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

Comparing libev/ev.pod (file contents):
Revision 1.77 by root, Sat Dec 8 22:11:14 2007 UTC vs.
Revision 1.116 by root, Mon Dec 31 01:34:09 2007 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.
… …
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.
117 119
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 ()>.
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
122You can find out the major and minor version numbers of the library 130You can find out the major and minor ABI version numbers of the library
123you linked against by calling the functions C<ev_version_major> and 131you linked against by calling the functions C<ev_version_major> and
124C<ev_version_minor>. If you want, you can compare against the global 132C<ev_version_minor>. If you want, you can compare against the global
125symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the 133symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
126version of the library your program was compiled against. 134version of the library your program was compiled against.
127 135
136These version numbers refer to the ABI version of the library, not the
137release version.
138
128Usually, it's a good idea to terminate if the major versions mismatch, 139Usually, it's a good idea to terminate if the major versions mismatch,
129as this indicates an incompatible change. Minor versions are usually 140as this indicates an incompatible change. Minor versions are usually
130compatible to older versions, so a larger minor version alone is usually 141compatible to older versions, so a larger minor version alone is usually
131not a problem. 142not a problem.
132 143
133Example: Make sure we haven't accidentally been linked against the wrong 144Example: Make sure we haven't accidentally been linked against the wrong
134version. 145version.
… …
295=item C<EVBACKEND_SELECT> (value 1, portable select backend) 306=item C<EVBACKEND_SELECT> (value 1, portable select backend)
296 307
297This is your standard select(2) backend. Not I<completely> standard, as 308This is your standard select(2) backend. Not I<completely> standard, as
298libev tries to roll its own fd_set with no limits on the number of fds, 309libev tries to roll its own fd_set with no limits on the number of fds,
299but if that fails, expect a fairly low limit on the number of fds when 310but if that fails, expect a fairly low limit on the number of fds when
300using this backend. It doesn't scale too well (O(highest_fd)), but its usually 311using this backend. It doesn't scale too well (O(highest_fd)), but its
301the fastest backend for a low number of fds. 312usually the fastest backend for a low number of (low-numbered :) fds.
313
314To get good performance out of this backend you need a high amount of
315parallelity (most of the file descriptors should be busy). If you are
316writing a server, you should C<accept ()> in a loop to accept as many
317connections as possible during one iteration. You might also want to have
318a look at C<ev_set_io_collect_interval ()> to increase the amount of
319readyness notifications you get per iteration.
302 320
303=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows) 321=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
304 322
305And this is your standard poll(2) backend. It's more complicated than 323And this is your standard poll(2) backend. It's more complicated
306select, but handles sparse fds better and has no artificial limit on the 324than select, but handles sparse fds better and has no artificial
307number of fds you can use (except it will slow down considerably with a 325limit on the number of fds you can use (except it will slow down
308lot of inactive fds). It scales similarly to select, i.e. O(total_fds). 326considerably with a lot of inactive fds). It scales similarly to select,
327i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
328performance tips.
309 329
310=item C<EVBACKEND_EPOLL> (value 4, Linux) 330=item C<EVBACKEND_EPOLL> (value 4, Linux)
311 331
312For few fds, this backend is a bit little slower than poll and select, 332For few fds, this backend is a bit little slower than poll and select,
313but it scales phenomenally better. While poll and select usually scale like 333but it scales phenomenally better. While poll and select usually scale
314O(total_fds) where n is the total number of fds (or the highest fd), epoll scales 334like O(total_fds) where n is the total number of fds (or the highest fd),
315either O(1) or O(active_fds). 335epoll scales either O(1) or O(active_fds). The epoll design has a number
336of shortcomings, such as silently dropping events in some hard-to-detect
337cases and rewiring a syscall per fd change, no fork support and bad
338support for dup.
316 339
317While stopping and starting an I/O watcher in the same iteration will 340While stopping, setting and starting an I/O watcher in the same iteration
318result in some caching, there is still a syscall per such incident 341will result in some caching, there is still a syscall per such incident
319(because the fd could point to a different file description now), so its 342(because the fd could point to a different file description now), so its
320best to avoid that. Also, dup()ed file descriptors might not work very 343best to avoid that. Also, C<dup ()>'ed file descriptors might not work
321well if you register events for both fds. 344very well if you register events for both fds.
322 345
323Please note that epoll sometimes generates spurious notifications, so you 346Please note that epoll sometimes generates spurious notifications, so you
324need to use non-blocking I/O or other means to avoid blocking when no data 347need to use non-blocking I/O or other means to avoid blocking when no data
325(or space) is available. 348(or space) is available.
326 349
350Best performance from this backend is achieved by not unregistering all
351watchers for a file descriptor until it has been closed, if possible, i.e.
352keep at least one watcher active per fd at all times.
353
354While nominally embeddeble in other event loops, this feature is broken in
355all kernel versions tested so far.
356
327=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones) 357=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
328 358
329Kqueue deserves special mention, as at the time of this writing, it 359Kqueue deserves special mention, as at the time of this writing, it
330was broken on all BSDs except NetBSD (usually it doesn't work with 360was broken on all BSDs except NetBSD (usually it doesn't work reliably
331anything but sockets and pipes, except on Darwin, where of course its 361with anything but sockets and pipes, except on Darwin, where of course
332completely useless). For this reason its not being "autodetected" 362it's completely useless). For this reason it's not being "autodetected"
333unless you explicitly specify it explicitly in the flags (i.e. using 363unless you explicitly specify it explicitly in the flags (i.e. using
334C<EVBACKEND_KQUEUE>). 364C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
365system like NetBSD.
366
367You still can embed kqueue into a normal poll or select backend and use it
368only for sockets (after having made sure that sockets work with kqueue on
369the target platform). See C<ev_embed> watchers for more info.
335 370
336It scales in the same way as the epoll backend, but the interface to the 371It scales in the same way as the epoll backend, but the interface to the
337kernel is more efficient (which says nothing about its actual speed, of 372kernel is more efficient (which says nothing about its actual speed, of
338course). While starting and stopping an I/O watcher does not cause an 373course). While stopping, setting and starting an I/O watcher does never
339extra syscall as with epoll, it still adds up to four event changes per 374cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to
340incident, so its best to avoid that. 375two event changes per incident, support for C<fork ()> is very bad and it
376drops fds silently in similarly hard-to-detect cases.
377
378This backend usually performs well under most conditions.
379
380While nominally embeddable in other event loops, this doesn't work
381everywhere, so you might need to test for this. And since it is broken
382almost everywhere, you should only use it when you have a lot of sockets
383(for which it usually works), by embedding it into another event loop
384(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and using it only for
385sockets.
341 386
342=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8) 387=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
343 388
344This is not implemented yet (and might never be). 389This is not implemented yet (and might never be, unless you send me an
390implementation). According to reports, C</dev/poll> only supports sockets
391and is not embeddable, which would limit the usefulness of this backend
392immensely.
345 393
346=item C<EVBACKEND_PORT> (value 32, Solaris 10) 394=item C<EVBACKEND_PORT> (value 32, Solaris 10)
347 395
348This uses the Solaris 10 port mechanism. As with everything on Solaris, 396This uses the Solaris 10 event port mechanism. As with everything on Solaris,
349it's really slow, but it still scales very well (O(active_fds)). 397it's really slow, but it still scales very well (O(active_fds)).
350 398
351Please note that solaris ports can result in a lot of spurious 399Please note that solaris event ports can deliver a lot of spurious
352notifications, so you need to use non-blocking I/O or other means to avoid 400notifications, so you need to use non-blocking I/O or other means to avoid
353blocking when no data (or space) is available. 401blocking when no data (or space) is available.
402
403While this backend scales well, it requires one system call per active
404file descriptor per loop iteration. For small and medium numbers of file
405descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
406might perform better.
354 407
355=item C<EVBACKEND_ALL> 408=item C<EVBACKEND_ALL>
356 409
357Try all backends (even potentially broken ones that wouldn't be tried 410Try all backends (even potentially broken ones that wouldn't be tried
358with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as 411with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
359C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>. 412C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
413
414It is definitely not recommended to use this flag.
360 415
361=back 416=back
362 417
363If one or more of these are ored into the flags value, then only these 418If one or more of these are ored into the flags value, then only these
364backends will be tried (in the reverse order as given here). If none are 419backends will be tried (in the reverse order as given here). If none are
… …
399Destroys the default loop again (frees all memory and kernel state 454Destroys the default loop again (frees all memory and kernel state
400etc.). None of the active event watchers will be stopped in the normal 455etc.). None of the active event watchers will be stopped in the normal
401sense, so e.g. C<ev_is_active> might still return true. It is your 456sense, so e.g. C<ev_is_active> might still return true. It is your
402responsibility to either stop all watchers cleanly yoursef I<before> 457responsibility to either stop all watchers cleanly yoursef I<before>
403calling this function, or cope with the fact afterwards (which is usually 458calling this function, or cope with the fact afterwards (which is usually
404the easiest thing, youc na just ignore the watchers and/or C<free ()> them 459the easiest thing, you can just ignore the watchers and/or C<free ()> them
405for example). 460for example).
461
462Note that certain global state, such as signal state, will not be freed by
463this function, and related watchers (such as signal and child watchers)
464would need to be stopped manually.
465
466In general it is not advisable to call this function except in the
467rare occasion where you really need to free e.g. the signal handling
468pipe fds. If you need dynamically allocated loops it is better to use
469C<ev_loop_new> and C<ev_loop_destroy>).
406 470
407=item ev_loop_destroy (loop) 471=item ev_loop_destroy (loop)
408 472
409Like C<ev_default_destroy>, but destroys an event loop created by an 473Like C<ev_default_destroy>, but destroys an event loop created by an
410earlier call to C<ev_loop_new>. 474earlier call to C<ev_loop_new>.
… …
455 519
456Returns the current "event loop time", which is the time the event loop 520Returns the current "event loop time", which is the time the event loop
457received events and started processing them. This timestamp does not 521received events and started processing them. This timestamp does not
458change as long as callbacks are being processed, and this is also the base 522change as long as callbacks are being processed, and this is also the base
459time used for relative timers. You can treat it as the timestamp of the 523time used for relative timers. You can treat it as the timestamp of the
460event occuring (or more correctly, libev finding out about it). 524event occurring (or more correctly, libev finding out about it).
461 525
462=item ev_loop (loop, int flags) 526=item ev_loop (loop, int flags)
463 527
464Finally, this is it, the event handler. This function usually is called 528Finally, this is it, the event handler. This function usually is called
465after you initialised all your watchers and you want to start handling 529after you initialised all your watchers and you want to start handling
… …
487usually a better approach for this kind of thing. 551usually a better approach for this kind of thing.
488 552
489Here are the gory details of what C<ev_loop> does: 553Here are the gory details of what C<ev_loop> does:
490 554
491 - Before the first iteration, call any pending watchers. 555 - Before the first iteration, call any pending watchers.
492 * If there are no active watchers (reference count is zero), return. 556 * If EVFLAG_FORKCHECK was used, check for a fork.
493 - Queue all prepare watchers and then call all outstanding watchers. 557 - If a fork was detected, queue and call all fork watchers.
558 - Queue and call all prepare watchers.
494 - If we have been forked, recreate the kernel state. 559 - If we have been forked, recreate the kernel state.
495 - Update the kernel state with all outstanding changes. 560 - Update the kernel state with all outstanding changes.
496 - Update the "event loop time". 561 - Update the "event loop time".
497 - Calculate for how long to block. 562 - Calculate for how long to sleep or block, if at all
563 (active idle watchers, EVLOOP_NONBLOCK or not having
564 any active watchers at all will result in not sleeping).
565 - Sleep if the I/O and timer collect interval say so.
498 - Block the process, waiting for any events. 566 - Block the process, waiting for any events.
499 - Queue all outstanding I/O (fd) events. 567 - Queue all outstanding I/O (fd) events.
500 - Update the "event loop time" and do time jump handling. 568 - Update the "event loop time" and do time jump handling.
501 - Queue all outstanding timers. 569 - Queue all outstanding timers.
502 - Queue all outstanding periodics. 570 - Queue all outstanding periodics.
503 - If no events are pending now, queue all idle watchers. 571 - If no events are pending now, queue all idle watchers.
504 - Queue all check watchers. 572 - Queue all check watchers.
505 - Call all queued watchers in reverse order (i.e. check watchers first). 573 - Call all queued watchers in reverse order (i.e. check watchers first).
506 Signals and child watchers are implemented as I/O watchers, and will 574 Signals and child watchers are implemented as I/O watchers, and will
507 be handled here by queueing them when their watcher gets executed. 575 be handled here by queueing them when their watcher gets executed.
508 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK 576 - If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
509 were used, return, otherwise continue with step *. 577 were used, or there are no active watchers, return, otherwise
578 continue with step *.
510 579
511Example: Queue some jobs and then loop until no events are outsanding 580Example: Queue some jobs and then loop until no events are outstanding
512anymore. 581anymore.
513 582
514 ... queue jobs here, make sure they register event watchers as long 583 ... queue jobs here, make sure they register event watchers as long
515 ... as they still have work to do (even an idle watcher will do..) 584 ... as they still have work to do (even an idle watcher will do..)
516 ev_loop (my_loop, 0); 585 ev_loop (my_loop, 0);
… …
520 589
521Can be used to make a call to C<ev_loop> return early (but only after it 590Can be used to make a call to C<ev_loop> return early (but only after it
522has processed all outstanding events). The C<how> argument must be either 591has processed all outstanding events). The C<how> argument must be either
523C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or 592C<EVUNLOOP_ONE>, which will make the innermost C<ev_loop> call return, or
524C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return. 593C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
594
595This "unloop state" will be cleared when entering C<ev_loop> again.
525 596
526=item ev_ref (loop) 597=item ev_ref (loop)
527 598
528=item ev_unref (loop) 599=item ev_unref (loop)
529 600
… …
534returning, ev_unref() after starting, and ev_ref() before stopping it. For 605returning, ev_unref() after starting, and ev_ref() before stopping it. For
535example, libev itself uses this for its internal signal pipe: It is not 606example, libev itself uses this for its internal signal pipe: It is not
536visible to the libev user and should not keep C<ev_loop> from exiting if 607visible to the libev user and should not keep C<ev_loop> from exiting if
537no event watchers registered by it are active. It is also an excellent 608no event watchers registered by it are active. It is also an excellent
538way to do this for generic recurring timers or from within third-party 609way to do this for generic recurring timers or from within third-party
539libraries. Just remember to I<unref after start> and I<ref before stop>. 610libraries. Just remember to I<unref after start> and I<ref before stop>
611(but only if the watcher wasn't active before, or was active before,
612respectively).
540 613
541Example: Create a signal watcher, but keep it from keeping C<ev_loop> 614Example: Create a signal watcher, but keep it from keeping C<ev_loop>
542running when nothing else is active. 615running when nothing else is active.
543 616
544 struct ev_signal exitsig; 617 struct ev_signal exitsig;
… …
548 621
549Example: For some weird reason, unregister the above signal handler again. 622Example: For some weird reason, unregister the above signal handler again.
550 623
551 ev_ref (loop); 624 ev_ref (loop);
552 ev_signal_stop (loop, &exitsig); 625 ev_signal_stop (loop, &exitsig);
626
627=item ev_set_io_collect_interval (loop, ev_tstamp interval)
628
629=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
630
631These advanced functions influence the time that libev will spend waiting
632for events. Both are by default C<0>, meaning that libev will try to
633invoke timer/periodic callbacks and I/O callbacks with minimum latency.
634
635Setting these to a higher value (the C<interval> I<must> be >= C<0>)
636allows libev to delay invocation of I/O and timer/periodic callbacks to
637increase efficiency of loop iterations.
638
639The background is that sometimes your program runs just fast enough to
640handle one (or very few) event(s) per loop iteration. While this makes
641the program responsive, it also wastes a lot of CPU time to poll for new
642events, especially with backends like C<select ()> which have a high
643overhead for the actual polling but can deliver many events at once.
644
645By setting a higher I<io collect interval> you allow libev to spend more
646time collecting I/O events, so you can handle more events per iteration,
647at the cost of increasing latency. Timeouts (both C<ev_periodic> and
648C<ev_timer>) will be not affected. Setting this to a non-null value will
649introduce an additional C<ev_sleep ()> call into most loop iterations.
650
651Likewise, by setting a higher I<timeout collect interval> you allow libev
652to spend more time collecting timeouts, at the expense of increased
653latency (the watcher callback will be called later). C<ev_io> watchers
654will not be affected. Setting this to a non-null value will not introduce
655any overhead in libev.
656
657Many (busy) programs can usually benefit by setting the io collect
658interval to a value near C<0.1> or so, which is often enough for
659interactive servers (of course not for games), likewise for timeouts. It
660usually doesn't make much sense to set it to a lower value than C<0.01>,
661as this approsaches the timing granularity of most systems.
553 662
554=back 663=back
555 664
556 665
557=head1 ANATOMY OF A WATCHER 666=head1 ANATOMY OF A WATCHER
… …
883In general you can register as many read and/or write event watchers per 992In general you can register as many read and/or write event watchers per
884fd as you want (as long as you don't confuse yourself). Setting all file 993fd as you want (as long as you don't confuse yourself). Setting all file
885descriptors to non-blocking mode is also usually a good idea (but not 994descriptors to non-blocking mode is also usually a good idea (but not
886required if you know what you are doing). 995required if you know what you are doing).
887 996
888You have to be careful with dup'ed file descriptors, though. Some backends
889(the linux epoll backend is a notable example) cannot handle dup'ed file
890descriptors correctly if you register interest in two or more fds pointing
891to the same underlying file/socket/etc. description (that is, they share
892the same underlying "file open").
893
894If you must do this, then force the use of a known-to-be-good backend 997If you must do this, then force the use of a known-to-be-good backend
895(at the time of this writing, this includes only C<EVBACKEND_SELECT> and 998(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
896C<EVBACKEND_POLL>). 999C<EVBACKEND_POLL>).
897 1000
898Another thing you have to watch out for is that it is quite easy to 1001Another thing you have to watch out for is that it is quite easy to
… …
908play around with an Xlib connection), then you have to seperately re-test 1011play around with an Xlib connection), then you have to seperately re-test
909whether a file descriptor is really ready with a known-to-be good interface 1012whether a file descriptor is really ready with a known-to-be good interface
910such as poll (fortunately in our Xlib example, Xlib already does this on 1013such as poll (fortunately in our Xlib example, Xlib already does this on
911its own, so its quite safe to use). 1014its own, so its quite safe to use).
912 1015
1016=head3 The special problem of disappearing file descriptors
1017
1018Some backends (e.g. kqueue, epoll) need to be told about closing a file
1019descriptor (either by calling C<close> explicitly or by any other means,
1020such as C<dup>). The reason is that you register interest in some file
1021descriptor, but when it goes away, the operating system will silently drop
1022this interest. If another file descriptor with the same number then is
1023registered with libev, there is no efficient way to see that this is, in
1024fact, a different file descriptor.
1025
1026To avoid having to explicitly tell libev about such cases, libev follows
1027the following policy: Each time C<ev_io_set> is being called, libev
1028will assume that this is potentially a new file descriptor, otherwise
1029it is assumed that the file descriptor stays the same. That means that
1030you I<have> to call C<ev_io_set> (or C<ev_io_init>) when you change the
1031descriptor even if the file descriptor number itself did not change.
1032
1033This is how one would do it normally anyway, the important point is that
1034the libev application should not optimise around libev but should leave
1035optimisations to libev.
1036
1037=head3 The special problem of dup'ed file descriptors
1038
1039Some backends (e.g. epoll), cannot register events for file descriptors,
1040but only events for the underlying file descriptions. That means when you
1041have C<dup ()>'ed file descriptors or weirder constellations, and register
1042events for them, only one file descriptor might actually receive events.
1043
1044There is no workaround possible except not registering events
1045for potentially C<dup ()>'ed file descriptors, or to resort to
1046C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1047
1048=head3 The special problem of fork
1049
1050Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
1051useless behaviour. Libev fully supports fork, but needs to be told about
1052it in the child.
1053
1054To support fork in your programs, you either have to call
1055C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
1056enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
1057C<EVBACKEND_POLL>.
1058
1059
1060=head3 Watcher-Specific Functions
1061
913=over 4 1062=over 4
914 1063
915=item ev_io_init (ev_io *, callback, int fd, int events) 1064=item ev_io_init (ev_io *, callback, int fd, int events)
916 1065
917=item ev_io_set (ev_io *, int fd, int events) 1066=item ev_io_set (ev_io *, int fd, int events)
… …
927=item int events [read-only] 1076=item int events [read-only]
928 1077
929The events being watched. 1078The events being watched.
930 1079
931=back 1080=back
1081
1082=head3 Examples
932 1083
933Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well 1084Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
934readable, but only once. Since it is likely line-buffered, you could 1085readable, but only once. Since it is likely line-buffered, you could
935attempt to read a whole line in the callback. 1086attempt to read a whole line in the callback.
936 1087
… …
969 ev_timer_set (&timer, after + ev_now () - ev_time (), 0.); 1120 ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
970 1121
971The callback is guarenteed to be invoked only when its timeout has passed, 1122The callback is guarenteed to be invoked only when its timeout has passed,
972but if multiple timers become ready during the same loop iteration then 1123but if multiple timers become ready during the same loop iteration then
973order of execution is undefined. 1124order of execution is undefined.
1125
1126=head3 Watcher-Specific Functions and Data Members
974 1127
975=over 4 1128=over 4
976 1129
977=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat) 1130=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
978 1131
… …
1032or C<ev_timer_again> is called and determines the next timeout (if any), 1185or C<ev_timer_again> is called and determines the next timeout (if any),
1033which is also when any modifications are taken into account. 1186which is also when any modifications are taken into account.
1034 1187
1035=back 1188=back
1036 1189
1190=head3 Examples
1191
1037Example: Create a timer that fires after 60 seconds. 1192Example: Create a timer that fires after 60 seconds.
1038 1193
1039 static void 1194 static void
1040 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents) 1195 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1041 { 1196 {
… …
1074but on wallclock time (absolute time). You can tell a periodic watcher 1229but on wallclock time (absolute time). You can tell a periodic watcher
1075to trigger "at" some specific point in time. For example, if you tell a 1230to trigger "at" some specific point in time. For example, if you tell a
1076periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now () 1231periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()
1077+ 10.>) and then reset your system clock to the last year, then it will 1232+ 10.>) and then reset your system clock to the last year, then it will
1078take a year to trigger the event (unlike an C<ev_timer>, which would trigger 1233take a year to trigger the event (unlike an C<ev_timer>, which would trigger
1079roughly 10 seconds later and of course not if you reset your system time 1234roughly 10 seconds later).
1080again).
1081 1235
1082They can also be used to implement vastly more complex timers, such as 1236They can also be used to implement vastly more complex timers, such as
1083triggering an event on eahc midnight, local time. 1237triggering an event on each midnight, local time or other, complicated,
1238rules.
1084 1239
1085As with timers, the callback is guarenteed to be invoked only when the 1240As with timers, the callback is guarenteed to be invoked only when the
1086time (C<at>) has been passed, but if multiple periodic timers become ready 1241time (C<at>) has been passed, but if multiple periodic timers become ready
1087during the same loop iteration then order of execution is undefined. 1242during the same loop iteration then order of execution is undefined.
1088 1243
1244=head3 Watcher-Specific Functions and Data Members
1245
1089=over 4 1246=over 4
1090 1247
1091=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb) 1248=item ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)
1092 1249
1093=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb) 1250=item ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)
… …
1095Lots of arguments, lets sort it out... There are basically three modes of 1252Lots of arguments, lets sort it out... There are basically three modes of
1096operation, and we will explain them from simplest to complex: 1253operation, and we will explain them from simplest to complex:
1097 1254
1098=over 4 1255=over 4
1099 1256
1100=item * absolute timer (interval = reschedule_cb = 0) 1257=item * absolute timer (at = time, interval = reschedule_cb = 0)
1101 1258
1102In this configuration the watcher triggers an event at the wallclock time 1259In this configuration the watcher triggers an event at the wallclock time
1103C<at> and doesn't repeat. It will not adjust when a time jump occurs, 1260C<at> and doesn't repeat. It will not adjust when a time jump occurs,
1104that is, if it is to be run at January 1st 2011 then it will run when the 1261that is, if it is to be run at January 1st 2011 then it will run when the
1105system time reaches or surpasses this time. 1262system time reaches or surpasses this time.
1106 1263
1107=item * non-repeating interval timer (interval > 0, reschedule_cb = 0) 1264=item * non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
1108 1265
1109In this mode the watcher will always be scheduled to time out at the next 1266In this mode the watcher will always be scheduled to time out at the next
1110C<at + N * interval> time (for some integer N) and then repeat, regardless 1267C<at + N * interval> time (for some integer N, which can also be negative)
1111of any time jumps. 1268and then repeat, regardless of any time jumps.
1112 1269
1113This can be used to create timers that do not drift with respect to system 1270This can be used to create timers that do not drift with respect to system
1114time: 1271time:
1115 1272
1116 ev_periodic_set (&periodic, 0., 3600., 0); 1273 ev_periodic_set (&periodic, 0., 3600., 0);
… …
1122 1279
1123Another way to think about it (for the mathematically inclined) is that 1280Another way to think about it (for the mathematically inclined) is that
1124C<ev_periodic> will try to run the callback in this mode at the next possible 1281C<ev_periodic> will try to run the callback in this mode at the next possible
1125time where C<time = at (mod interval)>, regardless of any time jumps. 1282time where C<time = at (mod interval)>, regardless of any time jumps.
1126 1283
1284For numerical stability it is preferable that the C<at> value is near
1285C<ev_now ()> (the current time), but there is no range requirement for
1286this value.
1287
1127=item * manual reschedule mode (reschedule_cb = callback) 1288=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)
1128 1289
1129In this mode the values for C<interval> and C<at> are both being 1290In this mode the values for C<interval> and C<at> are both being
1130ignored. Instead, each time the periodic watcher gets scheduled, the 1291ignored. Instead, each time the periodic watcher gets scheduled, the
1131reschedule callback will be called with the watcher as first, and the 1292reschedule callback will be called with the watcher as first, and the
1132current time as second argument. 1293current time as second argument.
1133 1294
1134NOTE: I<This callback MUST NOT stop or destroy any periodic watcher, 1295NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,
1135ever, or make any event loop modifications>. If you need to stop it, 1296ever, or make any event loop modifications>. If you need to stop it,
1136return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by 1297return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by
1137starting a prepare watcher). 1298starting an C<ev_prepare> watcher, which is legal).
1138 1299
1139Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w, 1300Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
1140ev_tstamp now)>, e.g.: 1301ev_tstamp now)>, e.g.:
1141 1302
1142 static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now) 1303 static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
… …
1165Simply stops and restarts the periodic watcher again. This is only useful 1326Simply stops and restarts the periodic watcher again. This is only useful
1166when you changed some parameters or the reschedule callback would return 1327when you changed some parameters or the reschedule callback would return
1167a different time than the last time it was called (e.g. in a crond like 1328a different time than the last time it was called (e.g. in a crond like
1168program when the crontabs have changed). 1329program when the crontabs have changed).
1169 1330
1331=item ev_tstamp offset [read-write]
1332
1333When repeating, this contains the offset value, otherwise this is the
1334absolute point in time (the C<at> value passed to C<ev_periodic_set>).
1335
1336Can be modified any time, but changes only take effect when the periodic
1337timer fires or C<ev_periodic_again> is being called.
1338
1170=item ev_tstamp interval [read-write] 1339=item ev_tstamp interval [read-write]
1171 1340
1172The current interval value. Can be modified any time, but changes only 1341The current interval value. Can be modified any time, but changes only
1173take effect when the periodic timer fires or C<ev_periodic_again> is being 1342take effect when the periodic timer fires or C<ev_periodic_again> is being
1174called. 1343called.
… …
1177 1346
1178The current reschedule callback, or C<0>, if this functionality is 1347The current reschedule callback, or C<0>, if this functionality is
1179switched off. Can be changed any time, but changes only take effect when 1348switched off. Can be changed any time, but changes only take effect when
1180the periodic timer fires or C<ev_periodic_again> is being called. 1349the periodic timer fires or C<ev_periodic_again> is being called.
1181 1350
1351=item ev_tstamp at [read-only]
1352
1353When active, contains the absolute time that the watcher is supposed to
1354trigger next.
1355
1182=back 1356=back
1357
1358=head3 Examples
1183 1359
1184Example: Call a callback every hour, or, more precisely, whenever the 1360Example: Call a callback every hour, or, more precisely, whenever the
1185system clock is divisible by 3600. The callback invocation times have 1361system clock is divisible by 3600. The callback invocation times have
1186potentially a lot of jittering, but good long-term stability. 1362potentially a lot of jittering, but good long-term stability.
1187 1363
… …
1227with the kernel (thus it coexists with your own signal handlers as long 1403with the kernel (thus it coexists with your own signal handlers as long
1228as you don't register any with libev). Similarly, when the last signal 1404as you don't register any with libev). Similarly, when the last signal
1229watcher for a signal is stopped libev will reset the signal handler to 1405watcher for a signal is stopped libev will reset the signal handler to
1230SIG_DFL (regardless of what it was set to before). 1406SIG_DFL (regardless of what it was set to before).
1231 1407
1408=head3 Watcher-Specific Functions and Data Members
1409
1232=over 4 1410=over 4
1233 1411
1234=item ev_signal_init (ev_signal *, callback, int signum) 1412=item ev_signal_init (ev_signal *, callback, int signum)
1235 1413
1236=item ev_signal_set (ev_signal *, int signum) 1414=item ev_signal_set (ev_signal *, int signum)
… …
1247 1425
1248=head2 C<ev_child> - watch out for process status changes 1426=head2 C<ev_child> - watch out for process status changes
1249 1427
1250Child watchers trigger when your process receives a SIGCHLD in response to 1428Child watchers trigger when your process receives a SIGCHLD in response to
1251some child status changes (most typically when a child of yours dies). 1429some child status changes (most typically when a child of yours dies).
1430
1431=head3 Watcher-Specific Functions and Data Members
1252 1432
1253=over 4 1433=over 4
1254 1434
1255=item ev_child_init (ev_child *, callback, int pid) 1435=item ev_child_init (ev_child *, callback, int pid)
1256 1436
… …
1275 1455
1276The process exit/trace status caused by C<rpid> (see your systems 1456The process exit/trace status caused by C<rpid> (see your systems
1277C<waitpid> and C<sys/wait.h> documentation for details). 1457C<waitpid> and C<sys/wait.h> documentation for details).
1278 1458
1279=back 1459=back
1460
1461=head3 Examples
1280 1462
1281Example: Try to exit cleanly on SIGINT and SIGTERM. 1463Example: Try to exit cleanly on SIGINT and SIGTERM.
1282 1464
1283 static void 1465 static void
1284 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents) 1466 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
… …
1325semantics of C<ev_stat> watchers, which means that libev sometimes needs 1507semantics of C<ev_stat> watchers, which means that libev sometimes needs
1326to fall back to regular polling again even with inotify, but changes are 1508to fall back to regular polling again even with inotify, but changes are
1327usually detected immediately, and if the file exists there will be no 1509usually detected immediately, and if the file exists there will be no
1328polling. 1510polling.
1329 1511
1512=head3 Inotify
1513
1514When C<inotify (7)> support has been compiled into libev (generally only
1515available on Linux) and present at runtime, it will be used to speed up
1516change detection where possible. The inotify descriptor will be created lazily
1517when the first C<ev_stat> watcher is being started.
1518
1519Inotify presense does not change the semantics of C<ev_stat> watchers
1520except that changes might be detected earlier, and in some cases, to avoid
1521making regular C<stat> calls. Even in the presense of inotify support
1522there are many cases where libev has to resort to regular C<stat> polling.
1523
1524(There is no support for kqueue, as apparently it cannot be used to
1525implement this functionality, due to the requirement of having a file
1526descriptor open on the object at all times).
1527
1528=head3 The special problem of stat time resolution
1529
1530The C<stat ()> syscall only supports full-second resolution portably, and
1531even on systems where the resolution is higher, many filesystems still
1532only support whole seconds.
1533
1534That means that, if the time is the only thing that changes, you might
1535miss updates: on the first update, C<ev_stat> detects a change and calls
1536your callback, which does something. When there is another update within
1537the same second, C<ev_stat> will be unable to detect it.
1538
1539The solution to this is to delay acting on a change for a second (or till
1540the next second boundary), using a roughly one-second delay C<ev_timer>
1541(C<ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)>). The C<.01>
1542is added to work around small timing inconsistencies of some operating
1543systems.
1544
1545=head3 Watcher-Specific Functions and Data Members
1546
1330=over 4 1547=over 4
1331 1548
1332=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval) 1549=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
1333 1550
1334=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval) 1551=item ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)
… …
1369=item const char *path [read-only] 1586=item const char *path [read-only]
1370 1587
1371The filesystem path that is being watched. 1588The filesystem path that is being watched.
1372 1589
1373=back 1590=back
1591
1592=head3 Examples
1374 1593
1375Example: Watch C</etc/passwd> for attribute changes. 1594Example: Watch C</etc/passwd> for attribute changes.
1376 1595
1377 static void 1596 static void
1378 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents) 1597 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
… …
1391 } 1610 }
1392 1611
1393 ... 1612 ...
1394 ev_stat passwd; 1613 ev_stat passwd;
1395 1614
1396 ev_stat_init (&passwd, passwd_cb, "/etc/passwd"); 1615 ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1397 ev_stat_start (loop, &passwd); 1616 ev_stat_start (loop, &passwd);
1617
1618Example: Like above, but additionally use a one-second delay so we do not
1619miss updates (however, frequent updates will delay processing, too, so
1620one might do the work both on C<ev_stat> callback invocation I<and> on
1621C<ev_timer> callback invocation).
1622
1623 static ev_stat passwd;
1624 static ev_timer timer;
1625
1626 static void
1627 timer_cb (EV_P_ ev_timer *w, int revents)
1628 {
1629 ev_timer_stop (EV_A_ w);
1630
1631 /* now it's one second after the most recent passwd change */
1632 }
1633
1634 static void
1635 stat_cb (EV_P_ ev_stat *w, int revents)
1636 {
1637 /* reset the one-second timer */
1638 ev_timer_again (EV_A_ &timer);
1639 }
1640
1641 ...
1642 ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
1643 ev_stat_start (loop, &passwd);
1644 ev_timer_init (&timer, timer_cb, 0., 1.01);
1398 1645
1399 1646
1400=head2 C<ev_idle> - when you've got nothing better to do... 1647=head2 C<ev_idle> - when you've got nothing better to do...
1401 1648
1402Idle watchers trigger events when no other events of the same or higher 1649Idle watchers trigger events when no other events of the same or higher
… …
1416Apart from keeping your process non-blocking (which is a useful 1663Apart from keeping your process non-blocking (which is a useful
1417effect on its own sometimes), idle watchers are a good place to do 1664effect on its own sometimes), idle watchers are a good place to do
1418"pseudo-background processing", or delay processing stuff to after the 1665"pseudo-background processing", or delay processing stuff to after the
1419event loop has handled all outstanding events. 1666event loop has handled all outstanding events.
1420 1667
1668=head3 Watcher-Specific Functions and Data Members
1669
1421=over 4 1670=over 4
1422 1671
1423=item ev_idle_init (ev_signal *, callback) 1672=item ev_idle_init (ev_signal *, callback)
1424 1673
1425Initialises and configures the idle watcher - it has no parameters of any 1674Initialises and configures the idle watcher - it has no parameters of any
1426kind. There is a C<ev_idle_set> macro, but using it is utterly pointless, 1675kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1427believe me. 1676believe me.
1428 1677
1429=back 1678=back
1679
1680=head3 Examples
1430 1681
1431Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the 1682Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1432callback, free it. Also, use no error checking, as usual. 1683callback, free it. Also, use no error checking, as usual.
1433 1684
1434 static void 1685 static void
… …
1486 1737
1487It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>) 1738It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
1488priority, to ensure that they are being run before any other watchers 1739priority, to ensure that they are being run before any other watchers
1489after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers, 1740after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,
1490too) should not activate ("feed") events into libev. While libev fully 1741too) should not activate ("feed") events into libev. While libev fully
1491supports this, they will be called before other C<ev_check> watchers did 1742supports this, they will be called before other C<ev_check> watchers
1492their job. As C<ev_check> watchers are often used to embed other event 1743did their job. As C<ev_check> watchers are often used to embed other
1493loops those other event loops might be in an unusable state until their 1744(non-libev) event loops those other event loops might be in an unusable
1494C<ev_check> watcher ran (always remind yourself to coexist peacefully with 1745state until their C<ev_check> watcher ran (always remind yourself to
1495others). 1746coexist peacefully with others).
1747
1748=head3 Watcher-Specific Functions and Data Members
1496 1749
1497=over 4 1750=over 4
1498 1751
1499=item ev_prepare_init (ev_prepare *, callback) 1752=item ev_prepare_init (ev_prepare *, callback)
1500 1753
… …
1503Initialises and configures the prepare or check watcher - they have no 1756Initialises and configures the prepare or check watcher - they have no
1504parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set> 1757parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1505macros, but using them is utterly, utterly and completely pointless. 1758macros, but using them is utterly, utterly and completely pointless.
1506 1759
1507=back 1760=back
1761
1762=head3 Examples
1508 1763
1509There are a number of principal ways to embed other event loops or modules 1764There are a number of principal ways to embed other event loops or modules
1510into libev. Here are some ideas on how to include libadns into libev 1765into libev. Here are some ideas on how to include libadns into libev
1511(there is a Perl module named C<EV::ADNS> that does this, which you could 1766(there is a Perl module named C<EV::ADNS> that does this, which you could
1512use for an actually working example. Another Perl module named C<EV::Glib> 1767use for an actually working example. Another Perl module named C<EV::Glib>
… …
1681portable one. 1936portable one.
1682 1937
1683So when you want to use this feature you will always have to be prepared 1938So when you want to use this feature you will always have to be prepared
1684that you cannot get an embeddable loop. The recommended way to get around 1939that you cannot get an embeddable loop. The recommended way to get around
1685this is to have a separate variables for your embeddable loop, try to 1940this is to have a separate variables for your embeddable loop, try to
1686create it, and if that fails, use the normal loop for everything: 1941create it, and if that fails, use the normal loop for everything.
1942
1943=head3 Watcher-Specific Functions and Data Members
1944
1945=over 4
1946
1947=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
1948
1949=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
1950
1951Configures the watcher to embed the given loop, which must be
1952embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1953invoked automatically, otherwise it is the responsibility of the callback
1954to invoke it (it will continue to be called until the sweep has been done,
1955if you do not want thta, you need to temporarily stop the embed watcher).
1956
1957=item ev_embed_sweep (loop, ev_embed *)
1958
1959Make a single, non-blocking sweep over the embedded loop. This works
1960similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1961apropriate way for embedded loops.
1962
1963=item struct ev_loop *other [read-only]
1964
1965The embedded event loop.
1966
1967=back
1968
1969=head3 Examples
1970
1971Example: Try to get an embeddable event loop and embed it into the default
1972event loop. If that is not possible, use the default loop. The default
1973loop is stored in C<loop_hi>, while the mebeddable loop is stored in
1974C<loop_lo> (which is C<loop_hi> in the acse no embeddable loop can be
1975used).
1687 1976
1688 struct ev_loop *loop_hi = ev_default_init (0); 1977 struct ev_loop *loop_hi = ev_default_init (0);
1689 struct ev_loop *loop_lo = 0; 1978 struct ev_loop *loop_lo = 0;
1690 struct ev_embed embed; 1979 struct ev_embed embed;
1691 1980
… …
1702 ev_embed_start (loop_hi, &embed); 1991 ev_embed_start (loop_hi, &embed);
1703 } 1992 }
1704 else 1993 else
1705 loop_lo = loop_hi; 1994 loop_lo = loop_hi;
1706 1995
1707=over 4 1996Example: Check if kqueue is available but not recommended and create
1997a kqueue backend for use with sockets (which usually work with any
1998kqueue implementation). Store the kqueue/socket-only event loop in
1999C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
1708 2000
1709=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop) 2001 struct ev_loop *loop = ev_default_init (0);
2002 struct ev_loop *loop_socket = 0;
2003 struct ev_embed embed;
2004
2005 if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
2006 if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
2007 {
2008 ev_embed_init (&embed, 0, loop_socket);
2009 ev_embed_start (loop, &embed);
2010 }
1710 2011
1711=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop) 2012 if (!loop_socket)
2013 loop_socket = loop;
1712 2014
1713Configures the watcher to embed the given loop, which must be 2015 // now use loop_socket for all sockets, and loop for everything else
1714embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1715invoked automatically, otherwise it is the responsibility of the callback
1716to invoke it (it will continue to be called until the sweep has been done,
1717if you do not want thta, you need to temporarily stop the embed watcher).
1718
1719=item ev_embed_sweep (loop, ev_embed *)
1720
1721Make a single, non-blocking sweep over the embedded loop. This works
1722similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1723apropriate way for embedded loops.
1724
1725=item struct ev_loop *loop [read-only]
1726
1727The embedded event loop.
1728
1729=back
1730 2016
1731 2017
1732=head2 C<ev_fork> - the audacity to resume the event loop after a fork 2018=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1733 2019
1734Fork watchers are called when a C<fork ()> was detected (usually because 2020Fork watchers are called when a C<fork ()> was detected (usually because
… …
1737event loop blocks next and before C<ev_check> watchers are being called, 2023event loop blocks next and before C<ev_check> watchers are being called,
1738and only in the child after the fork. If whoever good citizen calling 2024and only in the child after the fork. If whoever good citizen calling
1739C<ev_default_fork> cheats and calls it in the wrong process, the fork 2025C<ev_default_fork> cheats and calls it in the wrong process, the fork
1740handlers will be invoked, too, of course. 2026handlers will be invoked, too, of course.
1741 2027
2028=head3 Watcher-Specific Functions and Data Members
2029
1742=over 4 2030=over 4
1743 2031
1744=item ev_fork_init (ev_signal *, callback) 2032=item ev_fork_init (ev_signal *, callback)
1745 2033
1746Initialises and configures the fork watcher - it has no parameters of any 2034Initialises and configures the fork watcher - it has no parameters of any
… …
1962 2250
1963=item w->stop () 2251=item w->stop ()
1964 2252
1965Stops the watcher if it is active. Again, no C<loop> argument. 2253Stops the watcher if it is active. Again, no C<loop> argument.
1966 2254
1967=item w->again () C<ev::timer>, C<ev::periodic> only 2255=item w->again () (C<ev::timer>, C<ev::periodic> only)
1968 2256
1969For C<ev::timer> and C<ev::periodic>, this invokes the corresponding 2257For C<ev::timer> and C<ev::periodic>, this invokes the corresponding
1970C<ev_TYPE_again> function. 2258C<ev_TYPE_again> function.
1971 2259
1972=item w->sweep () C<ev::embed> only 2260=item w->sweep () (C<ev::embed> only)
1973 2261
1974Invokes C<ev_embed_sweep>. 2262Invokes C<ev_embed_sweep>.
1975 2263
1976=item w->update () C<ev::stat> only 2264=item w->update () (C<ev::stat> only)
1977 2265
1978Invokes C<ev_stat_stat>. 2266Invokes C<ev_stat_stat>.
1979 2267
1980=back 2268=back
1981 2269
… …
2001 } 2289 }
2002 2290
2003 2291
2004=head1 MACRO MAGIC 2292=head1 MACRO MAGIC
2005 2293
2006Libev can be compiled with a variety of options, the most fundemantal is 2294Libev can be compiled with a variety of options, the most fundamantal
2007C<EV_MULTIPLICITY>. This option determines whether (most) functions and 2295of which is C<EV_MULTIPLICITY>. This option determines whether (most)
2008callbacks have an initial C<struct ev_loop *> argument. 2296functions and callbacks have an initial C<struct ev_loop *> argument.
2009 2297
2010To make it easier to write programs that cope with either variant, the 2298To make it easier to write programs that cope with either variant, the
2011following macros are defined: 2299following macros are defined:
2012 2300
2013=over 4 2301=over 4
… …
2067Libev can (and often is) directly embedded into host 2355Libev can (and often is) directly embedded into host
2068applications. Examples of applications that embed it include the Deliantra 2356applications. Examples of applications that embed it include the Deliantra
2069Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe) 2357Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
2070and rxvt-unicode. 2358and rxvt-unicode.
2071 2359
2072The goal is to enable you to just copy the neecssary files into your 2360The goal is to enable you to just copy the necessary files into your
2073source directory without having to change even a single line in them, so 2361source directory without having to change even a single line in them, so
2074you can easily upgrade by simply copying (or having a checked-out copy of 2362you can easily upgrade by simply copying (or having a checked-out copy of
2075libev somewhere in your source tree). 2363libev somewhere in your source tree).
2076 2364
2077=head2 FILESETS 2365=head2 FILESETS
… …
2167 2455
2168If defined to be C<1>, libev will try to detect the availability of the 2456If defined to be C<1>, libev will try to detect the availability of the
2169monotonic clock option at both compiletime and runtime. Otherwise no use 2457monotonic clock option at both compiletime and runtime. Otherwise no use
2170of the monotonic clock option will be attempted. If you enable this, you 2458of the monotonic clock option will be attempted. If you enable this, you
2171usually have to link against librt or something similar. Enabling it when 2459usually have to link against librt or something similar. Enabling it when
2172the functionality isn't available is safe, though, althoguh you have 2460the functionality isn't available is safe, though, although you have
2173to make sure you link against any libraries where the C<clock_gettime> 2461to make sure you link against any libraries where the C<clock_gettime>
2174function is hiding in (often F<-lrt>). 2462function is hiding in (often F<-lrt>).
2175 2463
2176=item EV_USE_REALTIME 2464=item EV_USE_REALTIME
2177 2465
2178If defined to be C<1>, libev will try to detect the availability of the 2466If defined to be C<1>, libev will try to detect the availability of the
2179realtime clock option at compiletime (and assume its availability at 2467realtime clock option at compiletime (and assume its availability at
2180runtime if successful). Otherwise no use of the realtime clock option will 2468runtime if successful). Otherwise no use of the realtime clock option will
2181be attempted. This effectively replaces C<gettimeofday> by C<clock_get 2469be attempted. This effectively replaces C<gettimeofday> by C<clock_get
2182(CLOCK_REALTIME, ...)> and will not normally affect correctness. See tzhe note about libraries 2470(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the
2183in the description of C<EV_USE_MONOTONIC>, though. 2471note about libraries in the description of C<EV_USE_MONOTONIC>, though.
2472
2473=item EV_USE_NANOSLEEP
2474
2475If defined to be C<1>, libev will assume that C<nanosleep ()> is available
2476and will use it for delays. Otherwise it will use C<select ()>.
2184 2477
2185=item EV_USE_SELECT 2478=item EV_USE_SELECT
2186 2479
2187If undefined or defined to be C<1>, libev will compile in support for the 2480If undefined or defined to be C<1>, libev will compile in support for the
2188C<select>(2) backend. No attempt at autodetection will be done: if no 2481C<select>(2) backend. No attempt at autodetection will be done: if no
… …
2206wants osf handles on win32 (this is the case when the select to 2499wants osf handles on win32 (this is the case when the select to
2207be used is the winsock select). This means that it will call 2500be used is the winsock select). This means that it will call
2208C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise, 2501C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2209it is assumed that all these functions actually work on fds, even 2502it is assumed that all these functions actually work on fds, even
2210on win32. Should not be defined on non-win32 platforms. 2503on win32. Should not be defined on non-win32 platforms.
2504
2505=item EV_FD_TO_WIN32_HANDLE
2506
2507If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
2508file descriptors to socket handles. When not defining this symbol (the
2509default), then libev will call C<_get_osfhandle>, which is usually
2510correct. In some cases, programs use their own file descriptor management,
2511in which case they can provide this function to map fds to socket handles.
2211 2512
2212=item EV_USE_POLL 2513=item EV_USE_POLL
2213 2514
2214If defined to be C<1>, libev will compile in support for the C<poll>(2) 2515If defined to be C<1>, libev will compile in support for the C<poll>(2)
2215backend. Otherwise it will be enabled on non-win32 platforms. It 2516backend. Otherwise it will be enabled on non-win32 platforms. It
… …
2252be detected at runtime. 2553be detected at runtime.
2253 2554
2254=item EV_H 2555=item EV_H
2255 2556
2256The name of the F<ev.h> header file used to include it. The default if 2557The name of the F<ev.h> header file used to include it. The default if
2257undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This 2558undefined is C<"ev.h"> in F<event.h> and F<ev.c>. This can be used to
2258can be used to virtually rename the F<ev.h> header file in case of conflicts. 2559virtually rename the F<ev.h> header file in case of conflicts.
2259 2560
2260=item EV_CONFIG_H 2561=item EV_CONFIG_H
2261 2562
2262If C<EV_STANDALONE> isn't C<1>, this variable can be used to override 2563If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2263F<ev.c>'s idea of where to find the F<config.h> file, similarly to 2564F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2264C<EV_H>, above. 2565C<EV_H>, above.
2265 2566
2266=item EV_EVENT_H 2567=item EV_EVENT_H
2267 2568
2268Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea 2569Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2269of how the F<event.h> header can be found. 2570of how the F<event.h> header can be found, the dfeault is C<"event.h">.
2270 2571
2271=item EV_PROTOTYPES 2572=item EV_PROTOTYPES
2272 2573
2273If defined to be C<0>, then F<ev.h> will not define any function 2574If defined to be C<0>, then F<ev.h> will not define any function
2274prototypes, but still define all the structs and other symbols. This is 2575prototypes, but still define all the structs and other symbols. This is
… …
2340than enough. If you need to manage thousands of children you might want to 2641than enough. If you need to manage thousands of children you might want to
2341increase this value (I<must> be a power of two). 2642increase this value (I<must> be a power of two).
2342 2643
2343=item EV_INOTIFY_HASHSIZE 2644=item EV_INOTIFY_HASHSIZE
2344 2645
2345C<ev_staz> watchers use a small hash table to distribute workload by 2646C<ev_stat> watchers use a small hash table to distribute workload by
2346inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>), 2647inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
2347usually more than enough. If you need to manage thousands of C<ev_stat> 2648usually more than enough. If you need to manage thousands of C<ev_stat>
2348watchers you might want to increase this value (I<must> be a power of 2649watchers you might want to increase this value (I<must> be a power of
2349two). 2650two).
2350 2651
… …
2367 2668
2368=item ev_set_cb (ev, cb) 2669=item ev_set_cb (ev, cb)
2369 2670
2370Can be used to change the callback member declaration in each watcher, 2671Can be used to change the callback member declaration in each watcher,
2371and the way callbacks are invoked and set. Must expand to a struct member 2672and the way callbacks are invoked and set. Must expand to a struct member
2372definition and a statement, respectively. See the F<ev.v> header file for 2673definition and a statement, respectively. See the F<ev.h> header file for
2373their default definitions. One possible use for overriding these is to 2674their default definitions. One possible use for overriding these is to
2374avoid the C<struct ev_loop *> as first argument in all cases, or to use 2675avoid the C<struct ev_loop *> as first argument in all cases, or to use
2375method calls instead of plain function calls in C++. 2676method calls instead of plain function calls in C++.
2677
2678=head2 EXPORTED API SYMBOLS
2679
2680If you need to re-export the API (e.g. via a dll) and you need a list of
2681exported symbols, you can use the provided F<Symbol.*> files which list
2682all public symbols, one per line:
2683
2684 Symbols.ev for libev proper
2685 Symbols.event for the libevent emulation
2686
2687This can also be used to rename all public symbols to avoid clashes with
2688multiple versions of libev linked together (which is obviously bad in
2689itself, but sometimes it is inconvinient to avoid this).
2690
2691A sed command like this will create wrapper C<#define>'s that you need to
2692include before including F<ev.h>:
2693
2694 <Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
2695
2696This would create a file F<wrap.h> which essentially looks like this:
2697
2698 #define ev_backend myprefix_ev_backend
2699 #define ev_check_start myprefix_ev_check_start
2700 #define ev_check_stop myprefix_ev_check_stop
2701 ...
2376 2702
2377=head2 EXAMPLES 2703=head2 EXAMPLES
2378 2704
2379For a real-world example of a program the includes libev 2705For a real-world example of a program the includes libev
2380verbatim, you can have a look at the EV perl module 2706verbatim, you can have a look at the EV perl module
… …
2421 2747
2422=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers) 2748=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2423 2749
2424This means that, when you have a watcher that triggers in one hour and 2750This means that, when you have a watcher that triggers in one hour and
2425there are 100 watchers that would trigger before that then inserting will 2751there are 100 watchers that would trigger before that then inserting will
2426have to skip those 100 watchers. 2752have to skip roughly seven (C<ld 100>) of these watchers.
2427 2753
2428=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers) 2754=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2429 2755
2430That means that for changing a timer costs less than removing/adding them 2756That means that changing a timer costs less than removing/adding them
2431as only the relative motion in the event queue has to be paid for. 2757as only the relative motion in the event queue has to be paid for.
2432 2758
2433=item Starting io/check/prepare/idle/signal/child watchers: O(1) 2759=item Starting io/check/prepare/idle/signal/child watchers: O(1)
2434 2760
2435These just add the watcher into an array or at the head of a list. 2761These just add the watcher into an array or at the head of a list.
2762
2436=item Stopping check/prepare/idle watchers: O(1) 2763=item Stopping check/prepare/idle watchers: O(1)
2437 2764
2438=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE)) 2765=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2439 2766
2440These watchers are stored in lists then need to be walked to find the 2767These watchers are stored in lists then need to be walked to find the
2441correct watcher to remove. The lists are usually short (you don't usually 2768correct watcher to remove. The lists are usually short (you don't usually
2442have many watchers waiting for the same fd or signal). 2769have many watchers waiting for the same fd or signal).
2443 2770
2444=item Finding the next timer per loop iteration: O(1) 2771=item Finding the next timer in each loop iteration: O(1)
2772
2773By virtue of using a binary heap, the next timer is always found at the
2774beginning of the storage array.
2445 2775
2446=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd) 2776=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2447 2777
2448A change means an I/O watcher gets started or stopped, which requires 2778A change means an I/O watcher gets started or stopped, which requires
2449libev to recalculate its status (and possibly tell the kernel). 2779libev to recalculate its status (and possibly tell the kernel, depending
2780on backend and wether C<ev_io_set> was used).
2450 2781
2451=item Activating one watcher: O(1) 2782=item Activating one watcher (putting it into the pending state): O(1)
2452 2783
2453=item Priority handling: O(number_of_priorities) 2784=item Priority handling: O(number_of_priorities)
2454 2785
2455Priorities are implemented by allocating some space for each 2786Priorities are implemented by allocating some space for each
2456priority. When doing priority-based operations, libev usually has to 2787priority. When doing priority-based operations, libev usually has to
2457linearly search all the priorities. 2788linearly search all the priorities, but starting/stopping and activating
2789watchers becomes O(1) w.r.t. prioritiy handling.
2458 2790
2459=back 2791=back
2460 2792
2461 2793
2794=head1 Win32 platform limitations and workarounds
2795
2796Win32 doesn't support any of the standards (e.g. POSIX) that libev
2797requires, and its I/O model is fundamentally incompatible with the POSIX
2798model. Libev still offers limited functionality on this platform in
2799the form of the C<EVBACKEND_SELECT> backend, and only supports socket
2800descriptors. This only applies when using Win32 natively, not when using
2801e.g. cygwin.
2802
2803There is no supported compilation method available on windows except
2804embedding it into other applications.
2805
2806Due to the many, low, and arbitrary limits on the win32 platform and the
2807abysmal performance of winsockets, using a large number of sockets is not
2808recommended (and not reasonable). If your program needs to use more than
2809a hundred or so sockets, then likely it needs to use a totally different
2810implementation for windows, as libev offers the POSIX model, which cannot
2811be implemented efficiently on windows (microsoft monopoly games).
2812
2813=over 4
2814
2815=item The winsocket select function
2816
2817The winsocket C<select> function doesn't follow POSIX in that it requires
2818socket I<handles> and not socket I<file descriptors>. This makes select
2819very inefficient, and also requires a mapping from file descriptors
2820to socket handles. See the discussion of the C<EV_SELECT_USE_FD_SET>,
2821C<EV_SELECT_IS_WINSOCKET> and C<EV_FD_TO_WIN32_HANDLE> preprocessor
2822symbols for more info.
2823
2824The configuration for a "naked" win32 using the microsoft runtime
2825libraries and raw winsocket select is:
2826
2827 #define EV_USE_SELECT 1
2828 #define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
2829
2830Note that winsockets handling of fd sets is O(n), so you can easily get a
2831complexity in the O(n²) range when using win32.
2832
2833=item Limited number of file descriptors
2834
2835Windows has numerous arbitrary (and low) limits on things. Early versions
2836of winsocket's select only supported waiting for a max. of C<64> handles
2837(probably owning to the fact that all windows kernels can only wait for
2838C<64> things at the same time internally; microsoft recommends spawning a
2839chain of threads and wait for 63 handles and the previous thread in each).
2840
2841Newer versions support more handles, but you need to define C<FD_SETSIZE>
2842to some high number (e.g. C<2048>) before compiling the winsocket select
2843call (which might be in libev or elsewhere, for example, perl does its own
2844select emulation on windows).
2845
2846Another limit is the number of file descriptors in the microsoft runtime
2847libraries, which by default is C<64> (there must be a hidden I<64> fetish
2848or something like this inside microsoft). You can increase this by calling
2849C<_setmaxstdio>, which can increase this limit to C<2048> (another
2850arbitrary limit), but is broken in many versions of the microsoft runtime
2851libraries.
2852
2853This might get you to about C<512> or C<2048> sockets (depending on
2854windows version and/or the phase of the moon). To get more, you need to
2855wrap all I/O functions and provide your own fd management, but the cost of
2856calling select (O(n²)) will likely make this unworkable.
2857
2858=back
2859
2860
2462=head1 AUTHOR 2861=head1 AUTHOR
2463 2862
2464Marc Lehmann <libev@schmorp.de>. 2863Marc Lehmann <libev@schmorp.de>.
2465 2864

Diff Legend

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