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

Comparing libev/ev.pod (file contents):
Revision 1.80 by root, Sun Dec 9 19:47:30 2007 UTC vs.
Revision 1.131 by root, Tue Feb 19 17:09:28 2008 UTC

… …
4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 #include <ev.h> 7 #include <ev.h>
8 8
9=head1 EXAMPLE PROGRAM 9=head2 EXAMPLE PROGRAM
10 10
11 #include <ev.h> 11 #include <ev.h>
12 12
13 ev_io stdin_watcher; 13 ev_io stdin_watcher;
14 ev_timer timeout_watcher; 14 ev_timer timeout_watcher;
… …
53The newest version of this document is also available as a html-formatted 53The newest version of this document is also available as a html-formatted
54web page you might find easier to navigate when reading it for the first 54web page you might find easier to navigate when reading it for the first
55time: L<http://cvs.schmorp.de/libev/ev.html>. 55time: L<http://cvs.schmorp.de/libev/ev.html>.
56 56
57Libev is an event loop: you register interest in certain events (such as a 57Libev is an event loop: you register interest in certain events (such as a
58file descriptor being readable or a timeout occuring), and it will manage 58file descriptor being readable or a timeout occurring), and it will manage
59these event sources and provide your program with events. 59these event sources and provide your program with events.
60 60
61To do this, it must take more or less complete control over your process 61To do this, it must take more or less complete control over your process
62(or thread) by executing the I<event loop> handler, and will then 62(or thread) by executing the I<event loop> handler, and will then
63communicate events via a callback mechanism. 63communicate events via a callback mechanism.
… …
65You register interest in certain events by registering so-called I<event 65You register interest in certain events by registering so-called I<event
66watchers>, which are relatively small C structures you initialise with the 66watchers>, which are relatively small C structures you initialise with the
67details of the event, and then hand it over to libev by I<starting> the 67details of the event, and then hand it over to libev by I<starting> the
68watcher. 68watcher.
69 69
70=head1 FEATURES 70=head2 FEATURES
71 71
72Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the 72Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
73BSD-specific C<kqueue> and the Solaris-specific event port mechanisms 73BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
74for file descriptor events (C<ev_io>), the Linux C<inotify> interface 74for file descriptor events (C<ev_io>), the Linux C<inotify> interface
75(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers 75(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers
… …
82 82
83It also is quite fast (see this 83It also is quite fast (see this
84L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent 84L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
85for example). 85for example).
86 86
87=head1 CONVENTIONS 87=head2 CONVENTIONS
88 88
89Libev is very configurable. In this manual the default configuration will 89Libev is very configurable. In this manual the default configuration will
90be described, which supports multiple event loops. For more info about 90be described, which supports multiple event loops. For more info about
91various configuration options please have a look at B<EMBED> section in 91various configuration options please have a look at B<EMBED> section in
92this manual. If libev was configured without support for multiple event 92this manual. If libev was configured without support for multiple event
93loops, then all functions taking an initial argument of name C<loop> 93loops, then all functions taking an initial argument of name C<loop>
94(which is always of type C<struct ev_loop *>) will not have this argument. 94(which is always of type C<struct ev_loop *>) will not have this argument.
95 95
96=head1 TIME REPRESENTATION 96=head2 TIME REPRESENTATION
97 97
98Libev represents time as a single floating point number, representing the 98Libev represents time as a single floating point number, representing the
99(fractional) number of seconds since the (POSIX) epoch (somewhere near 99(fractional) number of seconds since the (POSIX) epoch (somewhere near
100the beginning of 1970, details are complicated, don't ask). This type is 100the beginning of 1970, details are complicated, don't ask). This type is
101called C<ev_tstamp>, which is what you should use too. It usually aliases 101called C<ev_tstamp>, which is what you should use too. It usually aliases
102to the C<double> type in C, and when you need to do any calculations on 102to the C<double> type in C, and when you need to do any calculations on
103it, you should treat it as such. 103it, you should treat it as some floatingpoint value. Unlike the name
104component C<stamp> might indicate, it is also used for time differences
105throughout libev.
104 106
105=head1 GLOBAL FUNCTIONS 107=head1 GLOBAL FUNCTIONS
106 108
107These functions can be called anytime, even before initialising the 109These functions can be called anytime, even before initialising the
108library in any way. 110library in any way.
… …
112=item ev_tstamp ev_time () 114=item ev_tstamp ev_time ()
113 115
114Returns the current time as libev would use it. Please note that the 116Returns the current time as libev would use it. Please note that the
115C<ev_now> function is usually faster and also often returns the timestamp 117C<ev_now> function is usually faster and also often returns the timestamp
116you actually want to know. 118you actually want to know.
119
120=item ev_sleep (ev_tstamp interval)
121
122Sleep for the given interval: The current thread will be blocked until
123either it is interrupted or the given time interval has passed. Basically
124this is a subsecond-resolution C<sleep ()>.
117 125
118=item int ev_version_major () 126=item int ev_version_major ()
119 127
120=item int ev_version_minor () 128=item int ev_version_minor ()
121 129
… …
252flags. If that is troubling you, check C<ev_backend ()> afterwards). 260flags. If that is troubling you, check C<ev_backend ()> afterwards).
253 261
254If you don't know what event loop to use, use the one returned from this 262If you don't know what event loop to use, use the one returned from this
255function. 263function.
256 264
265The default loop is the only loop that can handle C<ev_signal> and
266C<ev_child> watchers, and to do this, it always registers a handler
267for C<SIGCHLD>. If this is a problem for your app you can either
268create a dynamic loop with C<ev_loop_new> that doesn't do that, or you
269can simply overwrite the C<SIGCHLD> signal handler I<after> calling
270C<ev_default_init>.
271
257The flags argument can be used to specify special behaviour or specific 272The flags argument can be used to specify special behaviour or specific
258backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>). 273backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
259 274
260The following flags are supported: 275The following flags are supported:
261 276
… …
298=item C<EVBACKEND_SELECT> (value 1, portable select backend) 313=item C<EVBACKEND_SELECT> (value 1, portable select backend)
299 314
300This is your standard select(2) backend. Not I<completely> standard, as 315This is your standard select(2) backend. Not I<completely> standard, as
301libev tries to roll its own fd_set with no limits on the number of fds, 316libev tries to roll its own fd_set with no limits on the number of fds,
302but if that fails, expect a fairly low limit on the number of fds when 317but if that fails, expect a fairly low limit on the number of fds when
303using this backend. It doesn't scale too well (O(highest_fd)), but its usually 318using this backend. It doesn't scale too well (O(highest_fd)), but its
304the fastest backend for a low number of fds. 319usually the fastest backend for a low number of (low-numbered :) fds.
320
321To get good performance out of this backend you need a high amount of
322parallelity (most of the file descriptors should be busy). If you are
323writing a server, you should C<accept ()> in a loop to accept as many
324connections as possible during one iteration. You might also want to have
325a look at C<ev_set_io_collect_interval ()> to increase the amount of
326readyness notifications you get per iteration.
305 327
306=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows) 328=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
307 329
308And this is your standard poll(2) backend. It's more complicated than 330And this is your standard poll(2) backend. It's more complicated
309select, but handles sparse fds better and has no artificial limit on the 331than select, but handles sparse fds better and has no artificial
310number of fds you can use (except it will slow down considerably with a 332limit on the number of fds you can use (except it will slow down
311lot of inactive fds). It scales similarly to select, i.e. O(total_fds). 333considerably with a lot of inactive fds). It scales similarly to select,
334i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
335performance tips.
312 336
313=item C<EVBACKEND_EPOLL> (value 4, Linux) 337=item C<EVBACKEND_EPOLL> (value 4, Linux)
314 338
315For few fds, this backend is a bit little slower than poll and select, 339For few fds, this backend is a bit little slower than poll and select,
316but it scales phenomenally better. While poll and select usually scale like 340but it scales phenomenally better. While poll and select usually scale
317O(total_fds) where n is the total number of fds (or the highest fd), epoll scales 341like O(total_fds) where n is the total number of fds (or the highest fd),
318either O(1) or O(active_fds). 342epoll scales either O(1) or O(active_fds). The epoll design has a number
343of shortcomings, such as silently dropping events in some hard-to-detect
344cases and rewiring a syscall per fd change, no fork support and bad
345support for dup.
319 346
320While stopping and starting an I/O watcher in the same iteration will 347While stopping, setting and starting an I/O watcher in the same iteration
321result in some caching, there is still a syscall per such incident 348will result in some caching, there is still a syscall per such incident
322(because the fd could point to a different file description now), so its 349(because the fd could point to a different file description now), so its
323best to avoid that. Also, dup()ed file descriptors might not work very 350best to avoid that. Also, C<dup ()>'ed file descriptors might not work
324well if you register events for both fds. 351very well if you register events for both fds.
325 352
326Please note that epoll sometimes generates spurious notifications, so you 353Please note that epoll sometimes generates spurious notifications, so you
327need to use non-blocking I/O or other means to avoid blocking when no data 354need to use non-blocking I/O or other means to avoid blocking when no data
328(or space) is available. 355(or space) is available.
329 356
357Best performance from this backend is achieved by not unregistering all
358watchers for a file descriptor until it has been closed, if possible, i.e.
359keep at least one watcher active per fd at all times.
360
361While nominally embeddeble in other event loops, this feature is broken in
362all kernel versions tested so far.
363
330=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones) 364=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
331 365
332Kqueue deserves special mention, as at the time of this writing, it 366Kqueue deserves special mention, as at the time of this writing, it
333was broken on all BSDs except NetBSD (usually it doesn't work with 367was broken on all BSDs except NetBSD (usually it doesn't work reliably
334anything but sockets and pipes, except on Darwin, where of course its 368with anything but sockets and pipes, except on Darwin, where of course
335completely useless). For this reason its not being "autodetected" 369it's completely useless). For this reason it's not being "autodetected"
336unless you explicitly specify it explicitly in the flags (i.e. using 370unless you explicitly specify it explicitly in the flags (i.e. using
337C<EVBACKEND_KQUEUE>). 371C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
372system like NetBSD.
373
374You still can embed kqueue into a normal poll or select backend and use it
375only for sockets (after having made sure that sockets work with kqueue on
376the target platform). See C<ev_embed> watchers for more info.
338 377
339It scales in the same way as the epoll backend, but the interface to the 378It scales in the same way as the epoll backend, but the interface to the
340kernel is more efficient (which says nothing about its actual speed, of 379kernel is more efficient (which says nothing about its actual speed, of
341course). While starting and stopping an I/O watcher does not cause an 380course). While stopping, setting and starting an I/O watcher does never
342extra syscall as with epoll, it still adds up to four event changes per 381cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to
343incident, so its best to avoid that. 382two event changes per incident, support for C<fork ()> is very bad and it
383drops fds silently in similarly hard-to-detect cases.
384
385This backend usually performs well under most conditions.
386
387While nominally embeddable in other event loops, this doesn't work
388everywhere, so you might need to test for this. And since it is broken
389almost everywhere, you should only use it when you have a lot of sockets
390(for which it usually works), by embedding it into another event loop
391(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and using it only for
392sockets.
344 393
345=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8) 394=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
346 395
347This is not implemented yet (and might never be). 396This is not implemented yet (and might never be, unless you send me an
397implementation). According to reports, C</dev/poll> only supports sockets
398and is not embeddable, which would limit the usefulness of this backend
399immensely.
348 400
349=item C<EVBACKEND_PORT> (value 32, Solaris 10) 401=item C<EVBACKEND_PORT> (value 32, Solaris 10)
350 402
351This uses the Solaris 10 port mechanism. As with everything on Solaris, 403This uses the Solaris 10 event port mechanism. As with everything on Solaris,
352it's really slow, but it still scales very well (O(active_fds)). 404it's really slow, but it still scales very well (O(active_fds)).
353 405
354Please note that solaris ports can result in a lot of spurious 406Please note that solaris event ports can deliver a lot of spurious
355notifications, so you need to use non-blocking I/O or other means to avoid 407notifications, so you need to use non-blocking I/O or other means to avoid
356blocking when no data (or space) is available. 408blocking when no data (or space) is available.
409
410While this backend scales well, it requires one system call per active
411file descriptor per loop iteration. For small and medium numbers of file
412descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
413might perform better.
414
415On the positive side, ignoring the spurious readyness notifications, this
416backend actually performed to specification in all tests and is fully
417embeddable, which is a rare feat among the OS-specific backends.
357 418
358=item C<EVBACKEND_ALL> 419=item C<EVBACKEND_ALL>
359 420
360Try all backends (even potentially broken ones that wouldn't be tried 421Try all backends (even potentially broken ones that wouldn't be tried
361with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as 422with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
362C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>. 423C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
363 424
425It is definitely not recommended to use this flag.
426
364=back 427=back
365 428
366If one or more of these are ored into the flags value, then only these 429If one or more of these are ored into the flags value, then only these
367backends will be tried (in the reverse order as given here). If none are 430backends will be tried (in the reverse order as listed here). If none are
368specified, most compiled-in backend will be tried, usually in reverse 431specified, all backends in C<ev_recommended_backends ()> will be tried.
369order of their flag values :)
370 432
371The most typical usage is like this: 433The most typical usage is like this:
372 434
373 if (!ev_default_loop (0)) 435 if (!ev_default_loop (0))
374 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?"); 436 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
… …
402Destroys the default loop again (frees all memory and kernel state 464Destroys the default loop again (frees all memory and kernel state
403etc.). None of the active event watchers will be stopped in the normal 465etc.). None of the active event watchers will be stopped in the normal
404sense, so e.g. C<ev_is_active> might still return true. It is your 466sense, so e.g. C<ev_is_active> might still return true. It is your
405responsibility to either stop all watchers cleanly yoursef I<before> 467responsibility to either stop all watchers cleanly yoursef I<before>
406calling this function, or cope with the fact afterwards (which is usually 468calling this function, or cope with the fact afterwards (which is usually
407the easiest thing, youc na just ignore the watchers and/or C<free ()> them 469the easiest thing, you can just ignore the watchers and/or C<free ()> them
408for example). 470for example).
471
472Note that certain global state, such as signal state, will not be freed by
473this function, and related watchers (such as signal and child watchers)
474would need to be stopped manually.
475
476In general it is not advisable to call this function except in the
477rare occasion where you really need to free e.g. the signal handling
478pipe fds. If you need dynamically allocated loops it is better to use
479C<ev_loop_new> and C<ev_loop_destroy>).
409 480
410=item ev_loop_destroy (loop) 481=item ev_loop_destroy (loop)
411 482
412Like C<ev_default_destroy>, but destroys an event loop created by an 483Like C<ev_default_destroy>, but destroys an event loop created by an
413earlier call to C<ev_loop_new>. 484earlier call to C<ev_loop_new>.
414 485
415=item ev_default_fork () 486=item ev_default_fork ()
416 487
488This function sets a flag that causes subsequent C<ev_loop> iterations
417This function reinitialises the kernel state for backends that have 489to reinitialise the kernel state for backends that have one. Despite the
418one. Despite the name, you can call it anytime, but it makes most sense 490name, you can call it anytime, but it makes most sense after forking, in
419after forking, in either the parent or child process (or both, but that 491the child process (or both child and parent, but that again makes little
420again makes little sense). 492sense). You I<must> call it in the child before using any of the libev
493functions, and it will only take effect at the next C<ev_loop> iteration.
421 494
422You I<must> call this function in the child process after forking if and 495On the other hand, you only need to call this function in the child
423only if you want to use the event library in both processes. If you just 496process if and only if you want to use the event library in the child. If
424fork+exec, you don't have to call it. 497you just fork+exec, you don't have to call it at all.
425 498
426The function itself is quite fast and it's usually not a problem to call 499The function itself is quite fast and it's usually not a problem to call
427it just in case after a fork. To make this easy, the function will fit in 500it just in case after a fork. To make this easy, the function will fit in
428quite nicely into a call to C<pthread_atfork>: 501quite nicely into a call to C<pthread_atfork>:
429 502
430 pthread_atfork (0, 0, ev_default_fork); 503 pthread_atfork (0, 0, ev_default_fork);
431 504
432At the moment, C<EVBACKEND_SELECT> and C<EVBACKEND_POLL> are safe to use
433without calling this function, so if you force one of those backends you
434do not need to care.
435
436=item ev_loop_fork (loop) 505=item ev_loop_fork (loop)
437 506
438Like C<ev_default_fork>, but acts on an event loop created by 507Like C<ev_default_fork>, but acts on an event loop created by
439C<ev_loop_new>. Yes, you have to call this on every allocated event loop 508C<ev_loop_new>. Yes, you have to call this on every allocated event loop
440after fork, and how you do this is entirely your own problem. 509after fork, and how you do this is entirely your own problem.
510
511=item int ev_is_default_loop (loop)
512
513Returns true when the given loop actually is the default loop, false otherwise.
441 514
442=item unsigned int ev_loop_count (loop) 515=item unsigned int ev_loop_count (loop)
443 516
444Returns the count of loop iterations for the loop, which is identical to 517Returns the count of loop iterations for the loop, which is identical to
445the number of times libev did poll for new events. It starts at C<0> and 518the number of times libev did poll for new events. It starts at C<0> and
… …
458 531
459Returns the current "event loop time", which is the time the event loop 532Returns the current "event loop time", which is the time the event loop
460received events and started processing them. This timestamp does not 533received events and started processing them. This timestamp does not
461change as long as callbacks are being processed, and this is also the base 534change as long as callbacks are being processed, and this is also the base
462time used for relative timers. You can treat it as the timestamp of the 535time used for relative timers. You can treat it as the timestamp of the
463event occuring (or more correctly, libev finding out about it). 536event occurring (or more correctly, libev finding out about it).
464 537
465=item ev_loop (loop, int flags) 538=item ev_loop (loop, int flags)
466 539
467Finally, this is it, the event handler. This function usually is called 540Finally, this is it, the event handler. This function usually is called
468after you initialised all your watchers and you want to start handling 541after you initialised all your watchers and you want to start handling
… …
490usually a better approach for this kind of thing. 563usually a better approach for this kind of thing.
491 564
492Here are the gory details of what C<ev_loop> does: 565Here are the gory details of what C<ev_loop> does:
493 566
494 - Before the first iteration, call any pending watchers. 567 - Before the first iteration, call any pending watchers.
495 * If there are no active watchers (reference count is zero), return. 568 * If EVFLAG_FORKCHECK was used, check for a fork.
496 - Queue all prepare watchers and then call all outstanding watchers. 569 - If a fork was detected, queue and call all fork watchers.
570 - Queue and call all prepare watchers.
497 - If we have been forked, recreate the kernel state. 571 - If we have been forked, recreate the kernel state.
498 - Update the kernel state with all outstanding changes. 572 - Update the kernel state with all outstanding changes.
499 - Update the "event loop time". 573 - Update the "event loop time".
500 - Calculate for how long to block. 574 - Calculate for how long to sleep or block, if at all
575 (active idle watchers, EVLOOP_NONBLOCK or not having
576 any active watchers at all will result in not sleeping).
577 - Sleep if the I/O and timer collect interval say so.
501 - Block the process, waiting for any events. 578 - Block the process, waiting for any events.
502 - Queue all outstanding I/O (fd) events. 579 - Queue all outstanding I/O (fd) events.
503 - Update the "event loop time" and do time jump handling. 580 - Update the "event loop time" and do time jump handling.
504 - Queue all outstanding timers. 581 - Queue all outstanding timers.
505 - Queue all outstanding periodics. 582 - Queue all outstanding periodics.
506 - If no events are pending now, queue all idle watchers. 583 - If no events are pending now, queue all idle watchers.
507 - Queue all check watchers. 584 - Queue all check watchers.
508 - Call all queued watchers in reverse order (i.e. check watchers first). 585 - Call all queued watchers in reverse order (i.e. check watchers first).
509 Signals and child watchers are implemented as I/O watchers, and will 586 Signals and child watchers are implemented as I/O watchers, and will
510 be handled here by queueing them when their watcher gets executed. 587 be handled here by queueing them when their watcher gets executed.
511 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK 588 - If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
512 were used, return, otherwise continue with step *. 589 were used, or there are no active watchers, return, otherwise
590 continue with step *.
513 591
514Example: Queue some jobs and then loop until no events are outsanding 592Example: Queue some jobs and then loop until no events are outstanding
515anymore. 593anymore.
516 594
517 ... queue jobs here, make sure they register event watchers as long 595 ... queue jobs here, make sure they register event watchers as long
518 ... as they still have work to do (even an idle watcher will do..) 596 ... as they still have work to do (even an idle watcher will do..)
519 ev_loop (my_loop, 0); 597 ev_loop (my_loop, 0);
… …
523 601
524Can be used to make a call to C<ev_loop> return early (but only after it 602Can be used to make a call to C<ev_loop> return early (but only after it
525has processed all outstanding events). The C<how> argument must be either 603has processed all outstanding events). The C<how> argument must be either
526C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or 604C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
527C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return. 605C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
606
607This "unloop state" will be cleared when entering C<ev_loop> again.
528 608
529=item ev_ref (loop) 609=item ev_ref (loop)
530 610
531=item ev_unref (loop) 611=item ev_unref (loop)
532 612
… …
537returning, ev_unref() after starting, and ev_ref() before stopping it. For 617returning, ev_unref() after starting, and ev_ref() before stopping it. For
538example, libev itself uses this for its internal signal pipe: It is not 618example, libev itself uses this for its internal signal pipe: It is not
539visible to the libev user and should not keep C<ev_loop> from exiting if 619visible to the libev user and should not keep C<ev_loop> from exiting if
540no event watchers registered by it are active. It is also an excellent 620no event watchers registered by it are active. It is also an excellent
541way to do this for generic recurring timers or from within third-party 621way to do this for generic recurring timers or from within third-party
542libraries. Just remember to I<unref after start> and I<ref before stop>. 622libraries. Just remember to I<unref after start> and I<ref before stop>
623(but only if the watcher wasn't active before, or was active before,
624respectively).
543 625
544Example: Create a signal watcher, but keep it from keeping C<ev_loop> 626Example: Create a signal watcher, but keep it from keeping C<ev_loop>
545running when nothing else is active. 627running when nothing else is active.
546 628
547 struct ev_signal exitsig; 629 struct ev_signal exitsig;
… …
551 633
552Example: For some weird reason, unregister the above signal handler again. 634Example: For some weird reason, unregister the above signal handler again.
553 635
554 ev_ref (loop); 636 ev_ref (loop);
555 ev_signal_stop (loop, &exitsig); 637 ev_signal_stop (loop, &exitsig);
638
639=item ev_set_io_collect_interval (loop, ev_tstamp interval)
640
641=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
642
643These advanced functions influence the time that libev will spend waiting
644for events. Both are by default C<0>, meaning that libev will try to
645invoke timer/periodic callbacks and I/O callbacks with minimum latency.
646
647Setting these to a higher value (the C<interval> I<must> be >= C<0>)
648allows libev to delay invocation of I/O and timer/periodic callbacks to
649increase efficiency of loop iterations.
650
651The background is that sometimes your program runs just fast enough to
652handle one (or very few) event(s) per loop iteration. While this makes
653the program responsive, it also wastes a lot of CPU time to poll for new
654events, especially with backends like C<select ()> which have a high
655overhead for the actual polling but can deliver many events at once.
656
657By setting a higher I<io collect interval> you allow libev to spend more
658time collecting I/O events, so you can handle more events per iteration,
659at the cost of increasing latency. Timeouts (both C<ev_periodic> and
660C<ev_timer>) will be not affected. Setting this to a non-null value will
661introduce an additional C<ev_sleep ()> call into most loop iterations.
662
663Likewise, by setting a higher I<timeout collect interval> you allow libev
664to spend more time collecting timeouts, at the expense of increased
665latency (the watcher callback will be called later). C<ev_io> watchers
666will not be affected. Setting this to a non-null value will not introduce
667any overhead in libev.
668
669Many (busy) programs can usually benefit by setting the io collect
670interval to a value near C<0.1> or so, which is often enough for
671interactive servers (of course not for games), likewise for timeouts. It
672usually doesn't make much sense to set it to a lower value than C<0.01>,
673as this approsaches the timing granularity of most systems.
556 674
557=back 675=back
558 676
559 677
560=head1 ANATOMY OF A WATCHER 678=head1 ANATOMY OF A WATCHER
… …
659 777
660=item C<EV_FORK> 778=item C<EV_FORK>
661 779
662The event loop has been resumed in the child process after fork (see 780The event loop has been resumed in the child process after fork (see
663C<ev_fork>). 781C<ev_fork>).
782
783=item C<EV_ASYNC>
784
785The given async watcher has been asynchronously notified (see C<ev_async>).
664 786
665=item C<EV_ERROR> 787=item C<EV_ERROR>
666 788
667An unspecified error has occured, the watcher has been stopped. This might 789An unspecified error has occured, the watcher has been stopped. This might
668happen because the watcher could not be properly started because libev 790happen because the watcher could not be properly started because libev
… …
886In general you can register as many read and/or write event watchers per 1008In general you can register as many read and/or write event watchers per
887fd as you want (as long as you don't confuse yourself). Setting all file 1009fd as you want (as long as you don't confuse yourself). Setting all file
888descriptors to non-blocking mode is also usually a good idea (but not 1010descriptors to non-blocking mode is also usually a good idea (but not
889required if you know what you are doing). 1011required if you know what you are doing).
890 1012
891You have to be careful with dup'ed file descriptors, though. Some backends
892(the linux epoll backend is a notable example) cannot handle dup'ed file
893descriptors correctly if you register interest in two or more fds pointing
894to the same underlying file/socket/etc. description (that is, they share
895the same underlying "file open").
896
897If you must do this, then force the use of a known-to-be-good backend 1013If you must do this, then force the use of a known-to-be-good backend
898(at the time of this writing, this includes only C<EVBACKEND_SELECT> and 1014(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
899C<EVBACKEND_POLL>). 1015C<EVBACKEND_POLL>).
900 1016
901Another thing you have to watch out for is that it is quite easy to 1017Another thing you have to watch out for is that it is quite easy to
… …
911play around with an Xlib connection), then you have to seperately re-test 1027play around with an Xlib connection), then you have to seperately re-test
912whether a file descriptor is really ready with a known-to-be good interface 1028whether a file descriptor is really ready with a known-to-be good interface
913such as poll (fortunately in our Xlib example, Xlib already does this on 1029such as poll (fortunately in our Xlib example, Xlib already does this on
914its own, so its quite safe to use). 1030its own, so its quite safe to use).
915 1031
1032=head3 The special problem of disappearing file descriptors
1033
1034Some backends (e.g. kqueue, epoll) need to be told about closing a file
1035descriptor (either by calling C<close> explicitly or by any other means,
1036such as C<dup>). The reason is that you register interest in some file
1037descriptor, but when it goes away, the operating system will silently drop
1038this interest. If another file descriptor with the same number then is
1039registered with libev, there is no efficient way to see that this is, in
1040fact, a different file descriptor.
1041
1042To avoid having to explicitly tell libev about such cases, libev follows
1043the following policy: Each time C<ev_io_set> is being called, libev
1044will assume that this is potentially a new file descriptor, otherwise
1045it is assumed that the file descriptor stays the same. That means that
1046you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
1047descriptor even if the file descriptor number itself did not change.
1048
1049This is how one would do it normally anyway, the important point is that
1050the libev application should not optimise around libev but should leave
1051optimisations to libev.
1052
1053=head3 The special problem of dup'ed file descriptors
1054
1055Some backends (e.g. epoll), cannot register events for file descriptors,
1056but only events for the underlying file descriptions. That means when you
1057have C<dup ()>'ed file descriptors or weirder constellations, and register
1058events for them, only one file descriptor might actually receive events.
1059
1060There is no workaround possible except not registering events
1061for potentially C<dup ()>'ed file descriptors, or to resort to
1062C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1063
1064=head3 The special problem of fork
1065
1066Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
1067useless behaviour. Libev fully supports fork, but needs to be told about
1068it in the child.
1069
1070To support fork in your programs, you either have to call
1071C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
1072enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
1073C<EVBACKEND_POLL>.
1074
1075
1076=head3 Watcher-Specific Functions
1077
916=over 4 1078=over 4
917 1079
918=item ev_io_init (ev_io *, callback, int fd, int events) 1080=item ev_io_init (ev_io *, callback, int fd, int events)
919 1081
920=item ev_io_set (ev_io *, int fd, int events) 1082=item ev_io_set (ev_io *, int fd, int events)
… …
930=item int events [read-only] 1092=item int events [read-only]
931 1093
932The events being watched. 1094The events being watched.
933 1095
934=back 1096=back
1097
1098=head3 Examples
935 1099
936Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well 1100Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
937readable, but only once. Since it is likely line-buffered, you could 1101readable, but only once. Since it is likely line-buffered, you could
938attempt to read a whole line in the callback. 1102attempt to read a whole line in the callback.
939 1103
… …
972 ev_timer_set (&timer, after + ev_now () - ev_time (), 0.); 1136 ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
973 1137
974The callback is guarenteed to be invoked only when its timeout has passed, 1138The callback is guarenteed to be invoked only when its timeout has passed,
975but if multiple timers become ready during the same loop iteration then 1139but if multiple timers become ready during the same loop iteration then
976order of execution is undefined. 1140order of execution is undefined.
1141
1142=head3 Watcher-Specific Functions and Data Members
977 1143
978=over 4 1144=over 4
979 1145
980=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat) 1146=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
981 1147
… …
1035or C<ev_timer_again> is called and determines the next timeout (if any), 1201or C<ev_timer_again> is called and determines the next timeout (if any),
1036which is also when any modifications are taken into account. 1202which is also when any modifications are taken into account.
1037 1203
1038=back 1204=back
1039 1205
1206=head3 Examples
1207
1040Example: Create a timer that fires after 60 seconds. 1208Example: Create a timer that fires after 60 seconds.
1041 1209
1042 static void 1210 static void
1043 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents) 1211 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1044 { 1212 {
… …
1087 1255
1088As with timers, the callback is guarenteed to be invoked only when the 1256As with timers, the callback is guarenteed to be invoked only when the
1089time (C<at>) has been passed, but if multiple periodic timers become ready 1257time (C<at>) has been passed, but if multiple periodic timers become ready
1090during the same loop iteration then order of execution is undefined. 1258during the same loop iteration then order of execution is undefined.
1091 1259
1260=head3 Watcher-Specific Functions and Data Members
1261
1092=over 4 1262=over 4
1093 1263
1094=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb) 1264=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)
1095 1265
1096=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb) 1266=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)
… …
1192 1362
1193The current reschedule callback, or C<0>, if this functionality is 1363The current reschedule callback, or C<0>, if this functionality is
1194switched off. Can be changed any time, but changes only take effect when 1364switched off. Can be changed any time, but changes only take effect when
1195the periodic timer fires or C<ev_periodic_again> is being called. 1365the periodic timer fires or C<ev_periodic_again> is being called.
1196 1366
1367=item ev_tstamp at [read-only]
1368
1369When active, contains the absolute time that the watcher is supposed to
1370trigger next.
1371
1197=back 1372=back
1373
1374=head3 Examples
1198 1375
1199Example: Call a callback every hour, or, more precisely, whenever the 1376Example: Call a callback every hour, or, more precisely, whenever the
1200system clock is divisible by 3600. The callback invocation times have 1377system clock is divisible by 3600. The callback invocation times have
1201potentially a lot of jittering, but good long-term stability. 1378potentially a lot of jittering, but good long-term stability.
1202 1379
… …
1242with the kernel (thus it coexists with your own signal handlers as long 1419with the kernel (thus it coexists with your own signal handlers as long
1243as you don't register any with libev). Similarly, when the last signal 1420as you don't register any with libev). Similarly, when the last signal
1244watcher for a signal is stopped libev will reset the signal handler to 1421watcher for a signal is stopped libev will reset the signal handler to
1245SIG_DFL (regardless of what it was set to before). 1422SIG_DFL (regardless of what it was set to before).
1246 1423
1424=head3 Watcher-Specific Functions and Data Members
1425
1247=over 4 1426=over 4
1248 1427
1249=item ev_signal_init (ev_signal *, callback, int signum) 1428=item ev_signal_init (ev_signal *, callback, int signum)
1250 1429
1251=item ev_signal_set (ev_signal *, int signum) 1430=item ev_signal_set (ev_signal *, int signum)
… …
1263=head2 C<ev_child> - watch out for process status changes 1442=head2 C<ev_child> - watch out for process status changes
1264 1443
1265Child watchers trigger when your process receives a SIGCHLD in response to 1444Child watchers trigger when your process receives a SIGCHLD in response to
1266some child status changes (most typically when a child of yours dies). 1445some child status changes (most typically when a child of yours dies).
1267 1446
1447=head3 Watcher-Specific Functions and Data Members
1448
1268=over 4 1449=over 4
1269 1450
1270=item ev_child_init (ev_child *, callback, int pid) 1451=item ev_child_init (ev_child *, callback, int pid, int trace)
1271 1452
1272=item ev_child_set (ev_child *, int pid) 1453=item ev_child_set (ev_child *, int pid, int trace)
1273 1454
1274Configures the watcher to wait for status changes of process C<pid> (or 1455Configures the watcher to wait for status changes of process C<pid> (or
1275I<any> process if C<pid> is specified as C<0>). The callback can look 1456I<any> process if C<pid> is specified as C<0>). The callback can look
1276at the C<rstatus> member of the C<ev_child> watcher structure to see 1457at the C<rstatus> member of the C<ev_child> watcher structure to see
1277the status word (use the macros from C<sys/wait.h> and see your systems 1458the status word (use the macros from C<sys/wait.h> and see your systems
1278C<waitpid> documentation). The C<rpid> member contains the pid of the 1459C<waitpid> documentation). The C<rpid> member contains the pid of the
1279process causing the status change. 1460process causing the status change. C<trace> must be either C<0> (only
1461activate the watcher when the process terminates) or C<1> (additionally
1462activate the watcher when the process is stopped or continued).
1280 1463
1281=item int pid [read-only] 1464=item int pid [read-only]
1282 1465
1283The process id this watcher watches out for, or C<0>, meaning any process id. 1466The process id this watcher watches out for, or C<0>, meaning any process id.
1284 1467
… …
1290 1473
1291The process exit/trace status caused by C<rpid> (see your systems 1474The process exit/trace status caused by C<rpid> (see your systems
1292C<waitpid> and C<sys/wait.h> documentation for details). 1475C<waitpid> and C<sys/wait.h> documentation for details).
1293 1476
1294=back 1477=back
1478
1479=head3 Examples
1295 1480
1296Example: Try to exit cleanly on SIGINT and SIGTERM. 1481Example: Try to exit cleanly on SIGINT and SIGTERM.
1297 1482
1298 static void 1483 static void
1299 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents) 1484 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
… …
1340semantics of C<ev_stat> watchers, which means that libev sometimes needs 1525semantics of C<ev_stat> watchers, which means that libev sometimes needs
1341to fall back to regular polling again even with inotify, but changes are 1526to fall back to regular polling again even with inotify, but changes are
1342usually detected immediately, and if the file exists there will be no 1527usually detected immediately, and if the file exists there will be no
1343polling. 1528polling.
1344 1529
1530=head3 Inotify
1531
1532When C<inotify (7)> support has been compiled into libev (generally only
1533available on Linux) and present at runtime, it will be used to speed up
1534change detection where possible. The inotify descriptor will be created lazily
1535when the first C<ev_stat> watcher is being started.
1536
1537Inotify presense does not change the semantics of C<ev_stat> watchers
1538except that changes might be detected earlier, and in some cases, to avoid
1539making regular C<stat> calls. Even in the presense of inotify support
1540there are many cases where libev has to resort to regular C<stat> polling.
1541
1542(There is no support for kqueue, as apparently it cannot be used to
1543implement this functionality, due to the requirement of having a file
1544descriptor open on the object at all times).
1545
1546=head3 The special problem of stat time resolution
1547
1548The C<stat ()> syscall only supports full-second resolution portably, and
1549even on systems where the resolution is higher, many filesystems still
1550only support whole seconds.
1551
1552That means that, if the time is the only thing that changes, you might
1553miss updates: on the first update, C<ev_stat> detects a change and calls
1554your callback, which does something. When there is another update within
1555the same second, C<ev_stat> will be unable to detect it.
1556
1557The solution to this is to delay acting on a change for a second (or till
1558the next second boundary), using a roughly one-second delay C<ev_timer>
1559(C<ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)>). The C<.01>
1560is added to work around small timing inconsistencies of some operating
1561systems.
1562
1563=head3 Watcher-Specific Functions and Data Members
1564
1345=over 4 1565=over 4
1346 1566
1347=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval) 1567=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
1348 1568
1349=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval) 1569=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
… …
1384=item const char *path [read-only] 1604=item const char *path [read-only]
1385 1605
1386The filesystem path that is being watched. 1606The filesystem path that is being watched.
1387 1607
1388=back 1608=back
1609
1610=head3 Examples
1389 1611
1390Example: Watch C</etc/passwd> for attribute changes. 1612Example: Watch C</etc/passwd> for attribute changes.
1391 1613
1392 static void 1614 static void
1393 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents) 1615 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
… …
1406 } 1628 }
1407 1629
1408 ... 1630 ...
1409 ev_stat passwd; 1631 ev_stat passwd;
1410 1632
1411 ev_stat_init (&passwd, passwd_cb, "/etc/passwd"); 1633 ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1412 ev_stat_start (loop, &passwd); 1634 ev_stat_start (loop, &passwd);
1635
1636Example: Like above, but additionally use a one-second delay so we do not
1637miss updates (however, frequent updates will delay processing, too, so
1638one might do the work both on C<ev_stat> callback invocation I<and> on
1639C<ev_timer> callback invocation).
1640
1641 static ev_stat passwd;
1642 static ev_timer timer;
1643
1644 static void
1645 timer_cb (EV_P_ ev_timer *w, int revents)
1646 {
1647 ev_timer_stop (EV_A_ w);
1648
1649 /* now it's one second after the most recent passwd change */
1650 }
1651
1652 static void
1653 stat_cb (EV_P_ ev_stat *w, int revents)
1654 {
1655 /* reset the one-second timer */
1656 ev_timer_again (EV_A_ &timer);
1657 }
1658
1659 ...
1660 ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
1661 ev_stat_start (loop, &passwd);
1662 ev_timer_init (&timer, timer_cb, 0., 1.01);
1413 1663
1414 1664
1415=head2 C<ev_idle> - when you've got nothing better to do... 1665=head2 C<ev_idle> - when you've got nothing better to do...
1416 1666
1417Idle watchers trigger events when no other events of the same or higher 1667Idle watchers trigger events when no other events of the same or higher
… …
1431Apart from keeping your process non-blocking (which is a useful 1681Apart from keeping your process non-blocking (which is a useful
1432effect on its own sometimes), idle watchers are a good place to do 1682effect on its own sometimes), idle watchers are a good place to do
1433"pseudo-background processing", or delay processing stuff to after the 1683"pseudo-background processing", or delay processing stuff to after the
1434event loop has handled all outstanding events. 1684event loop has handled all outstanding events.
1435 1685
1686=head3 Watcher-Specific Functions and Data Members
1687
1436=over 4 1688=over 4
1437 1689
1438=item ev_idle_init (ev_signal *, callback) 1690=item ev_idle_init (ev_signal *, callback)
1439 1691
1440Initialises and configures the idle watcher - it has no parameters of any 1692Initialises and configures the idle watcher - it has no parameters of any
1441kind. There is a C<ev_idle_set> macro, but using it is utterly pointless, 1693kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1442believe me. 1694believe me.
1443 1695
1444=back 1696=back
1697
1698=head3 Examples
1445 1699
1446Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the 1700Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1447callback, free it. Also, use no error checking, as usual. 1701callback, free it. Also, use no error checking, as usual.
1448 1702
1449 static void 1703 static void
1450 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents) 1704 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1451 { 1705 {
1452 free (w); 1706 free (w);
1453 // now do something you wanted to do when the program has 1707 // now do something you wanted to do when the program has
1454 // no longer asnything immediate to do. 1708 // no longer anything immediate to do.
1455 } 1709 }
1456 1710
1457 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle)); 1711 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1458 ev_idle_init (idle_watcher, idle_cb); 1712 ev_idle_init (idle_watcher, idle_cb);
1459 ev_idle_start (loop, idle_cb); 1713 ev_idle_start (loop, idle_cb);
… …
1501 1755
1502It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>) 1756It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
1503priority, to ensure that they are being run before any other watchers 1757priority, to ensure that they are being run before any other watchers
1504after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers, 1758after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,
1505too) should not activate ("feed") events into libev. While libev fully 1759too) should not activate ("feed") events into libev. While libev fully
1506supports this, they will be called before other C<ev_check> watchers did 1760supports this, they will be called before other C<ev_check> watchers
1507their job. As C<ev_check> watchers are often used to embed other event 1761did their job. As C<ev_check> watchers are often used to embed other
1508loops those other event loops might be in an unusable state until their 1762(non-libev) event loops those other event loops might be in an unusable
1509C<ev_check> watcher ran (always remind yourself to coexist peacefully with 1763state until their C<ev_check> watcher ran (always remind yourself to
1510others). 1764coexist peacefully with others).
1765
1766=head3 Watcher-Specific Functions and Data Members
1511 1767
1512=over 4 1768=over 4
1513 1769
1514=item ev_prepare_init (ev_prepare *, callback) 1770=item ev_prepare_init (ev_prepare *, callback)
1515 1771
… …
1518Initialises and configures the prepare or check watcher - they have no 1774Initialises and configures the prepare or check watcher - they have no
1519parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set> 1775parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1520macros, but using them is utterly, utterly and completely pointless. 1776macros, but using them is utterly, utterly and completely pointless.
1521 1777
1522=back 1778=back
1779
1780=head3 Examples
1523 1781
1524There are a number of principal ways to embed other event loops or modules 1782There are a number of principal ways to embed other event loops or modules
1525into libev. Here are some ideas on how to include libadns into libev 1783into libev. Here are some ideas on how to include libadns into libev
1526(there is a Perl module named C<EV::ADNS> that does this, which you could 1784(there is a Perl module named C<EV::ADNS> that does this, which you could
1527use for an actually working example. Another Perl module named C<EV::Glib> 1785use for an actually working example. Another Perl module named C<EV::Glib>
… …
1696portable one. 1954portable one.
1697 1955
1698So when you want to use this feature you will always have to be prepared 1956So when you want to use this feature you will always have to be prepared
1699that you cannot get an embeddable loop. The recommended way to get around 1957that you cannot get an embeddable loop. The recommended way to get around
1700this is to have a separate variables for your embeddable loop, try to 1958this is to have a separate variables for your embeddable loop, try to
1701create it, and if that fails, use the normal loop for everything: 1959create it, and if that fails, use the normal loop for everything.
1960
1961=head3 Watcher-Specific Functions and Data Members
1962
1963=over 4
1964
1965=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
1966
1967=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
1968
1969Configures the watcher to embed the given loop, which must be
1970embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1971invoked automatically, otherwise it is the responsibility of the callback
1972to invoke it (it will continue to be called until the sweep has been done,
1973if you do not want thta, you need to temporarily stop the embed watcher).
1974
1975=item ev_embed_sweep (loop, ev_embed *)
1976
1977Make a single, non-blocking sweep over the embedded loop. This works
1978similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1979apropriate way for embedded loops.
1980
1981=item struct ev_loop *other [read-only]
1982
1983The embedded event loop.
1984
1985=back
1986
1987=head3 Examples
1988
1989Example: Try to get an embeddable event loop and embed it into the default
1990event loop. If that is not possible, use the default loop. The default
1991loop is stored in C<loop_hi>, while the mebeddable loop is stored in
1992C<loop_lo> (which is C<loop_hi> in the acse no embeddable loop can be
1993used).
1702 1994
1703 struct ev_loop *loop_hi = ev_default_init (0); 1995 struct ev_loop *loop_hi = ev_default_init (0);
1704 struct ev_loop *loop_lo = 0; 1996 struct ev_loop *loop_lo = 0;
1705 struct ev_embed embed; 1997 struct ev_embed embed;
1706 1998
… …
1717 ev_embed_start (loop_hi, &embed); 2009 ev_embed_start (loop_hi, &embed);
1718 } 2010 }
1719 else 2011 else
1720 loop_lo = loop_hi; 2012 loop_lo = loop_hi;
1721 2013
1722=over 4 2014Example: Check if kqueue is available but not recommended and create
2015a kqueue backend for use with sockets (which usually work with any
2016kqueue implementation). Store the kqueue/socket-only event loop in
2017C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
1723 2018
1724=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop) 2019 struct ev_loop *loop = ev_default_init (0);
2020 struct ev_loop *loop_socket = 0;
2021 struct ev_embed embed;
2022
2023 if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
2024 if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
2025 {
2026 ev_embed_init (&embed, 0, loop_socket);
2027 ev_embed_start (loop, &embed);
2028 }
1725 2029
1726=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop) 2030 if (!loop_socket)
2031 loop_socket = loop;
1727 2032
1728Configures the watcher to embed the given loop, which must be 2033 // now use loop_socket for all sockets, and loop for everything else
1729embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1730invoked automatically, otherwise it is the responsibility of the callback
1731to invoke it (it will continue to be called until the sweep has been done,
1732if you do not want thta, you need to temporarily stop the embed watcher).
1733
1734=item ev_embed_sweep (loop, ev_embed *)
1735
1736Make a single, non-blocking sweep over the embedded loop. This works
1737similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1738apropriate way for embedded loops.
1739
1740=item struct ev_loop *loop [read-only]
1741
1742The embedded event loop.
1743
1744=back
1745 2034
1746 2035
1747=head2 C<ev_fork> - the audacity to resume the event loop after a fork 2036=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1748 2037
1749Fork watchers are called when a C<fork ()> was detected (usually because 2038Fork watchers are called when a C<fork ()> was detected (usually because
… …
1752event loop blocks next and before C<ev_check> watchers are being called, 2041event loop blocks next and before C<ev_check> watchers are being called,
1753and only in the child after the fork. If whoever good citizen calling 2042and only in the child after the fork. If whoever good citizen calling
1754C<ev_default_fork> cheats and calls it in the wrong process, the fork 2043C<ev_default_fork> cheats and calls it in the wrong process, the fork
1755handlers will be invoked, too, of course. 2044handlers will be invoked, too, of course.
1756 2045
2046=head3 Watcher-Specific Functions and Data Members
2047
1757=over 4 2048=over 4
1758 2049
1759=item ev_fork_init (ev_signal *, callback) 2050=item ev_fork_init (ev_signal *, callback)
1760 2051
1761Initialises and configures the fork watcher - it has no parameters of any 2052Initialises and configures the fork watcher - it has no parameters of any
1762kind. There is a C<ev_fork_set> macro, but using it is utterly pointless, 2053kind. There is a C<ev_fork_set> macro, but using it is utterly pointless,
1763believe me. 2054believe me.
2055
2056=back
2057
2058
2059=head2 C<ev_async> - how to wake up another event loop
2060
2061In general, you cannot use an C<ev_loop> from multiple threads or other
2062asynchronous sources such as signal handlers (as opposed to multiple event
2063loops - those are of course safe to use in different threads).
2064
2065Sometimes, however, you need to wake up another event loop you do not
2066control, for example because it belongs to another thread. This is what
2067C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
2068can signal it by calling C<ev_async_send>, which is thread- and signal
2069safe.
2070
2071This functionality is very similar to C<ev_signal> watchers, as signals,
2072too, are asynchronous in nature, and signals, too, will be compressed
2073(i.e. the number of callback invocations may be less than the number of
2074C<ev_async_sent> calls).
2075
2076Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
2077just the default loop.
2078
2079=head3 Queueing
2080
2081C<ev_async> does not support queueing of data in any way. The reason
2082is that the author does not know of a simple (or any) algorithm for a
2083multiple-writer-single-reader queue that works in all cases and doesn't
2084need elaborate support such as pthreads.
2085
2086That means that if you want to queue data, you have to provide your own
2087queue. But at least I can tell you would implement locking around your
2088queue:
2089
2090=over 4
2091
2092=item queueing from a signal handler context
2093
2094To implement race-free queueing, you simply add to the queue in the signal
2095handler but you block the signal handler in the watcher callback. Here is an example that does that for
2096some fictitiuous SIGUSR1 handler:
2097
2098 static ev_async mysig;
2099
2100 static void
2101 sigusr1_handler (void)
2102 {
2103 sometype data;
2104
2105 // no locking etc.
2106 queue_put (data);
2107 ev_async_send (DEFAULT_ &mysig);
2108 }
2109
2110 static void
2111 mysig_cb (EV_P_ ev_async *w, int revents)
2112 {
2113 sometype data;
2114 sigset_t block, prev;
2115
2116 sigemptyset (&block);
2117 sigaddset (&block, SIGUSR1);
2118 sigprocmask (SIG_BLOCK, &block, &prev);
2119
2120 while (queue_get (&data))
2121 process (data);
2122
2123 if (sigismember (&prev, SIGUSR1)
2124 sigprocmask (SIG_UNBLOCK, &block, 0);
2125 }
2126
2127(Note: pthreads in theory requires you to use C<pthread_setmask>
2128instead of C<sigprocmask> when you use threads, but libev doesn't do it
2129either...).
2130
2131=item queueing from a thread context
2132
2133The strategy for threads is different, as you cannot (easily) block
2134threads but you can easily preempt them, so to queue safely you need to
2135employ a traditional mutex lock, such as in this pthread example:
2136
2137 static ev_async mysig;
2138 static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
2139
2140 static void
2141 otherthread (void)
2142 {
2143 // only need to lock the actual queueing operation
2144 pthread_mutex_lock (&mymutex);
2145 queue_put (data);
2146 pthread_mutex_unlock (&mymutex);
2147
2148 ev_async_send (DEFAULT_ &mysig);
2149 }
2150
2151 static void
2152 mysig_cb (EV_P_ ev_async *w, int revents)
2153 {
2154 pthread_mutex_lock (&mymutex);
2155
2156 while (queue_get (&data))
2157 process (data);
2158
2159 pthread_mutex_unlock (&mymutex);
2160 }
2161
2162=back
2163
2164
2165=head3 Watcher-Specific Functions and Data Members
2166
2167=over 4
2168
2169=item ev_async_init (ev_async *, callback)
2170
2171Initialises and configures the async watcher - it has no parameters of any
2172kind. There is a C<ev_asynd_set> macro, but using it is utterly pointless,
2173believe me.
2174
2175=item ev_async_send (loop, ev_async *)
2176
2177Sends/signals/activates the given C<ev_async> watcher, that is, feeds
2178an C<EV_ASYNC> event on the watcher into the event loop. Unlike
2179C<ev_feed_event>, this call is safe to do in other threads, signal or
2180similar contexts (see the dicusssion of C<EV_ATOMIC_T> in the embedding
2181section below on what exactly this means).
2182
2183This call incurs the overhead of a syscall only once per loop iteration,
2184so while the overhead might be noticable, it doesn't apply to repeated
2185calls to C<ev_async_send>.
1764 2186
1765=back 2187=back
1766 2188
1767 2189
1768=head1 OTHER FUNCTIONS 2190=head1 OTHER FUNCTIONS
… …
1977 2399
1978=item w->stop () 2400=item w->stop ()
1979 2401
1980Stops the watcher if it is active. Again, no C<loop> argument. 2402Stops the watcher if it is active. Again, no C<loop> argument.
1981 2403
1982=item w->again () C<ev::timer>, C<ev::periodic> only 2404=item w->again () (C<ev::timer>, C<ev::periodic> only)
1983 2405
1984For C<ev::timer> and C<ev::periodic>, this invokes the corresponding 2406For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
1985C<ev_TYPE_again> function. 2407C<ev_TYPE_again> function.
1986 2408
1987=item w->sweep () C<ev::embed> only 2409=item w->sweep () (C<ev::embed> only)
1988 2410
1989Invokes C<ev_embed_sweep>. 2411Invokes C<ev_embed_sweep>.
1990 2412
1991=item w->update () C<ev::stat> only 2413=item w->update () (C<ev::stat> only)
1992 2414
1993Invokes C<ev_stat_stat>. 2415Invokes C<ev_stat_stat>.
1994 2416
1995=back 2417=back
1996 2418
… …
1999Example: Define a class with an IO and idle watcher, start one of them in 2421Example: Define a class with an IO and idle watcher, start one of them in
2000the constructor. 2422the constructor.
2001 2423
2002 class myclass 2424 class myclass
2003 { 2425 {
2004 ev_io io; void io_cb (ev::io &w, int revents); 2426 ev::io io; void io_cb (ev::io &w, int revents);
2005 ev_idle idle void idle_cb (ev::idle &w, int revents); 2427 ev:idle idle void idle_cb (ev::idle &w, int revents);
2006 2428
2007 myclass (); 2429 myclass (int fd)
2008 }
2009
2010 myclass::myclass (int fd)
2011 { 2430 {
2012 io .set <myclass, &myclass::io_cb > (this); 2431 io .set <myclass, &myclass::io_cb > (this);
2013 idle.set <myclass, &myclass::idle_cb> (this); 2432 idle.set <myclass, &myclass::idle_cb> (this);
2014 2433
2015 io.start (fd, ev::READ); 2434 io.start (fd, ev::READ);
2435 }
2016 } 2436 };
2017 2437
2018 2438
2019=head1 MACRO MAGIC 2439=head1 MACRO MAGIC
2020 2440
2021Libev can be compiled with a variety of options, the most fundemantal is 2441Libev can be compiled with a variety of options, the most fundamantal
2022C<EV_MULTIPLICITY>. This option determines whether (most) functions and 2442of which is C<EV_MULTIPLICITY>. This option determines whether (most)
2023callbacks have an initial C<struct ev_loop *> argument. 2443functions and callbacks have an initial C<struct ev_loop *> argument.
2024 2444
2025To make it easier to write programs that cope with either variant, the 2445To make it easier to write programs that cope with either variant, the
2026following macros are defined: 2446following macros are defined:
2027 2447
2028=over 4 2448=over 4
… …
2082Libev can (and often is) directly embedded into host 2502Libev can (and often is) directly embedded into host
2083applications. Examples of applications that embed it include the Deliantra 2503applications. Examples of applications that embed it include the Deliantra
2084Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe) 2504Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
2085and rxvt-unicode. 2505and rxvt-unicode.
2086 2506
2087The goal is to enable you to just copy the neecssary files into your 2507The goal is to enable you to just copy the necessary files into your
2088source directory without having to change even a single line in them, so 2508source directory without having to change even a single line in them, so
2089you can easily upgrade by simply copying (or having a checked-out copy of 2509you can easily upgrade by simply copying (or having a checked-out copy of
2090libev somewhere in your source tree). 2510libev somewhere in your source tree).
2091 2511
2092=head2 FILESETS 2512=head2 FILESETS
… …
2182 2602
2183If defined to be C<1>, libev will try to detect the availability of the 2603If defined to be C<1>, libev will try to detect the availability of the
2184monotonic clock option at both compiletime and runtime. Otherwise no use 2604monotonic clock option at both compiletime and runtime. Otherwise no use
2185of the monotonic clock option will be attempted. If you enable this, you 2605of the monotonic clock option will be attempted. If you enable this, you
2186usually have to link against librt or something similar. Enabling it when 2606usually have to link against librt or something similar. Enabling it when
2187the functionality isn't available is safe, though, althoguh you have 2607the functionality isn't available is safe, though, although you have
2188to make sure you link against any libraries where the C<clock_gettime> 2608to make sure you link against any libraries where the C<clock_gettime>
2189function is hiding in (often F<-lrt>). 2609function is hiding in (often F<-lrt>).
2190 2610
2191=item EV_USE_REALTIME 2611=item EV_USE_REALTIME
2192 2612
2193If defined to be C<1>, libev will try to detect the availability of the 2613If defined to be C<1>, libev will try to detect the availability of the
2194realtime clock option at compiletime (and assume its availability at 2614realtime clock option at compiletime (and assume its availability at
2195runtime if successful). Otherwise no use of the realtime clock option will 2615runtime if successful). Otherwise no use of the realtime clock option will
2196be attempted. This effectively replaces C<gettimeofday> by C<clock_get 2616be attempted. This effectively replaces C<gettimeofday> by C<clock_get
2197(CLOCK_REALTIME, ...)> and will not normally affect correctness. See tzhe note about libraries 2617(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the
2198in the description of C<EV_USE_MONOTONIC>, though. 2618note about libraries in the description of C<EV_USE_MONOTONIC>, though.
2619
2620=item EV_USE_NANOSLEEP
2621
2622If defined to be C<1>, libev will assume that C<nanosleep ()> is available
2623and will use it for delays. Otherwise it will use C<select ()>.
2199 2624
2200=item EV_USE_SELECT 2625=item EV_USE_SELECT
2201 2626
2202If undefined or defined to be C<1>, libev will compile in support for the 2627If undefined or defined to be C<1>, libev will compile in support for the
2203C<select>(2) backend. No attempt at autodetection will be done: if no 2628C<select>(2) backend. No attempt at autodetection will be done: if no
… …
2221wants osf handles on win32 (this is the case when the select to 2646wants osf handles on win32 (this is the case when the select to
2222be used is the winsock select). This means that it will call 2647be used is the winsock select). This means that it will call
2223C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise, 2648C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2224it is assumed that all these functions actually work on fds, even 2649it is assumed that all these functions actually work on fds, even
2225on win32. Should not be defined on non-win32 platforms. 2650on win32. Should not be defined on non-win32 platforms.
2651
2652=item EV_FD_TO_WIN32_HANDLE
2653
2654If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
2655file descriptors to socket handles. When not defining this symbol (the
2656default), then libev will call C<_get_osfhandle>, which is usually
2657correct. In some cases, programs use their own file descriptor management,
2658in which case they can provide this function to map fds to socket handles.
2226 2659
2227=item EV_USE_POLL 2660=item EV_USE_POLL
2228 2661
2229If defined to be C<1>, libev will compile in support for the C<poll>(2) 2662If defined to be C<1>, libev will compile in support for the C<poll>(2)
2230backend. Otherwise it will be enabled on non-win32 platforms. It 2663backend. Otherwise it will be enabled on non-win32 platforms. It
… …
2264 2697
2265If defined to be C<1>, libev will compile in support for the Linux inotify 2698If defined to be C<1>, libev will compile in support for the Linux inotify
2266interface to speed up C<ev_stat> watchers. Its actual availability will 2699interface to speed up C<ev_stat> watchers. Its actual availability will
2267be detected at runtime. 2700be detected at runtime.
2268 2701
2702=item EV_ATOMIC_T
2703
2704Libev requires an integer type (suitable for storing C<0> or C<1>) whose
2705access is atomic with respect to other threads or signal contexts. No such
2706type is easily found in the C language, so you can provide your own type
2707that you know is safe for your purposes. It is used both for signal handler "locking"
2708as well as for signal and thread safety in C<ev_async> watchers.
2709
2710In the absense of this define, libev will use C<sig_atomic_t volatile>
2711(from F<signal.h>), which is usually good enough on most platforms.
2712
2269=item EV_H 2713=item EV_H
2270 2714
2271The name of the F<ev.h> header file used to include it. The default if 2715The name of the F<ev.h> header file used to include it. The default if
2272undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This 2716undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
2273can be used to virtually rename the F<ev.h> header file in case of conflicts. 2717used to virtually rename the F<ev.h> header file in case of conflicts.
2274 2718
2275=item EV_CONFIG_H 2719=item EV_CONFIG_H
2276 2720
2277If C<EV_STANDALONE> isn't C<1>, this variable can be used to override 2721If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2278F<ev.c>'s idea of where to find the F<config.h> file, similarly to 2722F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2279C<EV_H>, above. 2723C<EV_H>, above.
2280 2724
2281=item EV_EVENT_H 2725=item EV_EVENT_H
2282 2726
2283Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea 2727Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2284of how the F<event.h> header can be found. 2728of how the F<event.h> header can be found, the default is C<"event.h">.
2285 2729
2286=item EV_PROTOTYPES 2730=item EV_PROTOTYPES
2287 2731
2288If defined to be C<0>, then F<ev.h> will not define any function 2732If defined to be C<0>, then F<ev.h> will not define any function
2289prototypes, but still define all the structs and other symbols. This is 2733prototypes, but still define all the structs and other symbols. This is
… …
2340=item EV_FORK_ENABLE 2784=item EV_FORK_ENABLE
2341 2785
2342If undefined or defined to be C<1>, then fork watchers are supported. If 2786If undefined or defined to be C<1>, then fork watchers are supported. If
2343defined to be C<0>, then they are not. 2787defined to be C<0>, then they are not.
2344 2788
2789=item EV_ASYNC_ENABLE
2790
2791If undefined or defined to be C<1>, then async watchers are supported. If
2792defined to be C<0>, then they are not.
2793
2345=item EV_MINIMAL 2794=item EV_MINIMAL
2346 2795
2347If you need to shave off some kilobytes of code at the expense of some 2796If you need to shave off some kilobytes of code at the expense of some
2348speed, define this symbol to C<1>. Currently only used for gcc to override 2797speed, define this symbol to C<1>. Currently only used for gcc to override
2349some inlining decisions, saves roughly 30% codesize of amd64. 2798some inlining decisions, saves roughly 30% codesize of amd64.
… …
2355than enough. If you need to manage thousands of children you might want to 2804than enough. If you need to manage thousands of children you might want to
2356increase this value (I<must> be a power of two). 2805increase this value (I<must> be a power of two).
2357 2806
2358=item EV_INOTIFY_HASHSIZE 2807=item EV_INOTIFY_HASHSIZE
2359 2808
2360C<ev_staz> watchers use a small hash table to distribute workload by 2809C<ev_stat> watchers use a small hash table to distribute workload by
2361inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>), 2810inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
2362usually more than enough. If you need to manage thousands of C<ev_stat> 2811usually more than enough. If you need to manage thousands of C<ev_stat>
2363watchers you might want to increase this value (I<must> be a power of 2812watchers you might want to increase this value (I<must> be a power of
2364two). 2813two).
2365 2814
… …
2382 2831
2383=item ev_set_cb (ev, cb) 2832=item ev_set_cb (ev, cb)
2384 2833
2385Can be used to change the callback member declaration in each watcher, 2834Can be used to change the callback member declaration in each watcher,
2386and the way callbacks are invoked and set. Must expand to a struct member 2835and the way callbacks are invoked and set. Must expand to a struct member
2387definition and a statement, respectively. See the F<ev.v> header file for 2836definition and a statement, respectively. See the F<ev.h> header file for
2388their default definitions. One possible use for overriding these is to 2837their default definitions. One possible use for overriding these is to
2389avoid the C<struct ev_loop *> as first argument in all cases, or to use 2838avoid the C<struct ev_loop *> as first argument in all cases, or to use
2390method calls instead of plain function calls in C++. 2839method calls instead of plain function calls in C++.
2840
2841=head2 EXPORTED API SYMBOLS
2842
2843If you need to re-export the API (e.g. via a dll) and you need a list of
2844exported symbols, you can use the provided F<Symbol.*> files which list
2845all public symbols, one per line:
2846
2847 Symbols.ev for libev proper
2848 Symbols.event for the libevent emulation
2849
2850This can also be used to rename all public symbols to avoid clashes with
2851multiple versions of libev linked together (which is obviously bad in
2852itself, but sometimes it is inconvinient to avoid this).
2853
2854A sed command like this will create wrapper C<#define>'s that you need to
2855include before including F<ev.h>:
2856
2857 <Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
2858
2859This would create a file F<wrap.h> which essentially looks like this:
2860
2861 #define ev_backend myprefix_ev_backend
2862 #define ev_check_start myprefix_ev_check_start
2863 #define ev_check_stop myprefix_ev_check_stop
2864 ...
2391 2865
2392=head2 EXAMPLES 2866=head2 EXAMPLES
2393 2867
2394For a real-world example of a program the includes libev 2868For a real-world example of a program the includes libev
2395verbatim, you can have a look at the EV perl module 2869verbatim, you can have a look at the EV perl module
… …
2436 2910
2437=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers) 2911=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2438 2912
2439This means that, when you have a watcher that triggers in one hour and 2913This means that, when you have a watcher that triggers in one hour and
2440there are 100 watchers that would trigger before that then inserting will 2914there are 100 watchers that would trigger before that then inserting will
2441have to skip those 100 watchers. 2915have to skip roughly seven (C<ld 100>) of these watchers.
2442 2916
2443=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers) 2917=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2444 2918
2445That means that for changing a timer costs less than removing/adding them 2919That means that changing a timer costs less than removing/adding them
2446as only the relative motion in the event queue has to be paid for. 2920as only the relative motion in the event queue has to be paid for.
2447 2921
2448=item Starting io/check/prepare/idle/signal/child watchers: O(1) 2922=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
2449 2923
2450These just add the watcher into an array or at the head of a list. 2924These just add the watcher into an array or at the head of a list.
2925
2451=item Stopping check/prepare/idle watchers: O(1) 2926=item Stopping check/prepare/idle/fork/async watchers: O(1)
2452 2927
2453=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE)) 2928=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2454 2929
2455These watchers are stored in lists then need to be walked to find the 2930These watchers are stored in lists then need to be walked to find the
2456correct watcher to remove. The lists are usually short (you don't usually 2931correct watcher to remove. The lists are usually short (you don't usually
2457have many watchers waiting for the same fd or signal). 2932have many watchers waiting for the same fd or signal).
2458 2933
2459=item Finding the next timer per loop iteration: O(1) 2934=item Finding the next timer in each loop iteration: O(1)
2935
2936By virtue of using a binary heap, the next timer is always found at the
2937beginning of the storage array.
2460 2938
2461=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd) 2939=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2462 2940
2463A change means an I/O watcher gets started or stopped, which requires 2941A change means an I/O watcher gets started or stopped, which requires
2464libev to recalculate its status (and possibly tell the kernel). 2942libev to recalculate its status (and possibly tell the kernel, depending
2943on backend and wether C<ev_io_set> was used).
2465 2944
2466=item Activating one watcher: O(1) 2945=item Activating one watcher (putting it into the pending state): O(1)
2467 2946
2468=item Priority handling: O(number_of_priorities) 2947=item Priority handling: O(number_of_priorities)
2469 2948
2470Priorities are implemented by allocating some space for each 2949Priorities are implemented by allocating some space for each
2471priority. When doing priority-based operations, libev usually has to 2950priority. When doing priority-based operations, libev usually has to
2472linearly search all the priorities. 2951linearly search all the priorities, but starting/stopping and activating
2952watchers becomes O(1) w.r.t. priority handling.
2953
2954=item Sending an ev_async: O(1)
2955
2956=item Processing ev_async_send: O(number_of_async_watchers)
2957
2958=item Processing signals: O(max_signal_number)
2959
2960Sending involves a syscall I<iff> there were no other C<ev_async_send>
2961calls in the current loop iteration. Checking for async and signal events
2962involves iterating over all running async watchers or all signal numbers.
2473 2963
2474=back 2964=back
2475 2965
2476 2966
2967=head1 Win32 platform limitations and workarounds
2968
2969Win32 doesn't support any of the standards (e.g. POSIX) that libev
2970requires, and its I/O model is fundamentally incompatible with the POSIX
2971model. Libev still offers limited functionality on this platform in
2972the form of the C<EVBACKEND_SELECT> backend, and only supports socket
2973descriptors. This only applies when using Win32 natively, not when using
2974e.g. cygwin.
2975
2976There is no supported compilation method available on windows except
2977embedding it into other applications.
2978
2979Due to the many, low, and arbitrary limits on the win32 platform and the
2980abysmal performance of winsockets, using a large number of sockets is not
2981recommended (and not reasonable). If your program needs to use more than
2982a hundred or so sockets, then likely it needs to use a totally different
2983implementation for windows, as libev offers the POSIX model, which cannot
2984be implemented efficiently on windows (microsoft monopoly games).
2985
2986=over 4
2987
2988=item The winsocket select function
2989
2990The winsocket C<select> function doesn't follow POSIX in that it requires
2991socket I<handles> and not socket I<file descriptors>. This makes select
2992very inefficient, and also requires a mapping from file descriptors
2993to socket handles. See the discussion of the C<EV_SELECT_USE_FD_SET>,
2994C<EV_SELECT_IS_WINSOCKET> and C<EV_FD_TO_WIN32_HANDLE> preprocessor
2995symbols for more info.
2996
2997The configuration for a "naked" win32 using the microsoft runtime
2998libraries and raw winsocket select is:
2999
3000 #define EV_USE_SELECT 1
3001 #define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
3002
3003Note that winsockets handling of fd sets is O(n), so you can easily get a
3004complexity in the O(n²) range when using win32.
3005
3006=item Limited number of file descriptors
3007
3008Windows has numerous arbitrary (and low) limits on things. Early versions
3009of winsocket's select only supported waiting for a max. of C<64> handles
3010(probably owning to the fact that all windows kernels can only wait for
3011C<64> things at the same time internally; microsoft recommends spawning a
3012chain of threads and wait for 63 handles and the previous thread in each).
3013
3014Newer versions support more handles, but you need to define C<FD_SETSIZE>
3015to some high number (e.g. C<2048>) before compiling the winsocket select
3016call (which might be in libev or elsewhere, for example, perl does its own
3017select emulation on windows).
3018
3019Another limit is the number of file descriptors in the microsoft runtime
3020libraries, which by default is C<64> (there must be a hidden I<64> fetish
3021or something like this inside microsoft). You can increase this by calling
3022C<_setmaxstdio>, which can increase this limit to C<2048> (another
3023arbitrary limit), but is broken in many versions of the microsoft runtime
3024libraries.
3025
3026This might get you to about C<512> or C<2048> sockets (depending on
3027windows version and/or the phase of the moon). To get more, you need to
3028wrap all I/O functions and provide your own fd management, but the cost of
3029calling select (O(n²)) will likely make this unworkable.
3030
3031=back
3032
3033
2477=head1 AUTHOR 3034=head1 AUTHOR
2478 3035
2479Marc Lehmann <libev@schmorp.de>. 3036Marc Lehmann <libev@schmorp.de>.
2480 3037

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines