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

Comparing libev/ev.pod (file contents):
Revision 1.85 by root, Mon Dec 17 07:24:12 2007 UTC vs.
Revision 1.136 by root, Thu Mar 13 13:06:16 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 // a single header file is required
11 #include <ev.h> 12 #include <ev.h>
12 13
14 // every watcher type has its own typedef'd struct
15 // with the name ev_<type>
13 ev_io stdin_watcher; 16 ev_io stdin_watcher;
14 ev_timer timeout_watcher; 17 ev_timer timeout_watcher;
15 18
19 // all watcher callbacks have a similar signature
16 /* called when data readable on stdin */ 20 // this callback is called when data is readable on stdin
17 static void 21 static void
18 stdin_cb (EV_P_ struct ev_io *w, int revents) 22 stdin_cb (EV_P_ struct ev_io *w, int revents)
19 { 23 {
20 /* puts ("stdin ready"); */ 24 puts ("stdin ready");
21 ev_io_stop (EV_A_ w); /* just a syntax example */ 25 // for one-shot events, one must manually stop the watcher
22 ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */ 26 // with its corresponding stop function.
27 ev_io_stop (EV_A_ w);
28
29 // this causes all nested ev_loop's to stop iterating
30 ev_unloop (EV_A_ EVUNLOOP_ALL);
23 } 31 }
24 32
33 // another callback, this time for a time-out
25 static void 34 static void
26 timeout_cb (EV_P_ struct ev_timer *w, int revents) 35 timeout_cb (EV_P_ struct ev_timer *w, int revents)
27 { 36 {
28 /* puts ("timeout"); */ 37 puts ("timeout");
29 ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */ 38 // this causes the innermost ev_loop to stop iterating
39 ev_unloop (EV_A_ EVUNLOOP_ONE);
30 } 40 }
31 41
32 int 42 int
33 main (void) 43 main (void)
34 { 44 {
45 // use the default event loop unless you have special needs
35 struct ev_loop *loop = ev_default_loop (0); 46 struct ev_loop *loop = ev_default_loop (0);
36 47
37 /* initialise an io watcher, then start it */ 48 // initialise an io watcher, then start it
49 // this one will watch for stdin to become readable
38 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ); 50 ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
39 ev_io_start (loop, &stdin_watcher); 51 ev_io_start (loop, &stdin_watcher);
40 52
53 // initialise a timer watcher, then start it
41 /* simple non-repeating 5.5 second timeout */ 54 // simple non-repeating 5.5 second timeout
42 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.); 55 ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
43 ev_timer_start (loop, &timeout_watcher); 56 ev_timer_start (loop, &timeout_watcher);
44 57
45 /* loop till timeout or data ready */ 58 // now wait for events to arrive
46 ev_loop (loop, 0); 59 ev_loop (loop, 0);
47 60
61 // unloop was called, so exit
48 return 0; 62 return 0;
49 } 63 }
50 64
51=head1 DESCRIPTION 65=head1 DESCRIPTION
52 66
53The newest version of this document is also available as a html-formatted 67The newest version of this document is also available as an html-formatted
54web page you might find easier to navigate when reading it for the first 68web page you might find easier to navigate when reading it for the first
55time: L<http://cvs.schmorp.de/libev/ev.html>. 69time: L<http://cvs.schmorp.de/libev/ev.html>.
56 70
57Libev is an event loop: you register interest in certain events (such as a 71Libev 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 72file descriptor being readable or a timeout occurring), and it will manage
59these event sources and provide your program with events. 73these event sources and provide your program with events.
60 74
61To do this, it must take more or less complete control over your process 75To 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 76(or thread) by executing the I<event loop> handler, and will then
63communicate events via a callback mechanism. 77communicate events via a callback mechanism.
65You register interest in certain events by registering so-called I<event 79You register interest in certain events by registering so-called I<event
66watchers>, which are relatively small C structures you initialise with the 80watchers>, 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 81details of the event, and then hand it over to libev by I<starting> the
68watcher. 82watcher.
69 83
70=head1 FEATURES 84=head2 FEATURES
71 85
72Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the 86Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
73BSD-specific C<kqueue> and the Solaris-specific event port mechanisms 87BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
74for file descriptor events (C<ev_io>), the Linux C<inotify> interface 88for file descriptor events (C<ev_io>), the Linux C<inotify> interface
75(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers 89(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers
82 96
83It also is quite fast (see this 97It also is quite fast (see this
84L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent 98L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
85for example). 99for example).
86 100
87=head1 CONVENTIONS 101=head2 CONVENTIONS
88 102
89Libev is very configurable. In this manual the default configuration will 103Libev is very configurable. In this manual the default (and most common)
90be described, which supports multiple event loops. For more info about 104configuration will be described, which supports multiple event loops. For
91various configuration options please have a look at B<EMBED> section in 105more info about various configuration options please have a look at
92this manual. If libev was configured without support for multiple event 106B<EMBED> section in this manual. If libev was configured without support
93loops, then all functions taking an initial argument of name C<loop> 107for multiple event loops, then all functions taking an initial argument of
94(which is always of type C<struct ev_loop *>) will not have this argument. 108name C<loop> (which is always of type C<struct ev_loop *>) will not have
109this argument.
95 110
96=head1 TIME REPRESENTATION 111=head2 TIME REPRESENTATION
97 112
98Libev represents time as a single floating point number, representing the 113Libev represents time as a single floating point number, representing the
99(fractional) number of seconds since the (POSIX) epoch (somewhere near 114(fractional) number of seconds since the (POSIX) epoch (somewhere near
100the beginning of 1970, details are complicated, don't ask). This type is 115the 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 116called 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 117to the C<double> type in C, and when you need to do any calculations on
103it, you should treat it as such. 118it, you should treat it as some floatingpoint value. Unlike the name
119component C<stamp> might indicate, it is also used for time differences
120throughout libev.
104 121
105=head1 GLOBAL FUNCTIONS 122=head1 GLOBAL FUNCTIONS
106 123
107These functions can be called anytime, even before initialising the 124These functions can be called anytime, even before initialising the
108library in any way. 125library in any way.
112=item ev_tstamp ev_time () 129=item ev_tstamp ev_time ()
113 130
114Returns the current time as libev would use it. Please note that the 131Returns 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 132C<ev_now> function is usually faster and also often returns the timestamp
116you actually want to know. 133you actually want to know.
134
135=item ev_sleep (ev_tstamp interval)
136
137Sleep for the given interval: The current thread will be blocked until
138either it is interrupted or the given time interval has passed. Basically
139this is a subsecond-resolution C<sleep ()>.
117 140
118=item int ev_version_major () 141=item int ev_version_major ()
119 142
120=item int ev_version_minor () 143=item int ev_version_minor ()
121 144
252flags. If that is troubling you, check C<ev_backend ()> afterwards). 275flags. If that is troubling you, check C<ev_backend ()> afterwards).
253 276
254If you don't know what event loop to use, use the one returned from this 277If you don't know what event loop to use, use the one returned from this
255function. 278function.
256 279
280The default loop is the only loop that can handle C<ev_signal> and
281C<ev_child> watchers, and to do this, it always registers a handler
282for C<SIGCHLD>. If this is a problem for your app you can either
283create a dynamic loop with C<ev_loop_new> that doesn't do that, or you
284can simply overwrite the C<SIGCHLD> signal handler I<after> calling
285C<ev_default_init>.
286
257The flags argument can be used to specify special behaviour or specific 287The 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>). 288backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
259 289
260The following flags are supported: 290The following flags are supported:
261 291
282enabling this flag. 312enabling this flag.
283 313
284This works by calling C<getpid ()> on every iteration of the loop, 314This works by calling C<getpid ()> on every iteration of the loop,
285and thus this might slow down your event loop if you do a lot of loop 315and thus this might slow down your event loop if you do a lot of loop
286iterations and little real work, but is usually not noticeable (on my 316iterations and little real work, but is usually not noticeable (on my
287Linux system for example, C<getpid> is actually a simple 5-insn sequence 317GNU/Linux system for example, C<getpid> is actually a simple 5-insn sequence
288without a syscall and thus I<very> fast, but my Linux system also has 318without a syscall and thus I<very> fast, but my GNU/Linux system also has
289C<pthread_atfork> which is even faster). 319C<pthread_atfork> which is even faster).
290 320
291The big advantage of this flag is that you can forget about fork (and 321The big advantage of this flag is that you can forget about fork (and
292forget about forgetting to tell libev about forking) when you use this 322forget about forgetting to tell libev about forking) when you use this
293flag. 323flag.
298=item C<EVBACKEND_SELECT> (value 1, portable select backend) 328=item C<EVBACKEND_SELECT> (value 1, portable select backend)
299 329
300This is your standard select(2) backend. Not I<completely> standard, as 330This 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, 331libev 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 332but 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 333using this backend. It doesn't scale too well (O(highest_fd)), but its
304the fastest backend for a low number of fds. 334usually the fastest backend for a low number of (low-numbered :) fds.
335
336To get good performance out of this backend you need a high amount of
337parallelity (most of the file descriptors should be busy). If you are
338writing a server, you should C<accept ()> in a loop to accept as many
339connections as possible during one iteration. You might also want to have
340a look at C<ev_set_io_collect_interval ()> to increase the amount of
341readyness notifications you get per iteration.
305 342
306=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows) 343=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
307 344
308And this is your standard poll(2) backend. It's more complicated than 345And this is your standard poll(2) backend. It's more complicated
309select, but handles sparse fds better and has no artificial limit on the 346than select, but handles sparse fds better and has no artificial
310number of fds you can use (except it will slow down considerably with a 347limit 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). 348considerably with a lot of inactive fds). It scales similarly to select,
349i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
350performance tips.
312 351
313=item C<EVBACKEND_EPOLL> (value 4, Linux) 352=item C<EVBACKEND_EPOLL> (value 4, Linux)
314 353
315For few fds, this backend is a bit little slower than poll and select, 354For 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 355but 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 356like O(total_fds) where n is the total number of fds (or the highest fd),
318either O(1) or O(active_fds). 357epoll scales either O(1) or O(active_fds). The epoll design has a number
358of shortcomings, such as silently dropping events in some hard-to-detect
359cases and rewiring a syscall per fd change, no fork support and bad
360support for dup.
319 361
320While stopping and starting an I/O watcher in the same iteration will 362While stopping, setting and starting an I/O watcher in the same iteration
321result in some caching, there is still a syscall per such incident 363will 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 364(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 365best to avoid that. Also, C<dup ()>'ed file descriptors might not work
324well if you register events for both fds. 366very well if you register events for both fds.
325 367
326Please note that epoll sometimes generates spurious notifications, so you 368Please 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 369need to use non-blocking I/O or other means to avoid blocking when no data
328(or space) is available. 370(or space) is available.
329 371
372Best performance from this backend is achieved by not unregistering all
373watchers for a file descriptor until it has been closed, if possible, i.e.
374keep at least one watcher active per fd at all times.
375
376While nominally embeddeble in other event loops, this feature is broken in
377all kernel versions tested so far.
378
330=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones) 379=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
331 380
332Kqueue deserves special mention, as at the time of this writing, it 381Kqueue deserves special mention, as at the time of this writing, it
333was broken on all BSDs except NetBSD (usually it doesn't work with 382was broken on all BSDs except NetBSD (usually it doesn't work reliably
334anything but sockets and pipes, except on Darwin, where of course its 383with anything but sockets and pipes, except on Darwin, where of course
335completely useless). For this reason its not being "autodetected" 384it's completely useless). For this reason it's not being "autodetected"
336unless you explicitly specify it explicitly in the flags (i.e. using 385unless you explicitly specify it explicitly in the flags (i.e. using
337C<EVBACKEND_KQUEUE>). 386C<EVBACKEND_KQUEUE>) or libev was compiled on a known-to-be-good (-enough)
387system like NetBSD.
388
389You still can embed kqueue into a normal poll or select backend and use it
390only for sockets (after having made sure that sockets work with kqueue on
391the target platform). See C<ev_embed> watchers for more info.
338 392
339It scales in the same way as the epoll backend, but the interface to the 393It 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 394kernel is more efficient (which says nothing about its actual speed, of
341course). While starting and stopping an I/O watcher does not cause an 395course). 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 396cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to
343incident, so its best to avoid that. 397two event changes per incident, support for C<fork ()> is very bad and it
398drops fds silently in similarly hard-to-detect cases.
399
400This backend usually performs well under most conditions.
401
402While nominally embeddable in other event loops, this doesn't work
403everywhere, so you might need to test for this. And since it is broken
404almost everywhere, you should only use it when you have a lot of sockets
405(for which it usually works), by embedding it into another event loop
406(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and using it only for
407sockets.
344 408
345=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8) 409=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
346 410
347This is not implemented yet (and might never be). 411This is not implemented yet (and might never be, unless you send me an
412implementation). According to reports, C</dev/poll> only supports sockets
413and is not embeddable, which would limit the usefulness of this backend
414immensely.
348 415
349=item C<EVBACKEND_PORT> (value 32, Solaris 10) 416=item C<EVBACKEND_PORT> (value 32, Solaris 10)
350 417
351This uses the Solaris 10 port mechanism. As with everything on Solaris, 418This 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)). 419it's really slow, but it still scales very well (O(active_fds)).
353 420
354Please note that solaris ports can result in a lot of spurious 421Please 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 422notifications, so you need to use non-blocking I/O or other means to avoid
356blocking when no data (or space) is available. 423blocking when no data (or space) is available.
424
425While this backend scales well, it requires one system call per active
426file descriptor per loop iteration. For small and medium numbers of file
427descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
428might perform better.
429
430On the positive side, ignoring the spurious readyness notifications, this
431backend actually performed to specification in all tests and is fully
432embeddable, which is a rare feat among the OS-specific backends.
357 433
358=item C<EVBACKEND_ALL> 434=item C<EVBACKEND_ALL>
359 435
360Try all backends (even potentially broken ones that wouldn't be tried 436Try 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 437with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
362C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>. 438C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
363 439
440It is definitely not recommended to use this flag.
441
364=back 442=back
365 443
366If one or more of these are ored into the flags value, then only these 444If 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 445backends will be tried (in the reverse order as listed here). If none are
368specified, most compiled-in backend will be tried, usually in reverse 446specified, all backends in C<ev_recommended_backends ()> will be tried.
369order of their flag values :)
370 447
371The most typical usage is like this: 448The most typical usage is like this:
372 449
373 if (!ev_default_loop (0)) 450 if (!ev_default_loop (0))
374 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?"); 451 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
402Destroys the default loop again (frees all memory and kernel state 479Destroys the default loop again (frees all memory and kernel state
403etc.). None of the active event watchers will be stopped in the normal 480etc.). 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 481sense, so e.g. C<ev_is_active> might still return true. It is your
405responsibility to either stop all watchers cleanly yoursef I<before> 482responsibility to either stop all watchers cleanly yoursef I<before>
406calling this function, or cope with the fact afterwards (which is usually 483calling 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 484the easiest thing, you can just ignore the watchers and/or C<free ()> them
408for example). 485for example).
486
487Note that certain global state, such as signal state, will not be freed by
488this function, and related watchers (such as signal and child watchers)
489would need to be stopped manually.
490
491In general it is not advisable to call this function except in the
492rare occasion where you really need to free e.g. the signal handling
493pipe fds. If you need dynamically allocated loops it is better to use
494C<ev_loop_new> and C<ev_loop_destroy>).
409 495
410=item ev_loop_destroy (loop) 496=item ev_loop_destroy (loop)
411 497
412Like C<ev_default_destroy>, but destroys an event loop created by an 498Like C<ev_default_destroy>, but destroys an event loop created by an
413earlier call to C<ev_loop_new>. 499earlier call to C<ev_loop_new>.
414 500
415=item ev_default_fork () 501=item ev_default_fork ()
416 502
503This function sets a flag that causes subsequent C<ev_loop> iterations
417This function reinitialises the kernel state for backends that have 504to 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 505name, 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 506the child process (or both child and parent, but that again makes little
420again makes little sense). 507sense). You I<must> call it in the child before using any of the libev
508functions, and it will only take effect at the next C<ev_loop> iteration.
421 509
422You I<must> call this function in the child process after forking if and 510On 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 511process if and only if you want to use the event library in the child. If
424fork+exec, you don't have to call it. 512you just fork+exec, you don't have to call it at all.
425 513
426The function itself is quite fast and it's usually not a problem to call 514The 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 515it just in case after a fork. To make this easy, the function will fit in
428quite nicely into a call to C<pthread_atfork>: 516quite nicely into a call to C<pthread_atfork>:
429 517
430 pthread_atfork (0, 0, ev_default_fork); 518 pthread_atfork (0, 0, ev_default_fork);
431 519
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) 520=item ev_loop_fork (loop)
437 521
438Like C<ev_default_fork>, but acts on an event loop created by 522Like 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 523C<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. 524after fork, and how you do this is entirely your own problem.
525
526=item int ev_is_default_loop (loop)
527
528Returns true when the given loop actually is the default loop, false otherwise.
441 529
442=item unsigned int ev_loop_count (loop) 530=item unsigned int ev_loop_count (loop)
443 531
444Returns the count of loop iterations for the loop, which is identical to 532Returns 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 533the number of times libev did poll for new events. It starts at C<0> and
458 546
459Returns the current "event loop time", which is the time the event loop 547Returns the current "event loop time", which is the time the event loop
460received events and started processing them. This timestamp does not 548received events and started processing them. This timestamp does not
461change as long as callbacks are being processed, and this is also the base 549change 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 550time used for relative timers. You can treat it as the timestamp of the
463event occuring (or more correctly, libev finding out about it). 551event occurring (or more correctly, libev finding out about it).
464 552
465=item ev_loop (loop, int flags) 553=item ev_loop (loop, int flags)
466 554
467Finally, this is it, the event handler. This function usually is called 555Finally, this is it, the event handler. This function usually is called
468after you initialised all your watchers and you want to start handling 556after you initialised all your watchers and you want to start handling
490usually a better approach for this kind of thing. 578usually a better approach for this kind of thing.
491 579
492Here are the gory details of what C<ev_loop> does: 580Here are the gory details of what C<ev_loop> does:
493 581
494 - Before the first iteration, call any pending watchers. 582 - Before the first iteration, call any pending watchers.
495 * If there are no active watchers (reference count is zero), return. 583 * If EVFLAG_FORKCHECK was used, check for a fork.
496 - Queue all prepare watchers and then call all outstanding watchers. 584 - If a fork was detected, queue and call all fork watchers.
585 - Queue and call all prepare watchers.
497 - If we have been forked, recreate the kernel state. 586 - If we have been forked, recreate the kernel state.
498 - Update the kernel state with all outstanding changes. 587 - Update the kernel state with all outstanding changes.
499 - Update the "event loop time". 588 - Update the "event loop time".
500 - Calculate for how long to block. 589 - Calculate for how long to sleep or block, if at all
590 (active idle watchers, EVLOOP_NONBLOCK or not having
591 any active watchers at all will result in not sleeping).
592 - Sleep if the I/O and timer collect interval say so.
501 - Block the process, waiting for any events. 593 - Block the process, waiting for any events.
502 - Queue all outstanding I/O (fd) events. 594 - Queue all outstanding I/O (fd) events.
503 - Update the "event loop time" and do time jump handling. 595 - Update the "event loop time" and do time jump handling.
504 - Queue all outstanding timers. 596 - Queue all outstanding timers.
505 - Queue all outstanding periodics. 597 - Queue all outstanding periodics.
506 - If no events are pending now, queue all idle watchers. 598 - If no events are pending now, queue all idle watchers.
507 - Queue all check watchers. 599 - Queue all check watchers.
508 - Call all queued watchers in reverse order (i.e. check watchers first). 600 - 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 601 Signals and child watchers are implemented as I/O watchers, and will
510 be handled here by queueing them when their watcher gets executed. 602 be handled here by queueing them when their watcher gets executed.
511 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK 603 - If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
512 were used, return, otherwise continue with step *. 604 were used, or there are no active watchers, return, otherwise
605 continue with step *.
513 606
514Example: Queue some jobs and then loop until no events are outsanding 607Example: Queue some jobs and then loop until no events are outstanding
515anymore. 608anymore.
516 609
517 ... queue jobs here, make sure they register event watchers as long 610 ... 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..) 611 ... as they still have work to do (even an idle watcher will do..)
519 ev_loop (my_loop, 0); 612 ev_loop (my_loop, 0);
523 616
524Can be used to make a call to C<ev_loop> return early (but only after it 617Can 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 618has 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 619C<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. 620C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
621
622This "unloop state" will be cleared when entering C<ev_loop> again.
528 623
529=item ev_ref (loop) 624=item ev_ref (loop)
530 625
531=item ev_unref (loop) 626=item ev_unref (loop)
532 627
537returning, ev_unref() after starting, and ev_ref() before stopping it. For 632returning, ev_unref() after starting, and ev_ref() before stopping it. For
538example, libev itself uses this for its internal signal pipe: It is not 633example, 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 634visible 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 635no 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 636way 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>. 637libraries. Just remember to I<unref after start> and I<ref before stop>
638(but only if the watcher wasn't active before, or was active before,
639respectively).
543 640
544Example: Create a signal watcher, but keep it from keeping C<ev_loop> 641Example: Create a signal watcher, but keep it from keeping C<ev_loop>
545running when nothing else is active. 642running when nothing else is active.
546 643
547 struct ev_signal exitsig; 644 struct ev_signal exitsig;
551 648
552Example: For some weird reason, unregister the above signal handler again. 649Example: For some weird reason, unregister the above signal handler again.
553 650
554 ev_ref (loop); 651 ev_ref (loop);
555 ev_signal_stop (loop, &exitsig); 652 ev_signal_stop (loop, &exitsig);
653
654=item ev_set_io_collect_interval (loop, ev_tstamp interval)
655
656=item ev_set_timeout_collect_interval (loop, ev_tstamp interval)
657
658These advanced functions influence the time that libev will spend waiting
659for events. Both are by default C<0>, meaning that libev will try to
660invoke timer/periodic callbacks and I/O callbacks with minimum latency.
661
662Setting these to a higher value (the C<interval> I<must> be >= C<0>)
663allows libev to delay invocation of I/O and timer/periodic callbacks to
664increase efficiency of loop iterations.
665
666The background is that sometimes your program runs just fast enough to
667handle one (or very few) event(s) per loop iteration. While this makes
668the program responsive, it also wastes a lot of CPU time to poll for new
669events, especially with backends like C<select ()> which have a high
670overhead for the actual polling but can deliver many events at once.
671
672By setting a higher I<io collect interval> you allow libev to spend more
673time collecting I/O events, so you can handle more events per iteration,
674at the cost of increasing latency. Timeouts (both C<ev_periodic> and
675C<ev_timer>) will be not affected. Setting this to a non-null value will
676introduce an additional C<ev_sleep ()> call into most loop iterations.
677
678Likewise, by setting a higher I<timeout collect interval> you allow libev
679to spend more time collecting timeouts, at the expense of increased
680latency (the watcher callback will be called later). C<ev_io> watchers
681will not be affected. Setting this to a non-null value will not introduce
682any overhead in libev.
683
684Many (busy) programs can usually benefit by setting the io collect
685interval to a value near C<0.1> or so, which is often enough for
686interactive servers (of course not for games), likewise for timeouts. It
687usually doesn't make much sense to set it to a lower value than C<0.01>,
688as this approsaches the timing granularity of most systems.
556 689
557=back 690=back
558 691
559 692
560=head1 ANATOMY OF A WATCHER 693=head1 ANATOMY OF A WATCHER
659 792
660=item C<EV_FORK> 793=item C<EV_FORK>
661 794
662The event loop has been resumed in the child process after fork (see 795The event loop has been resumed in the child process after fork (see
663C<ev_fork>). 796C<ev_fork>).
797
798=item C<EV_ASYNC>
799
800The given async watcher has been asynchronously notified (see C<ev_async>).
664 801
665=item C<EV_ERROR> 802=item C<EV_ERROR>
666 803
667An unspecified error has occured, the watcher has been stopped. This might 804An unspecified error has occured, the watcher has been stopped. This might
668happen because the watcher could not be properly started because libev 805happen because the watcher could not be properly started because libev
886In general you can register as many read and/or write event watchers per 1023In 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 1024fd 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 1025descriptors to non-blocking mode is also usually a good idea (but not
889required if you know what you are doing). 1026required if you know what you are doing).
890 1027
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 1028If 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 1029(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
899C<EVBACKEND_POLL>). 1030C<EVBACKEND_POLL>).
900 1031
901Another thing you have to watch out for is that it is quite easy to 1032Another thing you have to watch out for is that it is quite easy to
913such as poll (fortunately in our Xlib example, Xlib already does this on 1044such as poll (fortunately in our Xlib example, Xlib already does this on
914its own, so its quite safe to use). 1045its own, so its quite safe to use).
915 1046
916=head3 The special problem of disappearing file descriptors 1047=head3 The special problem of disappearing file descriptors
917 1048
918Some backends (e.g kqueue, epoll) need to be told about closing a file 1049Some backends (e.g. kqueue, epoll) need to be told about closing a file
919descriptor (either by calling C<close> explicitly or by any other means, 1050descriptor (either by calling C<close> explicitly or by any other means,
920such as C<dup>). The reason is that you register interest in some file 1051such as C<dup>). The reason is that you register interest in some file
921descriptor, but when it goes away, the operating system will silently drop 1052descriptor, but when it goes away, the operating system will silently drop
922this interest. If another file descriptor with the same number then is 1053this interest. If another file descriptor with the same number then is
923registered with libev, there is no efficient way to see that this is, in 1054registered with libev, there is no efficient way to see that this is, in
932 1063
933This is how one would do it normally anyway, the important point is that 1064This is how one would do it normally anyway, the important point is that
934the libev application should not optimise around libev but should leave 1065the libev application should not optimise around libev but should leave
935optimisations to libev. 1066optimisations to libev.
936 1067
1068=head3 The special problem of dup'ed file descriptors
1069
1070Some backends (e.g. epoll), cannot register events for file descriptors,
1071but only events for the underlying file descriptions. That means when you
1072have C<dup ()>'ed file descriptors or weirder constellations, and register
1073events for them, only one file descriptor might actually receive events.
1074
1075There is no workaround possible except not registering events
1076for potentially C<dup ()>'ed file descriptors, or to resort to
1077C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1078
1079=head3 The special problem of fork
1080
1081Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
1082useless behaviour. Libev fully supports fork, but needs to be told about
1083it in the child.
1084
1085To support fork in your programs, you either have to call
1086C<ev_default_fork ()> or C<ev_loop_fork ()> after a fork in the child,
1087enable C<EVFLAG_FORKCHECK>, or resort to C<EVBACKEND_SELECT> or
1088C<EVBACKEND_POLL>.
1089
937 1090
938=head3 Watcher-Specific Functions 1091=head3 Watcher-Specific Functions
939 1092
940=over 4 1093=over 4
941 1094
954=item int events [read-only] 1107=item int events [read-only]
955 1108
956The events being watched. 1109The events being watched.
957 1110
958=back 1111=back
1112
1113=head3 Examples
959 1114
960Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well 1115Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
961readable, but only once. Since it is likely line-buffered, you could 1116readable, but only once. Since it is likely line-buffered, you could
962attempt to read a whole line in the callback. 1117attempt to read a whole line in the callback.
963 1118
1016configure a timer to trigger every 10 seconds, then it will trigger at 1171configure a timer to trigger every 10 seconds, then it will trigger at
1017exactly 10 second intervals. If, however, your program cannot keep up with 1172exactly 10 second intervals. If, however, your program cannot keep up with
1018the timer (because it takes longer than those 10 seconds to do stuff) the 1173the timer (because it takes longer than those 10 seconds to do stuff) the
1019timer will not fire more than once per event loop iteration. 1174timer will not fire more than once per event loop iteration.
1020 1175
1021=item ev_timer_again (loop) 1176=item ev_timer_again (loop, ev_timer *)
1022 1177
1023This will act as if the timer timed out and restart it again if it is 1178This will act as if the timer timed out and restart it again if it is
1024repeating. The exact semantics are: 1179repeating. The exact semantics are:
1025 1180
1026If the timer is pending, its pending status is cleared. 1181If the timer is pending, its pending status is cleared.
1061or C<ev_timer_again> is called and determines the next timeout (if any), 1216or C<ev_timer_again> is called and determines the next timeout (if any),
1062which is also when any modifications are taken into account. 1217which is also when any modifications are taken into account.
1063 1218
1064=back 1219=back
1065 1220
1221=head3 Examples
1222
1066Example: Create a timer that fires after 60 seconds. 1223Example: Create a timer that fires after 60 seconds.
1067 1224
1068 static void 1225 static void
1069 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents) 1226 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1070 { 1227 {
1133In this configuration the watcher triggers an event at the wallclock time 1290In this configuration the watcher triggers an event at the wallclock time
1134C<at> and doesn't repeat. It will not adjust when a time jump occurs, 1291C<at> and doesn't repeat. It will not adjust when a time jump occurs,
1135that is, if it is to be run at January 1st 2011 then it will run when the 1292that is, if it is to be run at January 1st 2011 then it will run when the
1136system time reaches or surpasses this time. 1293system time reaches or surpasses this time.
1137 1294
1138=item * non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0) 1295=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
1139 1296
1140In this mode the watcher will always be scheduled to time out at the next 1297In this mode the watcher will always be scheduled to time out at the next
1141C<at + N * interval> time (for some integer N, which can also be negative) 1298C<at + N * interval> time (for some integer N, which can also be negative)
1142and then repeat, regardless of any time jumps. 1299and then repeat, regardless of any time jumps.
1143 1300
1226 1383
1227When active, contains the absolute time that the watcher is supposed to 1384When active, contains the absolute time that the watcher is supposed to
1228trigger next. 1385trigger next.
1229 1386
1230=back 1387=back
1388
1389=head3 Examples
1231 1390
1232Example: Call a callback every hour, or, more precisely, whenever the 1391Example: Call a callback every hour, or, more precisely, whenever the
1233system clock is divisible by 3600. The callback invocation times have 1392system clock is divisible by 3600. The callback invocation times have
1234potentially a lot of jittering, but good long-term stability. 1393potentially a lot of jittering, but good long-term stability.
1235 1394
1275with the kernel (thus it coexists with your own signal handlers as long 1434with the kernel (thus it coexists with your own signal handlers as long
1276as you don't register any with libev). Similarly, when the last signal 1435as you don't register any with libev). Similarly, when the last signal
1277watcher for a signal is stopped libev will reset the signal handler to 1436watcher for a signal is stopped libev will reset the signal handler to
1278SIG_DFL (regardless of what it was set to before). 1437SIG_DFL (regardless of what it was set to before).
1279 1438
1439If possible and supported, libev will install its handlers with
1440C<SA_RESTART> behaviour enabled, so syscalls should not be unduly
1441interrupted. If you have a problem with syscalls getting interrupted by
1442signals you can block all signals in an C<ev_check> watcher and unblock
1443them in an C<ev_prepare> watcher.
1444
1280=head3 Watcher-Specific Functions and Data Members 1445=head3 Watcher-Specific Functions and Data Members
1281 1446
1282=over 4 1447=over 4
1283 1448
1284=item ev_signal_init (ev_signal *, callback, int signum) 1449=item ev_signal_init (ev_signal *, callback, int signum)
1292 1457
1293The signal the watcher watches out for. 1458The signal the watcher watches out for.
1294 1459
1295=back 1460=back
1296 1461
1462=head3 Examples
1463
1464Example: Try to exit cleanly on SIGINT and SIGTERM.
1465
1466 static void
1467 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1468 {
1469 ev_unloop (loop, EVUNLOOP_ALL);
1470 }
1471
1472 struct ev_signal signal_watcher;
1473 ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1474 ev_signal_start (loop, &sigint_cb);
1475
1297 1476
1298=head2 C<ev_child> - watch out for process status changes 1477=head2 C<ev_child> - watch out for process status changes
1299 1478
1300Child watchers trigger when your process receives a SIGCHLD in response to 1479Child watchers trigger when your process receives a SIGCHLD in response to
1301some child status changes (most typically when a child of yours dies). 1480some child status changes (most typically when a child of yours dies). It
1481is permissible to install a child watcher I<after> the child has been
1482forked (which implies it might have already exited), as long as the event
1483loop isn't entered (or is continued from a watcher).
1484
1485Only the default event loop is capable of handling signals, and therefore
1486you can only rgeister child watchers in the default event loop.
1487
1488=head3 Process Interaction
1489
1490Libev grabs C<SIGCHLD> as soon as the default event loop is
1491initialised. This is necessary to guarantee proper behaviour even if
1492the first child watcher is started after the child exits. The occurance
1493of C<SIGCHLD> is recorded asynchronously, but child reaping is done
1494synchronously as part of the event loop processing. Libev always reaps all
1495children, even ones not watched.
1496
1497=head3 Overriding the Built-In Processing
1498
1499Libev offers no special support for overriding the built-in child
1500processing, but if your application collides with libev's default child
1501handler, you can override it easily by installing your own handler for
1502C<SIGCHLD> after initialising the default loop, and making sure the
1503default loop never gets destroyed. You are encouraged, however, to use an
1504event-based approach to child reaping and thus use libev's support for
1505that, so other libev users can use C<ev_child> watchers freely.
1302 1506
1303=head3 Watcher-Specific Functions and Data Members 1507=head3 Watcher-Specific Functions and Data Members
1304 1508
1305=over 4 1509=over 4
1306 1510
1307=item ev_child_init (ev_child *, callback, int pid) 1511=item ev_child_init (ev_child *, callback, int pid, int trace)
1308 1512
1309=item ev_child_set (ev_child *, int pid) 1513=item ev_child_set (ev_child *, int pid, int trace)
1310 1514
1311Configures the watcher to wait for status changes of process C<pid> (or 1515Configures the watcher to wait for status changes of process C<pid> (or
1312I<any> process if C<pid> is specified as C<0>). The callback can look 1516I<any> process if C<pid> is specified as C<0>). The callback can look
1313at the C<rstatus> member of the C<ev_child> watcher structure to see 1517at the C<rstatus> member of the C<ev_child> watcher structure to see
1314the status word (use the macros from C<sys/wait.h> and see your systems 1518the status word (use the macros from C<sys/wait.h> and see your systems
1315C<waitpid> documentation). The C<rpid> member contains the pid of the 1519C<waitpid> documentation). The C<rpid> member contains the pid of the
1316process causing the status change. 1520process causing the status change. C<trace> must be either C<0> (only
1521activate the watcher when the process terminates) or C<1> (additionally
1522activate the watcher when the process is stopped or continued).
1317 1523
1318=item int pid [read-only] 1524=item int pid [read-only]
1319 1525
1320The process id this watcher watches out for, or C<0>, meaning any process id. 1526The process id this watcher watches out for, or C<0>, meaning any process id.
1321 1527
1328The process exit/trace status caused by C<rpid> (see your systems 1534The process exit/trace status caused by C<rpid> (see your systems
1329C<waitpid> and C<sys/wait.h> documentation for details). 1535C<waitpid> and C<sys/wait.h> documentation for details).
1330 1536
1331=back 1537=back
1332 1538
1333Example: Try to exit cleanly on SIGINT and SIGTERM. 1539=head3 Examples
1540
1541Example: C<fork()> a new process and install a child handler to wait for
1542its completion.
1543
1544 ev_child cw;
1334 1545
1335 static void 1546 static void
1336 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents) 1547 child_cb (EV_P_ struct ev_child *w, int revents)
1337 { 1548 {
1338 ev_unloop (loop, EVUNLOOP_ALL); 1549 ev_child_stop (EV_A_ w);
1550 printf ("process %d exited with status %x\n", w->rpid, w->rstatus);
1339 } 1551 }
1340 1552
1341 struct ev_signal signal_watcher; 1553 pid_t pid = fork ();
1342 ev_signal_init (&signal_watcher, sigint_cb, SIGINT); 1554
1343 ev_signal_start (loop, &sigint_cb); 1555 if (pid < 0)
1556 // error
1557 else if (pid == 0)
1558 {
1559 // the forked child executes here
1560 exit (1);
1561 }
1562 else
1563 {
1564 ev_child_init (&cw, child_cb, pid, 0);
1565 ev_child_start (EV_DEFAULT_ &cw);
1566 }
1344 1567
1345 1568
1346=head2 C<ev_stat> - did the file attributes just change? 1569=head2 C<ev_stat> - did the file attributes just change?
1347 1570
1348This watches a filesystem path for attribute changes. That is, it calls 1571This watches a filesystem path for attribute changes. That is, it calls
1377semantics of C<ev_stat> watchers, which means that libev sometimes needs 1600semantics of C<ev_stat> watchers, which means that libev sometimes needs
1378to fall back to regular polling again even with inotify, but changes are 1601to fall back to regular polling again even with inotify, but changes are
1379usually detected immediately, and if the file exists there will be no 1602usually detected immediately, and if the file exists there will be no
1380polling. 1603polling.
1381 1604
1605=head3 Inotify
1606
1607When C<inotify (7)> support has been compiled into libev (generally only
1608available on Linux) and present at runtime, it will be used to speed up
1609change detection where possible. The inotify descriptor will be created lazily
1610when the first C<ev_stat> watcher is being started.
1611
1612Inotify presense does not change the semantics of C<ev_stat> watchers
1613except that changes might be detected earlier, and in some cases, to avoid
1614making regular C<stat> calls. Even in the presense of inotify support
1615there are many cases where libev has to resort to regular C<stat> polling.
1616
1617(There is no support for kqueue, as apparently it cannot be used to
1618implement this functionality, due to the requirement of having a file
1619descriptor open on the object at all times).
1620
1621=head3 The special problem of stat time resolution
1622
1623The C<stat ()> syscall only supports full-second resolution portably, and
1624even on systems where the resolution is higher, many filesystems still
1625only support whole seconds.
1626
1627That means that, if the time is the only thing that changes, you might
1628miss updates: on the first update, C<ev_stat> detects a change and calls
1629your callback, which does something. When there is another update within
1630the same second, C<ev_stat> will be unable to detect it.
1631
1632The solution to this is to delay acting on a change for a second (or till
1633the next second boundary), using a roughly one-second delay C<ev_timer>
1634(C<ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)>). The C<.01>
1635is added to work around small timing inconsistencies of some operating
1636systems.
1637
1382=head3 Watcher-Specific Functions and Data Members 1638=head3 Watcher-Specific Functions and Data Members
1383 1639
1384=over 4 1640=over 4
1385 1641
1386=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval) 1642=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
1395 1651
1396The callback will be receive C<EV_STAT> when a change was detected, 1652The callback will be receive C<EV_STAT> when a change was detected,
1397relative to the attributes at the time the watcher was started (or the 1653relative to the attributes at the time the watcher was started (or the
1398last change was detected). 1654last change was detected).
1399 1655
1400=item ev_stat_stat (ev_stat *) 1656=item ev_stat_stat (loop, ev_stat *)
1401 1657
1402Updates the stat buffer immediately with new values. If you change the 1658Updates the stat buffer immediately with new values. If you change the
1403watched path in your callback, you could call this fucntion to avoid 1659watched path in your callback, you could call this fucntion to avoid
1404detecting this change (while introducing a race condition). Can also be 1660detecting this change (while introducing a race condition). Can also be
1405useful simply to find out the new values. 1661useful simply to find out the new values.
1423=item const char *path [read-only] 1679=item const char *path [read-only]
1424 1680
1425The filesystem path that is being watched. 1681The filesystem path that is being watched.
1426 1682
1427=back 1683=back
1684
1685=head3 Examples
1428 1686
1429Example: Watch C</etc/passwd> for attribute changes. 1687Example: Watch C</etc/passwd> for attribute changes.
1430 1688
1431 static void 1689 static void
1432 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents) 1690 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1445 } 1703 }
1446 1704
1447 ... 1705 ...
1448 ev_stat passwd; 1706 ev_stat passwd;
1449 1707
1450 ev_stat_init (&passwd, passwd_cb, "/etc/passwd"); 1708 ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1451 ev_stat_start (loop, &passwd); 1709 ev_stat_start (loop, &passwd);
1710
1711Example: Like above, but additionally use a one-second delay so we do not
1712miss updates (however, frequent updates will delay processing, too, so
1713one might do the work both on C<ev_stat> callback invocation I<and> on
1714C<ev_timer> callback invocation).
1715
1716 static ev_stat passwd;
1717 static ev_timer timer;
1718
1719 static void
1720 timer_cb (EV_P_ ev_timer *w, int revents)
1721 {
1722 ev_timer_stop (EV_A_ w);
1723
1724 /* now it's one second after the most recent passwd change */
1725 }
1726
1727 static void
1728 stat_cb (EV_P_ ev_stat *w, int revents)
1729 {
1730 /* reset the one-second timer */
1731 ev_timer_again (EV_A_ &timer);
1732 }
1733
1734 ...
1735 ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
1736 ev_stat_start (loop, &passwd);
1737 ev_timer_init (&timer, timer_cb, 0., 1.01);
1452 1738
1453 1739
1454=head2 C<ev_idle> - when you've got nothing better to do... 1740=head2 C<ev_idle> - when you've got nothing better to do...
1455 1741
1456Idle watchers trigger events when no other events of the same or higher 1742Idle watchers trigger events when no other events of the same or higher
1482kind. There is a C<ev_idle_set> macro, but using it is utterly pointless, 1768kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1483believe me. 1769believe me.
1484 1770
1485=back 1771=back
1486 1772
1773=head3 Examples
1774
1487Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the 1775Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1488callback, free it. Also, use no error checking, as usual. 1776callback, free it. Also, use no error checking, as usual.
1489 1777
1490 static void 1778 static void
1491 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents) 1779 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1492 { 1780 {
1493 free (w); 1781 free (w);
1494 // now do something you wanted to do when the program has 1782 // now do something you wanted to do when the program has
1495 // no longer asnything immediate to do. 1783 // no longer anything immediate to do.
1496 } 1784 }
1497 1785
1498 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle)); 1786 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1499 ev_idle_init (idle_watcher, idle_cb); 1787 ev_idle_init (idle_watcher, idle_cb);
1500 ev_idle_start (loop, idle_cb); 1788 ev_idle_start (loop, idle_cb);
1542 1830
1543It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>) 1831It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
1544priority, to ensure that they are being run before any other watchers 1832priority, to ensure that they are being run before any other watchers
1545after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers, 1833after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,
1546too) should not activate ("feed") events into libev. While libev fully 1834too) should not activate ("feed") events into libev. While libev fully
1547supports this, they will be called before other C<ev_check> watchers did 1835supports this, they will be called before other C<ev_check> watchers
1548their job. As C<ev_check> watchers are often used to embed other event 1836did their job. As C<ev_check> watchers are often used to embed other
1549loops those other event loops might be in an unusable state until their 1837(non-libev) event loops those other event loops might be in an unusable
1550C<ev_check> watcher ran (always remind yourself to coexist peacefully with 1838state until their C<ev_check> watcher ran (always remind yourself to
1551others). 1839coexist peacefully with others).
1552 1840
1553=head3 Watcher-Specific Functions and Data Members 1841=head3 Watcher-Specific Functions and Data Members
1554 1842
1555=over 4 1843=over 4
1556 1844
1561Initialises and configures the prepare or check watcher - they have no 1849Initialises and configures the prepare or check watcher - they have no
1562parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set> 1850parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1563macros, but using them is utterly, utterly and completely pointless. 1851macros, but using them is utterly, utterly and completely pointless.
1564 1852
1565=back 1853=back
1854
1855=head3 Examples
1566 1856
1567There are a number of principal ways to embed other event loops or modules 1857There are a number of principal ways to embed other event loops or modules
1568into libev. Here are some ideas on how to include libadns into libev 1858into libev. Here are some ideas on how to include libadns into libev
1569(there is a Perl module named C<EV::ADNS> that does this, which you could 1859(there is a Perl module named C<EV::ADNS> that does this, which you could
1570use for an actually working example. Another Perl module named C<EV::Glib> 1860use for an actually working example. Another Perl module named C<EV::Glib>
1739portable one. 2029portable one.
1740 2030
1741So when you want to use this feature you will always have to be prepared 2031So when you want to use this feature you will always have to be prepared
1742that you cannot get an embeddable loop. The recommended way to get around 2032that you cannot get an embeddable loop. The recommended way to get around
1743this is to have a separate variables for your embeddable loop, try to 2033this is to have a separate variables for your embeddable loop, try to
1744create it, and if that fails, use the normal loop for everything: 2034create it, and if that fails, use the normal loop for everything.
2035
2036=head3 Watcher-Specific Functions and Data Members
2037
2038=over 4
2039
2040=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
2041
2042=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
2043
2044Configures the watcher to embed the given loop, which must be
2045embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
2046invoked automatically, otherwise it is the responsibility of the callback
2047to invoke it (it will continue to be called until the sweep has been done,
2048if you do not want thta, you need to temporarily stop the embed watcher).
2049
2050=item ev_embed_sweep (loop, ev_embed *)
2051
2052Make a single, non-blocking sweep over the embedded loop. This works
2053similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
2054apropriate way for embedded loops.
2055
2056=item struct ev_loop *other [read-only]
2057
2058The embedded event loop.
2059
2060=back
2061
2062=head3 Examples
2063
2064Example: Try to get an embeddable event loop and embed it into the default
2065event loop. If that is not possible, use the default loop. The default
2066loop is stored in C<loop_hi>, while the mebeddable loop is stored in
2067C<loop_lo> (which is C<loop_hi> in the acse no embeddable loop can be
2068used).
1745 2069
1746 struct ev_loop *loop_hi = ev_default_init (0); 2070 struct ev_loop *loop_hi = ev_default_init (0);
1747 struct ev_loop *loop_lo = 0; 2071 struct ev_loop *loop_lo = 0;
1748 struct ev_embed embed; 2072 struct ev_embed embed;
1749 2073
1760 ev_embed_start (loop_hi, &embed); 2084 ev_embed_start (loop_hi, &embed);
1761 } 2085 }
1762 else 2086 else
1763 loop_lo = loop_hi; 2087 loop_lo = loop_hi;
1764 2088
1765=head3 Watcher-Specific Functions and Data Members 2089Example: Check if kqueue is available but not recommended and create
2090a kqueue backend for use with sockets (which usually work with any
2091kqueue implementation). Store the kqueue/socket-only event loop in
2092C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
1766 2093
1767=over 4 2094 struct ev_loop *loop = ev_default_init (0);
2095 struct ev_loop *loop_socket = 0;
2096 struct ev_embed embed;
2097
2098 if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
2099 if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
2100 {
2101 ev_embed_init (&embed, 0, loop_socket);
2102 ev_embed_start (loop, &embed);
2103 }
1768 2104
1769=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop) 2105 if (!loop_socket)
2106 loop_socket = loop;
1770 2107
1771=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop) 2108 // now use loop_socket for all sockets, and loop for everything else
1772
1773Configures the watcher to embed the given loop, which must be
1774embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1775invoked automatically, otherwise it is the responsibility of the callback
1776to invoke it (it will continue to be called until the sweep has been done,
1777if you do not want thta, you need to temporarily stop the embed watcher).
1778
1779=item ev_embed_sweep (loop, ev_embed *)
1780
1781Make a single, non-blocking sweep over the embedded loop. This works
1782similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1783apropriate way for embedded loops.
1784
1785=item struct ev_loop *loop [read-only]
1786
1787The embedded event loop.
1788
1789=back
1790 2109
1791 2110
1792=head2 C<ev_fork> - the audacity to resume the event loop after a fork 2111=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1793 2112
1794Fork watchers are called when a C<fork ()> was detected (usually because 2113Fork watchers are called when a C<fork ()> was detected (usually because
1810believe me. 2129believe me.
1811 2130
1812=back 2131=back
1813 2132
1814 2133
2134=head2 C<ev_async> - how to wake up another event loop
2135
2136In general, you cannot use an C<ev_loop> from multiple threads or other
2137asynchronous sources such as signal handlers (as opposed to multiple event
2138loops - those are of course safe to use in different threads).
2139
2140Sometimes, however, you need to wake up another event loop you do not
2141control, for example because it belongs to another thread. This is what
2142C<ev_async> watchers do: as long as the C<ev_async> watcher is active, you
2143can signal it by calling C<ev_async_send>, which is thread- and signal
2144safe.
2145
2146This functionality is very similar to C<ev_signal> watchers, as signals,
2147too, are asynchronous in nature, and signals, too, will be compressed
2148(i.e. the number of callback invocations may be less than the number of
2149C<ev_async_sent> calls).
2150
2151Unlike C<ev_signal> watchers, C<ev_async> works with any event loop, not
2152just the default loop.
2153
2154=head3 Queueing
2155
2156C<ev_async> does not support queueing of data in any way. The reason
2157is that the author does not know of a simple (or any) algorithm for a
2158multiple-writer-single-reader queue that works in all cases and doesn't
2159need elaborate support such as pthreads.
2160
2161That means that if you want to queue data, you have to provide your own
2162queue. But at least I can tell you would implement locking around your
2163queue:
2164
2165=over 4
2166
2167=item queueing from a signal handler context
2168
2169To implement race-free queueing, you simply add to the queue in the signal
2170handler but you block the signal handler in the watcher callback. Here is an example that does that for
2171some fictitiuous SIGUSR1 handler:
2172
2173 static ev_async mysig;
2174
2175 static void
2176 sigusr1_handler (void)
2177 {
2178 sometype data;
2179
2180 // no locking etc.
2181 queue_put (data);
2182 ev_async_send (EV_DEFAULT_ &mysig);
2183 }
2184
2185 static void
2186 mysig_cb (EV_P_ ev_async *w, int revents)
2187 {
2188 sometype data;
2189 sigset_t block, prev;
2190
2191 sigemptyset (&block);
2192 sigaddset (&block, SIGUSR1);
2193 sigprocmask (SIG_BLOCK, &block, &prev);
2194
2195 while (queue_get (&data))
2196 process (data);
2197
2198 if (sigismember (&prev, SIGUSR1)
2199 sigprocmask (SIG_UNBLOCK, &block, 0);
2200 }
2201
2202(Note: pthreads in theory requires you to use C<pthread_setmask>
2203instead of C<sigprocmask> when you use threads, but libev doesn't do it
2204either...).
2205
2206=item queueing from a thread context
2207
2208The strategy for threads is different, as you cannot (easily) block
2209threads but you can easily preempt them, so to queue safely you need to
2210employ a traditional mutex lock, such as in this pthread example:
2211
2212 static ev_async mysig;
2213 static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
2214
2215 static void
2216 otherthread (void)
2217 {
2218 // only need to lock the actual queueing operation
2219 pthread_mutex_lock (&mymutex);
2220 queue_put (data);
2221 pthread_mutex_unlock (&mymutex);
2222
2223 ev_async_send (EV_DEFAULT_ &mysig);
2224 }
2225
2226 static void
2227 mysig_cb (EV_P_ ev_async *w, int revents)
2228 {
2229 pthread_mutex_lock (&mymutex);
2230
2231 while (queue_get (&data))
2232 process (data);
2233
2234 pthread_mutex_unlock (&mymutex);
2235 }
2236
2237=back
2238
2239
2240=head3 Watcher-Specific Functions and Data Members
2241
2242=over 4
2243
2244=item ev_async_init (ev_async *, callback)
2245
2246Initialises and configures the async watcher - it has no parameters of any
2247kind. There is a C<ev_asynd_set> macro, but using it is utterly pointless,
2248believe me.
2249
2250=item ev_async_send (loop, ev_async *)
2251
2252Sends/signals/activates the given C<ev_async> watcher, that is, feeds
2253an C<EV_ASYNC> event on the watcher into the event loop. Unlike
2254C<ev_feed_event>, this call is safe to do in other threads, signal or
2255similar contexts (see the dicusssion of C<EV_ATOMIC_T> in the embedding
2256section below on what exactly this means).
2257
2258This call incurs the overhead of a syscall only once per loop iteration,
2259so while the overhead might be noticable, it doesn't apply to repeated
2260calls to C<ev_async_send>.
2261
2262=back
2263
2264
1815=head1 OTHER FUNCTIONS 2265=head1 OTHER FUNCTIONS
1816 2266
1817There are some other functions of possible interest. Described. Here. Now. 2267There are some other functions of possible interest. Described. Here. Now.
1818 2268
1819=over 4 2269=over 4
2046Example: Define a class with an IO and idle watcher, start one of them in 2496Example: Define a class with an IO and idle watcher, start one of them in
2047the constructor. 2497the constructor.
2048 2498
2049 class myclass 2499 class myclass
2050 { 2500 {
2051 ev_io io; void io_cb (ev::io &w, int revents); 2501 ev::io io; void io_cb (ev::io &w, int revents);
2052 ev_idle idle void idle_cb (ev::idle &w, int revents); 2502 ev:idle idle void idle_cb (ev::idle &w, int revents);
2053 2503
2054 myclass (); 2504 myclass (int fd)
2055 }
2056
2057 myclass::myclass (int fd)
2058 { 2505 {
2059 io .set <myclass, &myclass::io_cb > (this); 2506 io .set <myclass, &myclass::io_cb > (this);
2060 idle.set <myclass, &myclass::idle_cb> (this); 2507 idle.set <myclass, &myclass::idle_cb> (this);
2061 2508
2062 io.start (fd, ev::READ); 2509 io.start (fd, ev::READ);
2510 }
2063 } 2511 };
2512
2513
2514=head1 OTHER LANGUAGE BINDINGS
2515
2516Libev does not offer other language bindings itself, but bindings for a
2517numbe rof languages exist in the form of third-party packages. If you know
2518any interesting language binding in addition to the ones listed here, drop
2519me a note.
2520
2521=over 4
2522
2523=item Perl
2524
2525The EV module implements the full libev API and is actually used to test
2526libev. EV is developed together with libev. Apart from the EV core module,
2527there are additional modules that implement libev-compatible interfaces
2528to C<libadns> (C<EV::ADNS>), C<Net::SNMP> (C<Net::SNMP::EV>) and the
2529C<libglib> event core (C<Glib::EV> and C<EV::Glib>).
2530
2531It can be found and installed via CPAN, its homepage is found at
2532L<http://software.schmorp.de/pkg/EV>.
2533
2534=item Ruby
2535
2536Tony Arcieri has written a ruby extension that offers access to a subset
2537of the libev API and adds filehandle abstractions, asynchronous DNS and
2538more on top of it. It can be found via gem servers. Its homepage is at
2539L<http://rev.rubyforge.org/>.
2540
2541=item D
2542
2543Leandro Lucarella has written a D language binding (F<ev.d>) for libev, to
2544be found at L<http://git.llucax.com.ar/?p=software/ev.d.git;a=summary>.
2545
2546=back
2064 2547
2065 2548
2066=head1 MACRO MAGIC 2549=head1 MACRO MAGIC
2067 2550
2068Libev can be compiled with a variety of options, the most fundamantal 2551Libev can be compiled with a variety of options, the most fundamantal
2129Libev can (and often is) directly embedded into host 2612Libev can (and often is) directly embedded into host
2130applications. Examples of applications that embed it include the Deliantra 2613applications. Examples of applications that embed it include the Deliantra
2131Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe) 2614Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
2132and rxvt-unicode. 2615and rxvt-unicode.
2133 2616
2134The goal is to enable you to just copy the neecssary files into your 2617The goal is to enable you to just copy the necessary files into your
2135source directory without having to change even a single line in them, so 2618source directory without having to change even a single line in them, so
2136you can easily upgrade by simply copying (or having a checked-out copy of 2619you can easily upgrade by simply copying (or having a checked-out copy of
2137libev somewhere in your source tree). 2620libev somewhere in your source tree).
2138 2621
2139=head2 FILESETS 2622=head2 FILESETS
2229 2712
2230If defined to be C<1>, libev will try to detect the availability of the 2713If defined to be C<1>, libev will try to detect the availability of the
2231monotonic clock option at both compiletime and runtime. Otherwise no use 2714monotonic clock option at both compiletime and runtime. Otherwise no use
2232of the monotonic clock option will be attempted. If you enable this, you 2715of the monotonic clock option will be attempted. If you enable this, you
2233usually have to link against librt or something similar. Enabling it when 2716usually have to link against librt or something similar. Enabling it when
2234the functionality isn't available is safe, though, althoguh you have 2717the functionality isn't available is safe, though, although you have
2235to make sure you link against any libraries where the C<clock_gettime> 2718to make sure you link against any libraries where the C<clock_gettime>
2236function is hiding in (often F<-lrt>). 2719function is hiding in (often F<-lrt>).
2237 2720
2238=item EV_USE_REALTIME 2721=item EV_USE_REALTIME
2239 2722
2240If defined to be C<1>, libev will try to detect the availability of the 2723If defined to be C<1>, libev will try to detect the availability of the
2241realtime clock option at compiletime (and assume its availability at 2724realtime clock option at compiletime (and assume its availability at
2242runtime if successful). Otherwise no use of the realtime clock option will 2725runtime if successful). Otherwise no use of the realtime clock option will
2243be attempted. This effectively replaces C<gettimeofday> by C<clock_get 2726be attempted. This effectively replaces C<gettimeofday> by C<clock_get
2244(CLOCK_REALTIME, ...)> and will not normally affect correctness. See tzhe note about libraries 2727(CLOCK_REALTIME, ...)> and will not normally affect correctness. See the
2245in the description of C<EV_USE_MONOTONIC>, though. 2728note about libraries in the description of C<EV_USE_MONOTONIC>, though.
2729
2730=item EV_USE_NANOSLEEP
2731
2732If defined to be C<1>, libev will assume that C<nanosleep ()> is available
2733and will use it for delays. Otherwise it will use C<select ()>.
2246 2734
2247=item EV_USE_SELECT 2735=item EV_USE_SELECT
2248 2736
2249If undefined or defined to be C<1>, libev will compile in support for the 2737If undefined or defined to be C<1>, libev will compile in support for the
2250C<select>(2) backend. No attempt at autodetection will be done: if no 2738C<select>(2) backend. No attempt at autodetection will be done: if no
2268wants osf handles on win32 (this is the case when the select to 2756wants osf handles on win32 (this is the case when the select to
2269be used is the winsock select). This means that it will call 2757be used is the winsock select). This means that it will call
2270C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise, 2758C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2271it is assumed that all these functions actually work on fds, even 2759it is assumed that all these functions actually work on fds, even
2272on win32. Should not be defined on non-win32 platforms. 2760on win32. Should not be defined on non-win32 platforms.
2761
2762=item EV_FD_TO_WIN32_HANDLE
2763
2764If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
2765file descriptors to socket handles. When not defining this symbol (the
2766default), then libev will call C<_get_osfhandle>, which is usually
2767correct. In some cases, programs use their own file descriptor management,
2768in which case they can provide this function to map fds to socket handles.
2273 2769
2274=item EV_USE_POLL 2770=item EV_USE_POLL
2275 2771
2276If defined to be C<1>, libev will compile in support for the C<poll>(2) 2772If defined to be C<1>, libev will compile in support for the C<poll>(2)
2277backend. Otherwise it will be enabled on non-win32 platforms. It 2773backend. Otherwise it will be enabled on non-win32 platforms. It
2311 2807
2312If defined to be C<1>, libev will compile in support for the Linux inotify 2808If defined to be C<1>, libev will compile in support for the Linux inotify
2313interface to speed up C<ev_stat> watchers. Its actual availability will 2809interface to speed up C<ev_stat> watchers. Its actual availability will
2314be detected at runtime. 2810be detected at runtime.
2315 2811
2812=item EV_ATOMIC_T
2813
2814Libev requires an integer type (suitable for storing C<0> or C<1>) whose
2815access is atomic with respect to other threads or signal contexts. No such
2816type is easily found in the C language, so you can provide your own type
2817that you know is safe for your purposes. It is used both for signal handler "locking"
2818as well as for signal and thread safety in C<ev_async> watchers.
2819
2820In the absense of this define, libev will use C<sig_atomic_t volatile>
2821(from F<signal.h>), which is usually good enough on most platforms.
2822
2316=item EV_H 2823=item EV_H
2317 2824
2318The name of the F<ev.h> header file used to include it. The default if 2825The name of the F<ev.h> header file used to include it. The default if
2319undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This 2826undefined is C<"ev.h"> in F<event.h>, F<ev.c> and F<ev++.h>. This can be
2320can be used to virtually rename the F<ev.h> header file in case of conflicts. 2827used to virtually rename the F<ev.h> header file in case of conflicts.
2321 2828
2322=item EV_CONFIG_H 2829=item EV_CONFIG_H
2323 2830
2324If C<EV_STANDALONE> isn't C<1>, this variable can be used to override 2831If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2325F<ev.c>'s idea of where to find the F<config.h> file, similarly to 2832F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2326C<EV_H>, above. 2833C<EV_H>, above.
2327 2834
2328=item EV_EVENT_H 2835=item EV_EVENT_H
2329 2836
2330Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea 2837Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2331of how the F<event.h> header can be found. 2838of how the F<event.h> header can be found, the default is C<"event.h">.
2332 2839
2333=item EV_PROTOTYPES 2840=item EV_PROTOTYPES
2334 2841
2335If defined to be C<0>, then F<ev.h> will not define any function 2842If defined to be C<0>, then F<ev.h> will not define any function
2336prototypes, but still define all the structs and other symbols. This is 2843prototypes, but still define all the structs and other symbols. This is
2387=item EV_FORK_ENABLE 2894=item EV_FORK_ENABLE
2388 2895
2389If undefined or defined to be C<1>, then fork watchers are supported. If 2896If undefined or defined to be C<1>, then fork watchers are supported. If
2390defined to be C<0>, then they are not. 2897defined to be C<0>, then they are not.
2391 2898
2899=item EV_ASYNC_ENABLE
2900
2901If undefined or defined to be C<1>, then async watchers are supported. If
2902defined to be C<0>, then they are not.
2903
2392=item EV_MINIMAL 2904=item EV_MINIMAL
2393 2905
2394If you need to shave off some kilobytes of code at the expense of some 2906If you need to shave off some kilobytes of code at the expense of some
2395speed, define this symbol to C<1>. Currently only used for gcc to override 2907speed, define this symbol to C<1>. Currently only used for gcc to override
2396some inlining decisions, saves roughly 30% codesize of amd64. 2908some inlining decisions, saves roughly 30% codesize of amd64.
2402than enough. If you need to manage thousands of children you might want to 2914than enough. If you need to manage thousands of children you might want to
2403increase this value (I<must> be a power of two). 2915increase this value (I<must> be a power of two).
2404 2916
2405=item EV_INOTIFY_HASHSIZE 2917=item EV_INOTIFY_HASHSIZE
2406 2918
2407C<ev_staz> watchers use a small hash table to distribute workload by 2919C<ev_stat> watchers use a small hash table to distribute workload by
2408inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>), 2920inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
2409usually more than enough. If you need to manage thousands of C<ev_stat> 2921usually more than enough. If you need to manage thousands of C<ev_stat>
2410watchers you might want to increase this value (I<must> be a power of 2922watchers you might want to increase this value (I<must> be a power of
2411two). 2923two).
2412 2924
2429 2941
2430=item ev_set_cb (ev, cb) 2942=item ev_set_cb (ev, cb)
2431 2943
2432Can be used to change the callback member declaration in each watcher, 2944Can be used to change the callback member declaration in each watcher,
2433and the way callbacks are invoked and set. Must expand to a struct member 2945and the way callbacks are invoked and set. Must expand to a struct member
2434definition and a statement, respectively. See the F<ev.v> header file for 2946definition and a statement, respectively. See the F<ev.h> header file for
2435their default definitions. One possible use for overriding these is to 2947their default definitions. One possible use for overriding these is to
2436avoid the C<struct ev_loop *> as first argument in all cases, or to use 2948avoid the C<struct ev_loop *> as first argument in all cases, or to use
2437method calls instead of plain function calls in C++. 2949method calls instead of plain function calls in C++.
2950
2951=head2 EXPORTED API SYMBOLS
2952
2953If you need to re-export the API (e.g. via a dll) and you need a list of
2954exported symbols, you can use the provided F<Symbol.*> files which list
2955all public symbols, one per line:
2956
2957 Symbols.ev for libev proper
2958 Symbols.event for the libevent emulation
2959
2960This can also be used to rename all public symbols to avoid clashes with
2961multiple versions of libev linked together (which is obviously bad in
2962itself, but sometimes it is inconvinient to avoid this).
2963
2964A sed command like this will create wrapper C<#define>'s that you need to
2965include before including F<ev.h>:
2966
2967 <Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
2968
2969This would create a file F<wrap.h> which essentially looks like this:
2970
2971 #define ev_backend myprefix_ev_backend
2972 #define ev_check_start myprefix_ev_check_start
2973 #define ev_check_stop myprefix_ev_check_stop
2974 ...
2438 2975
2439=head2 EXAMPLES 2976=head2 EXAMPLES
2440 2977
2441For a real-world example of a program the includes libev 2978For a real-world example of a program the includes libev
2442verbatim, you can have a look at the EV perl module 2979verbatim, you can have a look at the EV perl module
2483 3020
2484=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers) 3021=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2485 3022
2486This means that, when you have a watcher that triggers in one hour and 3023This means that, when you have a watcher that triggers in one hour and
2487there are 100 watchers that would trigger before that then inserting will 3024there are 100 watchers that would trigger before that then inserting will
2488have to skip those 100 watchers. 3025have to skip roughly seven (C<ld 100>) of these watchers.
2489 3026
2490=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers) 3027=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2491 3028
2492That means that for changing a timer costs less than removing/adding them 3029That means that changing a timer costs less than removing/adding them
2493as only the relative motion in the event queue has to be paid for. 3030as only the relative motion in the event queue has to be paid for.
2494 3031
2495=item Starting io/check/prepare/idle/signal/child watchers: O(1) 3032=item Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)
2496 3033
2497These just add the watcher into an array or at the head of a list. 3034These just add the watcher into an array or at the head of a list.
3035
2498=item Stopping check/prepare/idle watchers: O(1) 3036=item Stopping check/prepare/idle/fork/async watchers: O(1)
2499 3037
2500=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE)) 3038=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2501 3039
2502These watchers are stored in lists then need to be walked to find the 3040These watchers are stored in lists then need to be walked to find the
2503correct watcher to remove. The lists are usually short (you don't usually 3041correct watcher to remove. The lists are usually short (you don't usually
2504have many watchers waiting for the same fd or signal). 3042have many watchers waiting for the same fd or signal).
2505 3043
2506=item Finding the next timer per loop iteration: O(1) 3044=item Finding the next timer in each loop iteration: O(1)
3045
3046By virtue of using a binary heap, the next timer is always found at the
3047beginning of the storage array.
2507 3048
2508=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd) 3049=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2509 3050
2510A change means an I/O watcher gets started or stopped, which requires 3051A change means an I/O watcher gets started or stopped, which requires
2511libev to recalculate its status (and possibly tell the kernel). 3052libev to recalculate its status (and possibly tell the kernel, depending
3053on backend and wether C<ev_io_set> was used).
2512 3054
2513=item Activating one watcher: O(1) 3055=item Activating one watcher (putting it into the pending state): O(1)
2514 3056
2515=item Priority handling: O(number_of_priorities) 3057=item Priority handling: O(number_of_priorities)
2516 3058
2517Priorities are implemented by allocating some space for each 3059Priorities are implemented by allocating some space for each
2518priority. When doing priority-based operations, libev usually has to 3060priority. When doing priority-based operations, libev usually has to
2519linearly search all the priorities. 3061linearly search all the priorities, but starting/stopping and activating
3062watchers becomes O(1) w.r.t. priority handling.
3063
3064=item Sending an ev_async: O(1)
3065
3066=item Processing ev_async_send: O(number_of_async_watchers)
3067
3068=item Processing signals: O(max_signal_number)
3069
3070Sending involves a syscall I<iff> there were no other C<ev_async_send>
3071calls in the current loop iteration. Checking for async and signal events
3072involves iterating over all running async watchers or all signal numbers.
2520 3073
2521=back 3074=back
2522 3075
2523 3076
3077=head1 Win32 platform limitations and workarounds
3078
3079Win32 doesn't support any of the standards (e.g. POSIX) that libev
3080requires, and its I/O model is fundamentally incompatible with the POSIX
3081model. Libev still offers limited functionality on this platform in
3082the form of the C<EVBACKEND_SELECT> backend, and only supports socket
3083descriptors. This only applies when using Win32 natively, not when using
3084e.g. cygwin.
3085
3086There is no supported compilation method available on windows except
3087embedding it into other applications.
3088
3089Due to the many, low, and arbitrary limits on the win32 platform and the
3090abysmal performance of winsockets, using a large number of sockets is not
3091recommended (and not reasonable). If your program needs to use more than
3092a hundred or so sockets, then likely it needs to use a totally different
3093implementation for windows, as libev offers the POSIX model, which cannot
3094be implemented efficiently on windows (microsoft monopoly games).
3095
3096=over 4
3097
3098=item The winsocket select function
3099
3100The winsocket C<select> function doesn't follow POSIX in that it requires
3101socket I<handles> and not socket I<file descriptors>. This makes select
3102very inefficient, and also requires a mapping from file descriptors
3103to socket handles. See the discussion of the C<EV_SELECT_USE_FD_SET>,
3104C<EV_SELECT_IS_WINSOCKET> and C<EV_FD_TO_WIN32_HANDLE> preprocessor
3105symbols for more info.
3106
3107The configuration for a "naked" win32 using the microsoft runtime
3108libraries and raw winsocket select is:
3109
3110 #define EV_USE_SELECT 1
3111 #define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
3112
3113Note that winsockets handling of fd sets is O(n), so you can easily get a
3114complexity in the O(n²) range when using win32.
3115
3116=item Limited number of file descriptors
3117
3118Windows has numerous arbitrary (and low) limits on things. Early versions
3119of winsocket's select only supported waiting for a max. of C<64> handles
3120(probably owning to the fact that all windows kernels can only wait for
3121C<64> things at the same time internally; microsoft recommends spawning a
3122chain of threads and wait for 63 handles and the previous thread in each).
3123
3124Newer versions support more handles, but you need to define C<FD_SETSIZE>
3125to some high number (e.g. C<2048>) before compiling the winsocket select
3126call (which might be in libev or elsewhere, for example, perl does its own
3127select emulation on windows).
3128
3129Another limit is the number of file descriptors in the microsoft runtime
3130libraries, which by default is C<64> (there must be a hidden I<64> fetish
3131or something like this inside microsoft). You can increase this by calling
3132C<_setmaxstdio>, which can increase this limit to C<2048> (another
3133arbitrary limit), but is broken in many versions of the microsoft runtime
3134libraries.
3135
3136This might get you to about C<512> or C<2048> sockets (depending on
3137windows version and/or the phase of the moon). To get more, you need to
3138wrap all I/O functions and provide your own fd management, but the cost of
3139calling select (O(n²)) will likely make this unworkable.
3140
3141=back
3142
3143
2524=head1 AUTHOR 3144=head1 AUTHOR
2525 3145
2526Marc Lehmann <libev@schmorp.de>. 3146Marc Lehmann <libev@schmorp.de>.
2527 3147

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines