locking" 4535provide your own type that you know is safe for your purposes. It is used
3937as well as for signal and thread safety in C<ev_async> watchers. 4536both for signal handler "locking" as well as for signal and thread safety
4537in C<ev_async> watchers.
3938 4538
3939In the absence of this define, libev will use C<sig_atomic_t volatile> 4539In the absence of this define, libev will use C<sig_atomic_t volatile>
3940(from F<signal.h>), which is usually good enough on most platforms. 4540(from F<signal.h>), which is usually good enough on most platforms,
4541although strictly speaking using a type that also implies a memory fence
4542is required.
3941 4543
3942=item EV_H (h) 4544=item EV_H (h)
3943 4545
3944The name of the F<ev.h> header file used to include it. The default if 4546The name of the F<ev.h> header file used to include it. The default if
3945undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be 4547undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
3969will have the C<struct ev_loop *> as first argument, and you can create 4571will have the C<struct ev_loop *> as first argument, and you can create
3970additional independent event loops. Otherwise there will be no support 4572additional independent event loops. Otherwise there will be no support
3971for multiple event loops and there is no first event loop pointer 4573for multiple event loops and there is no first event loop pointer
3972argument. Instead, all functions act on the single default loop. 4574argument. Instead, all functions act on the single default loop.
3973 4575
4576Note that C<EV_DEFAULT> and C<EV_DEFAULT_> will no longer provide a
4577default loop when multiplicity is switched off - you always have to
4578initialise the loop manually in this case.
4579
3974=item EV_MINPRI 4580=item EV_MINPRI
3975 4581
3976=item EV_MAXPRI 4582=item EV_MAXPRI
3977 4583
3978The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to 4584The range of allowed priorities. C<EV_MINPRI> must be smaller or equal to
4075 4681
4076With an intelligent-enough linker (gcc+binutils are intelligent enough 4682With an intelligent-enough linker (gcc+binutils are intelligent enough
4077when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by 4683when you use C<-Wl,--gc-sections -ffunction-sections>) functions unused by
4078your program might be left out as well - a binary starting a timer and an 4684your program might be left out as well - a binary starting a timer and an
4079I/O watcher then might come out at only 5Kb. 4685I/O watcher then might come out at only 5Kb.
4686
4687=item EV_API_STATIC
4688
4689If this symbol is defined (by default it is not), then all identifiers
4690will have static linkage. This means that libev will not export any
4691identifiers, and you cannot link against libev anymore. This can be useful
4692when you embed libev, only want to use libev functions in a single file,
4693and do not want its identifiers to be visible.
4694
4695To use this, define C<EV_API_STATIC> and include F<ev.c> in the file that
4696wants to use libev.
4697
4698This option only works when libev is compiled with a C compiler, as C++
4699doesn't support the required declaration syntax.
4080 4700
4081=item EV_AVOID_STDIO 4701=item EV_AVOID_STDIO
4082 4702
4083If this is set to C<1> at compiletime, then libev will avoid using stdio 4703If this is set to C<1> at compiletime, then libev will avoid using stdio
4084functions (printf, scanf, perror etc.). This will increase the code size 4704functions (printf, scanf, perror etc.). This will increase the code size
4228And a F<ev_cpp.C> implementation file that contains libev proper and is compiled: 4848And a F<ev_cpp.C> implementation file that contains libev proper and is compiled:
4229 4849
4230 #include "ev_cpp.h" 4850 #include "ev_cpp.h"
4231 #include "ev.c" 4851 #include "ev.c"
4232 4852
4233=head1 INTERACTION WITH OTHER PROGRAMS OR LIBRARIES 4853=head1 INTERACTION WITH OTHER PROGRAMS, LIBRARIES OR THE ENVIRONMENT
4234 4854
4235=head2 THREADS AND COROUTINES 4855=head2 THREADS AND COROUTINES
4236 4856
4237=head3 THREADS 4857=head3 THREADS
4238 4858
4289default loop and triggering an C<ev_async> watcher from the default loop 4909default loop and triggering an C<ev_async> watcher from the default loop
4290watcher callback into the event loop interested in the signal. 4910watcher callback into the event loop interested in the signal.
4291 4911
4292=back 4912=back
4293 4913
4294=head4 THREAD LOCKING EXAMPLE 4914See also L<THREAD LOCKING EXAMPLE>.
4295
4296Here is a fictitious example of how to run an event loop in a different
4297thread than where callbacks are being invoked and watchers are
4298created/added/removed.
4299
4300For a real-world example, see the C<EV::Loop::Async> perl module,
4301which uses exactly this technique (which is suited for many high-level
4302languages).
4303
4304The example uses a pthread mutex to protect the loop data, a condition
4305variable to wait for callback invocations, an async watcher to notify the
4306event loop thread and an unspecified mechanism to wake up the main thread.
4307
4308First, you need to associate some data with the event loop:
4309
4310 typedef struct {
4311 mutex_t lock; /* global loop lock */
4312 ev_async async_w;
4313 thread_t tid;
4314 cond_t invoke_cv;
4315 } userdata;
4316
4317 void prepare_loop (EV_P)
4318 {
4319 // for simplicity, we use a static userdata struct.
4320 static userdata u;
4321
4322 ev_async_init (&u->async_w, async_cb);
4323 ev_async_start (EV_A_ &u->async_w);
4324
4325 pthread_mutex_init (&u->lock, 0);
4326 pthread_cond_init (&u->invoke_cv, 0);
4327
4328 // now associate this with the loop
4329 ev_set_userdata (EV_A_ u);
4330 ev_set_invoke_pending_cb (EV_A_ l_invoke);
4331 ev_set_loop_release_cb (EV_A_ l_release, l_acquire);
4332
4333 // then create the thread running ev_loop
4334 pthread_create (&u->tid, 0, l_run, EV_A);
4335 }
4336
4337The callback for the C<ev_async> watcher does nothing: the watcher is used
4338solely to wake up the event loop so it takes notice of any new watchers
4339that might have been added:
4340
4341 static void
4342 async_cb (EV_P_ ev_async *w, int revents)
4343 {
4344 // just used for the side effects
4345 }
4346
4347The C<l_release> and C<l_acquire> callbacks simply unlock/lock the mutex
4348protecting the loop data, respectively.
4349
4350 static void
4351 l_release (EV_P)
4352 {
4353 userdata *u = ev_userdata (EV_A);
4354 pthread_mutex_unlock (&u->lock);
4355 }
4356
4357 static void
4358 l_acquire (EV_P)
4359 {
4360 userdata *u = ev_userdata (EV_A);
4361 pthread_mutex_lock (&u->lock);
4362 }
4363
4364The event loop thread first acquires the mutex, and then jumps straight
4365into C<ev_run>:
4366
4367 void *
4368 l_run (void *thr_arg)
4369 {
4370 struct ev_loop *loop = (struct ev_loop *)thr_arg;
4371
4372 l_acquire (EV_A);
4373 pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
4374 ev_run (EV_A_ 0);
4375 l_release (EV_A);
4376
4377 return 0;
4378 }
4379
4380Instead of invoking all pending watchers, the C<l_invoke> callback will
4381signal the main thread via some unspecified mechanism (signals? pipe
4382writes? C<Async::Interrupt>?) and then waits until all pending watchers
4383have been called (in a while loop because a) spurious wakeups are possible
4384and b) skipping inter-thread-communication when there are no pending
4385watchers is very beneficial):
4386
4387 static void
4388 l_invoke (EV_P)
4389 {
4390 userdata *u = ev_userdata (EV_A);
4391
4392 while (ev_pending_count (EV_A))
4393 {
4394 wake_up_other_thread_in_some_magic_or_not_so_magic_way ();
4395 pthread_cond_wait (&u->invoke_cv, &u->lock);
4396 }
4397 }
4398
4399Now, whenever the main thread gets told to invoke pending watchers, it
4400will grab the lock, call C<ev_invoke_pending> and then signal the loop
4401thread to continue:
4402
4403 static void
4404 real_invoke_pending (EV_P)
4405 {
4406 userdata *u = ev_userdata (EV_A);
4407
4408 pthread_mutex_lock (&u->lock);
4409 ev_invoke_pending (EV_A);
4410 pthread_cond_signal (&u->invoke_cv);
4411 pthread_mutex_unlock (&u->lock);
4412 }
4413
4414Whenever you want to start/stop a watcher or do other modifications to an
4415event loop, you will now have to lock:
4416
4417 ev_timer timeout_watcher;
4418 userdata *u = ev_userdata (EV_A);
4419
4420 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
4421
4422 pthread_mutex_lock (&u->lock);
4423 ev_timer_start (EV_A_ &timeout_watcher);
4424 ev_async_send (EV_A_ &u->async_w);
4425 pthread_mutex_unlock (&u->lock);
4426
4427Note that sending the C<ev_async> watcher is required because otherwise
4428an event loop currently blocking in the kernel will have no knowledge
4429about the newly added timer. By waking up the loop it will pick up any new
4430watchers in the next event loop iteration.
4431 4915
4432=head3 COROUTINES 4916=head3 COROUTINES
4433 4917
4434Libev is very accommodating to coroutines ("cooperative threads"): 4918Libev is very accommodating to coroutines ("cooperative threads"):
4435libev fully supports nesting calls to its functions from different 4919libev fully supports nesting calls to its functions from different
4600requires, and its I/O model is fundamentally incompatible with the POSIX 5084requires, and its I/O model is fundamentally incompatible with the POSIX
4601model. Libev still offers limited functionality on this platform in 5085model. Libev still offers limited functionality on this platform in
4602the form of the C<EVBACKEND_SELECT> backend, and only supports socket 5086the form of the C<EVBACKEND_SELECT> backend, and only supports socket
4603descriptors. This only applies when using Win32 natively, not when using 5087descriptors. This only applies when using Win32 natively, not when using
4604e.g. cygwin. Actually, it only applies to the microsofts own compilers, 5088e.g. cygwin. Actually, it only applies to the microsofts own compilers,
4605as every compielr comes with a slightly differently broken/incompatible 5089as every compiler comes with a slightly differently broken/incompatible
4606environment. 5090environment.
4607 5091
4608Lifting these limitations would basically require the full 5092Lifting these limitations would basically require the full
4609re-implementation of the I/O system. If you are into this kind of thing, 5093re-implementation of the I/O system. If you are into this kind of thing,
4610then note that glib does exactly that for you in a very portable way (note 5094then note that glib does exactly that for you in a very portable way (note
4704structure (guaranteed by POSIX but not by ISO C for example), but it also 5188structure (guaranteed by POSIX but not by ISO C for example), but it also
4705assumes that the same (machine) code can be used to call any watcher 5189assumes that the same (machine) code can be used to call any watcher
4706callback: The watcher callbacks have different type signatures, but libev 5190callback: The watcher callbacks have different type signatures, but libev
4707calls them using an C<ev_watcher *> internally. 5191calls them using an C<ev_watcher *> internally.
4708 5192
5193=item pointer accesses must be thread-atomic
5194
5195Accessing a pointer value must be atomic, it must both be readable and
5196writable in one piece - this is the case on all current architectures.
5197
4709=item C<sig_atomic_t volatile> must be thread-atomic as well 5198=item C<sig_atomic_t volatile> must be thread-atomic as well
4710 5199
4711The type C<sig_atomic_t volatile> (or whatever is defined as 5200The type C<sig_atomic_t volatile> (or whatever is defined as
4712C<EV_ATOMIC_T>) must be atomic with respect to accesses from different 5201C<EV_ATOMIC_T>) must be atomic with respect to accesses from different
4713threads. This is not part of the specification for C<sig_atomic_t>, but is 5202threads. This is not part of the specification for C<sig_atomic_t>, but is
4738 5227
4739The type C<double> is used to represent timestamps. It is required to 5228The type C<double> is used to represent timestamps. It is required to
4740have at least 51 bits of mantissa (and 9 bits of exponent), which is 5229have at least 51 bits of mantissa (and 9 bits of exponent), which is
4741good enough for at least into the year 4000 with millisecond accuracy 5230good enough for at least into the year 4000 with millisecond accuracy
4742(the design goal for libev). This requirement is overfulfilled by 5231(the design goal for libev). This requirement is overfulfilled by
4743implementations using IEEE 754, which is basically all existing ones. With 5232implementations using IEEE 754, which is basically all existing ones.
5233
4744IEEE 754 doubles, you get microsecond accuracy until at least 2200. 5234With IEEE 754 doubles, you get microsecond accuracy until at least the
5235year 2255 (and millisecond accuracy till the year 287396 - by then, libev
5236is either obsolete or somebody patched it to use C<long double> or
5237something like that, just kidding).
4745 5238
4746=back 5239=back
4747 5240
4748If you know of other additional requirements drop me a note. 5241If you know of other additional requirements drop me a note.
4749 5242
4811=item Processing ev_async_send: O(number_of_async_watchers) 5304=item Processing ev_async_send: O(number_of_async_watchers)
4812 5305
4813=item Processing signals: O(max_signal_number) 5306=item Processing signals: O(max_signal_number)
4814 5307
4815Sending involves a system call I<iff> there were no other C<ev_async_send> 5308Sending involves a system call I<iff> there were no other C<ev_async_send>
4816calls in the current loop iteration. Checking for async and signal events 5309calls in the current loop iteration and the loop is currently
5310blocked. Checking for async and signal events involves iterating over all
4817involves iterating over all running async watchers or all signal numbers. 5311running async watchers or all signal numbers.
4818 5312
4819=back 5313=back
4820 5314
4821 5315
4822=head1 PORTING FROM LIBEV 3.X TO 4.X 5316=head1 PORTING FROM LIBEV 3.X TO 4.X
4823 5317
4824The major version 4 introduced some minor incompatible changes to the API. 5318The major version 4 introduced some incompatible changes to the API.
4825 5319
4826At the moment, the C<ev.h> header file tries to implement superficial 5320At the moment, the C<ev.h> header file provides compatibility definitions
4827compatibility, so most programs should still compile. Those might be 5321for all changes, so most programs should still compile. The compatibility
4828removed in later versions of libev, so better update early than late. 5322layer might be removed in later versions of libev, so better update to the
5323new API early than late.
4829 5324
4830=over 4 5325=over 4
5326
5327=item C<EV_COMPAT3> backwards compatibility mechanism
5328
5329The backward compatibility mechanism can be controlled by
5330C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
5331section.
5332
5333=item C<ev_default_destroy> and C<ev_default_fork> have been removed
5334
5335These calls can be replaced easily by their C<ev_loop_xxx> counterparts:
5336
5337 ev_loop_destroy (EV_DEFAULT_UC);
5338 ev_loop_fork (EV_DEFAULT);
4831 5339
4832=item function/symbol renames 5340=item function/symbol renames
4833 5341
4834A number of functions and symbols have been renamed: 5342A number of functions and symbols have been renamed:
4835 5343
4854ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme 5362ev_loop> anymore and C<EV_TIMER> now follows the same naming scheme
4855as all other watcher types. Note that C<ev_loop_fork> is still called 5363as all other watcher types. Note that C<ev_loop_fork> is still called
4856C<ev_loop_fork> because it would otherwise clash with the C<ev_fork> 5364C<ev_loop_fork> because it would otherwise clash with the C<ev_fork>
4857typedef. 5365typedef.
4858 5366
4859=item C<EV_COMPAT3> backwards compatibility mechanism
4860
4861The backward compatibility mechanism can be controlled by
4862C<EV_COMPAT3>. See L<PREPROCESSOR SYMBOLS/MACROS> in the L<EMBEDDING>
4863section.
4864
4865=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES> 5367=item C<EV_MINIMAL> mechanism replaced by C<EV_FEATURES>
4866 5368
4867The preprocessor symbol C<EV_MINIMAL> has been replaced by a different 5369The preprocessor symbol C<EV_MINIMAL> has been replaced by a different
4868mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile 5370mechanism, C<EV_FEATURES>. Programs using C<EV_MINIMAL> usually compile
4869and work, but the library code will of course be larger. 5371and work, but the library code will of course be larger.
4931The physical time that is observed. It is apparently strictly monotonic :) 5433The physical time that is observed. It is apparently strictly monotonic :)
4932 5434
4933=item wall-clock time 5435=item wall-clock time
4934 5436
4935The time and date as shown on clocks. Unlike real time, it can actually 5437The time and date as shown on clocks. Unlike real time, it can actually
4936be wrong and jump forwards and backwards, e.g. when the you adjust your 5438be wrong and jump forwards and backwards, e.g. when you adjust your
4937clock. 5439clock.
4938 5440
4939=item watcher 5441=item watcher
4940 5442
4941A data structure that describes interest in certain events. Watchers need 5443A data structure that describes interest in certain events. Watchers need
4943 5445
4944=back 5446=back
4945 5447
4946=head1 AUTHOR 5448=head1 AUTHOR
4947 5449
4948Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael Magnusson. 5450Marc Lehmann <libev@schmorp.de>, with repeated corrections by Mikael
5451Magnusson and Emanuele Giaquinta, and minor corrections by many others.
4949 5452

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines