ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libev/ev.3
(Generate patch)

Comparing libev/ev.3 (file contents):
Revision 1.2 by root, Sun Nov 18 03:43:24 2007 UTC vs.
Revision 1.46 by root, Sun Dec 9 19:42:57 2007 UTC

… …
127.\} 127.\}
128.rm #[ #] #H #V #F C 128.rm #[ #] #H #V #F C
129.\" ======================================================================== 129.\" ========================================================================
130.\" 130.\"
131.IX Title ""<STANDARD INPUT>" 1" 131.IX Title ""<STANDARD INPUT>" 1"
132.TH "<STANDARD INPUT>" 1 "2007-11-18" "perl v5.8.8" "User Contributed Perl Documentation" 132.TH "<STANDARD INPUT>" 1 "2007-12-09" "perl v5.8.8" "User Contributed Perl Documentation"
133.SH "NAME" 133.SH "NAME"
134libev \- a high performance full\-featured event loop written in C 134libev \- a high performance full\-featured event loop written in C
135.SH "SYNOPSIS" 135.SH "SYNOPSIS"
136.IX Header "SYNOPSIS" 136.IX Header "SYNOPSIS"
137.Vb 1 137.Vb 1
138\& #include <ev.h> 138\& #include <ev.h>
139.Ve 139.Ve
140.SH "EXAMPLE PROGRAM"
141.IX Header "EXAMPLE PROGRAM"
142.Vb 1
143\& #include <ev.h>
144.Ve
145.PP
146.Vb 2
147\& ev_io stdin_watcher;
148\& ev_timer timeout_watcher;
149.Ve
150.PP
151.Vb 8
152\& /* called when data readable on stdin */
153\& static void
154\& stdin_cb (EV_P_ struct ev_io *w, int revents)
155\& {
156\& /* puts ("stdin ready"); */
157\& ev_io_stop (EV_A_ w); /* just a syntax example */
158\& ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
159\& }
160.Ve
161.PP
162.Vb 6
163\& static void
164\& timeout_cb (EV_P_ struct ev_timer *w, int revents)
165\& {
166\& /* puts ("timeout"); */
167\& ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
168\& }
169.Ve
170.PP
171.Vb 4
172\& int
173\& main (void)
174\& {
175\& struct ev_loop *loop = ev_default_loop (0);
176.Ve
177.PP
178.Vb 3
179\& /* initialise an io watcher, then start it */
180\& ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
181\& ev_io_start (loop, &stdin_watcher);
182.Ve
183.PP
184.Vb 3
185\& /* simple non-repeating 5.5 second timeout */
186\& ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
187\& ev_timer_start (loop, &timeout_watcher);
188.Ve
189.PP
190.Vb 2
191\& /* loop till timeout or data ready */
192\& ev_loop (loop, 0);
193.Ve
194.PP
195.Vb 2
196\& return 0;
197\& }
198.Ve
140.SH "DESCRIPTION" 199.SH "DESCRIPTION"
141.IX Header "DESCRIPTION" 200.IX Header "DESCRIPTION"
201The newest version of this document is also available as a html-formatted
202web page you might find easier to navigate when reading it for the first
203time: <http://cvs.schmorp.de/libev/ev.html>.
204.PP
142Libev is an event loop: you register interest in certain events (such as a 205Libev is an event loop: you register interest in certain events (such as a
143file descriptor being readable or a timeout occuring), and it will manage 206file descriptor being readable or a timeout occuring), and it will manage
144these event sources and provide your program with events. 207these event sources and provide your program with events.
145.PP 208.PP
146To do this, it must take more or less complete control over your process 209To do this, it must take more or less complete control over your process
… …
151watchers\fR, which are relatively small C structures you initialise with the 214watchers\fR, which are relatively small C structures you initialise with the
152details of the event, and then hand it over to libev by \fIstarting\fR the 215details of the event, and then hand it over to libev by \fIstarting\fR the
153watcher. 216watcher.
154.SH "FEATURES" 217.SH "FEATURES"
155.IX Header "FEATURES" 218.IX Header "FEATURES"
156Libev supports select, poll, the linux-specific epoll and the bsd-specific 219Libev supports \f(CW\*(C`select\*(C'\fR, \f(CW\*(C`poll\*(C'\fR, the Linux-specific \f(CW\*(C`epoll\*(C'\fR, the
157kqueue mechanisms for file descriptor events, relative timers, absolute 220BSD-specific \f(CW\*(C`kqueue\*(C'\fR and the Solaris-specific event port mechanisms
158timers with customised rescheduling, signal events, process status change 221for file descriptor events (\f(CW\*(C`ev_io\*(C'\fR), the Linux \f(CW\*(C`inotify\*(C'\fR interface
159events (related to \s-1SIGCHLD\s0), and event watchers dealing with the event 222(for \f(CW\*(C`ev_stat\*(C'\fR), relative timers (\f(CW\*(C`ev_timer\*(C'\fR), absolute timers
160loop mechanism itself (idle, prepare and check watchers). It also is quite 223with customised rescheduling (\f(CW\*(C`ev_periodic\*(C'\fR), synchronous signals
161fast (see this benchmark comparing 224(\f(CW\*(C`ev_signal\*(C'\fR), process status change events (\f(CW\*(C`ev_child\*(C'\fR), and event
162it to libevent for example). 225watchers dealing with the event loop mechanism itself (\f(CW\*(C`ev_idle\*(C'\fR,
226\&\f(CW\*(C`ev_embed\*(C'\fR, \f(CW\*(C`ev_prepare\*(C'\fR and \f(CW\*(C`ev_check\*(C'\fR watchers) as well as
227file watchers (\f(CW\*(C`ev_stat\*(C'\fR) and even limited support for fork events
228(\f(CW\*(C`ev_fork\*(C'\fR).
229.PP
230It also is quite fast (see this
231benchmark comparing it to libevent
232for example).
163.SH "CONVENTIONS" 233.SH "CONVENTIONS"
164.IX Header "CONVENTIONS" 234.IX Header "CONVENTIONS"
165Libev is very configurable. In this manual the default configuration 235Libev is very configurable. In this manual the default configuration will
166will be described, which supports multiple event loops. For more info 236be described, which supports multiple event loops. For more info about
167about various configuration options please have a look at the file 237various configuration options please have a look at \fB\s-1EMBED\s0\fR section in
168\&\fI\s-1README\s0.embed\fR in the libev distribution. If libev was configured without 238this manual. If libev was configured without support for multiple event
169support for multiple event loops, then all functions taking an initial 239loops, then all functions taking an initial argument of name \f(CW\*(C`loop\*(C'\fR
170argument of name \f(CW\*(C`loop\*(C'\fR (which is always of type \f(CW\*(C`struct ev_loop *\*(C'\fR) 240(which is always of type \f(CW\*(C`struct ev_loop *\*(C'\fR) will not have this argument.
171will not have this argument.
172.SH "TIME REPRESENTATION" 241.SH "TIME REPRESENTATION"
173.IX Header "TIME REPRESENTATION" 242.IX Header "TIME REPRESENTATION"
174Libev represents time as a single floating point number, representing the 243Libev represents time as a single floating point number, representing the
175(fractional) number of seconds since the (\s-1POSIX\s0) epoch (somewhere near 244(fractional) number of seconds since the (\s-1POSIX\s0) epoch (somewhere near
176the beginning of 1970, details are complicated, don't ask). This type is 245the beginning of 1970, details are complicated, don't ask). This type is
177called \f(CW\*(C`ev_tstamp\*(C'\fR, which is what you should use too. It usually aliases 246called \f(CW\*(C`ev_tstamp\*(C'\fR, which is what you should use too. It usually aliases
178to the double type in C. 247to the \f(CW\*(C`double\*(C'\fR type in C, and when you need to do any calculations on
248it, you should treat it as such.
179.SH "GLOBAL FUNCTIONS" 249.SH "GLOBAL FUNCTIONS"
180.IX Header "GLOBAL FUNCTIONS" 250.IX Header "GLOBAL FUNCTIONS"
181These functions can be called anytime, even before initialising the 251These functions can be called anytime, even before initialising the
182library in any way. 252library in any way.
183.IP "ev_tstamp ev_time ()" 4 253.IP "ev_tstamp ev_time ()" 4
… …
199.Sp 269.Sp
200Usually, it's a good idea to terminate if the major versions mismatch, 270Usually, it's a good idea to terminate if the major versions mismatch,
201as this indicates an incompatible change. Minor versions are usually 271as this indicates an incompatible change. Minor versions are usually
202compatible to older versions, so a larger minor version alone is usually 272compatible to older versions, so a larger minor version alone is usually
203not a problem. 273not a problem.
274.Sp
275Example: Make sure we haven't accidentally been linked against the wrong
276version.
277.Sp
278.Vb 3
279\& assert (("libev version mismatch",
280\& ev_version_major () == EV_VERSION_MAJOR
281\& && ev_version_minor () >= EV_VERSION_MINOR));
282.Ve
283.IP "unsigned int ev_supported_backends ()" 4
284.IX Item "unsigned int ev_supported_backends ()"
285Return the set of all backends (i.e. their corresponding \f(CW\*(C`EV_BACKEND_*\*(C'\fR
286value) compiled into this binary of libev (independent of their
287availability on the system you are running on). See \f(CW\*(C`ev_default_loop\*(C'\fR for
288a description of the set values.
289.Sp
290Example: make sure we have the epoll method, because yeah this is cool and
291a must have and can we have a torrent of it please!!!11
292.Sp
293.Vb 2
294\& assert (("sorry, no epoll, no sex",
295\& ev_supported_backends () & EVBACKEND_EPOLL));
296.Ve
297.IP "unsigned int ev_recommended_backends ()" 4
298.IX Item "unsigned int ev_recommended_backends ()"
299Return the set of all backends compiled into this binary of libev and also
300recommended for this platform. This set is often smaller than the one
301returned by \f(CW\*(C`ev_supported_backends\*(C'\fR, as for example kqueue is broken on
302most BSDs and will not be autodetected unless you explicitly request it
303(assuming you know what you are doing). This is the set of backends that
304libev will probe for if you specify no backends explicitly.
305.IP "unsigned int ev_embeddable_backends ()" 4
306.IX Item "unsigned int ev_embeddable_backends ()"
307Returns the set of backends that are embeddable in other event loops. This
308is the theoretical, all\-platform, value. To find which backends
309might be supported on the current system, you would need to look at
310\&\f(CW\*(C`ev_embeddable_backends () & ev_supported_backends ()\*(C'\fR, likewise for
311recommended ones.
312.Sp
313See the description of \f(CW\*(C`ev_embed\*(C'\fR watchers for more info.
204.IP "ev_set_allocator (void *(*cb)(void *ptr, long size))" 4 314.IP "ev_set_allocator (void *(*cb)(void *ptr, long size))" 4
205.IX Item "ev_set_allocator (void *(*cb)(void *ptr, long size))" 315.IX Item "ev_set_allocator (void *(*cb)(void *ptr, long size))"
206Sets the allocation function to use (the prototype is similar to the 316Sets the allocation function to use (the prototype is similar \- the
207realloc C function, the semantics are identical). It is used to allocate 317semantics is identical \- to the realloc C function). It is used to
208and free memory (no surprises here). If it returns zero when memory 318allocate and free memory (no surprises here). If it returns zero when
209needs to be allocated, the library might abort or take some potentially 319memory needs to be allocated, the library might abort or take some
210destructive action. The default is your system realloc function. 320potentially destructive action. The default is your system realloc
321function.
211.Sp 322.Sp
212You could override this function in high-availability programs to, say, 323You could override this function in high-availability programs to, say,
213free some memory if it cannot allocate memory, to use a special allocator, 324free some memory if it cannot allocate memory, to use a special allocator,
214or even to sleep a while and retry until some memory is available. 325or even to sleep a while and retry until some memory is available.
326.Sp
327Example: Replace the libev allocator with one that waits a bit and then
328retries).
329.Sp
330.Vb 6
331\& static void *
332\& persistent_realloc (void *ptr, size_t size)
333\& {
334\& for (;;)
335\& {
336\& void *newptr = realloc (ptr, size);
337.Ve
338.Sp
339.Vb 2
340\& if (newptr)
341\& return newptr;
342.Ve
343.Sp
344.Vb 3
345\& sleep (60);
346\& }
347\& }
348.Ve
349.Sp
350.Vb 2
351\& ...
352\& ev_set_allocator (persistent_realloc);
353.Ve
215.IP "ev_set_syserr_cb (void (*cb)(const char *msg));" 4 354.IP "ev_set_syserr_cb (void (*cb)(const char *msg));" 4
216.IX Item "ev_set_syserr_cb (void (*cb)(const char *msg));" 355.IX Item "ev_set_syserr_cb (void (*cb)(const char *msg));"
217Set the callback function to call on a retryable syscall error (such 356Set the callback function to call on a retryable syscall error (such
218as failed select, poll, epoll_wait). The message is a printable string 357as failed select, poll, epoll_wait). The message is a printable string
219indicating the system call or subsystem causing the problem. If this 358indicating the system call or subsystem causing the problem. If this
220callback is set, then libev will expect it to remedy the sitution, no 359callback is set, then libev will expect it to remedy the sitution, no
221matter what, when it returns. That is, libev will generally retry the 360matter what, when it returns. That is, libev will generally retry the
222requested operation, or, if the condition doesn't go away, do bad stuff 361requested operation, or, if the condition doesn't go away, do bad stuff
223(such as abort). 362(such as abort).
363.Sp
364Example: This is basically the same thing that libev does internally, too.
365.Sp
366.Vb 6
367\& static void
368\& fatal_error (const char *msg)
369\& {
370\& perror (msg);
371\& abort ();
372\& }
373.Ve
374.Sp
375.Vb 2
376\& ...
377\& ev_set_syserr_cb (fatal_error);
378.Ve
224.SH "FUNCTIONS CONTROLLING THE EVENT LOOP" 379.SH "FUNCTIONS CONTROLLING THE EVENT LOOP"
225.IX Header "FUNCTIONS CONTROLLING THE EVENT LOOP" 380.IX Header "FUNCTIONS CONTROLLING THE EVENT LOOP"
226An event loop is described by a \f(CW\*(C`struct ev_loop *\*(C'\fR. The library knows two 381An event loop is described by a \f(CW\*(C`struct ev_loop *\*(C'\fR. The library knows two
227types of such loops, the \fIdefault\fR loop, which supports signals and child 382types of such loops, the \fIdefault\fR loop, which supports signals and child
228events, and dynamically created loops which do not. 383events, and dynamically created loops which do not.
… …
236.IP "struct ev_loop *ev_default_loop (unsigned int flags)" 4 391.IP "struct ev_loop *ev_default_loop (unsigned int flags)" 4
237.IX Item "struct ev_loop *ev_default_loop (unsigned int flags)" 392.IX Item "struct ev_loop *ev_default_loop (unsigned int flags)"
238This will initialise the default event loop if it hasn't been initialised 393This will initialise the default event loop if it hasn't been initialised
239yet and return it. If the default loop could not be initialised, returns 394yet and return it. If the default loop could not be initialised, returns
240false. If it already was initialised it simply returns it (and ignores the 395false. If it already was initialised it simply returns it (and ignores the
241flags). 396flags. If that is troubling you, check \f(CW\*(C`ev_backend ()\*(C'\fR afterwards).
242.Sp 397.Sp
243If you don't know what event loop to use, use the one returned from this 398If you don't know what event loop to use, use the one returned from this
244function. 399function.
245.Sp 400.Sp
246The flags argument can be used to specify special behaviour or specific 401The flags argument can be used to specify special behaviour or specific
247backends to use, and is usually specified as 0 (or \s-1EVFLAG_AUTO\s0). 402backends to use, and is usually specified as \f(CW0\fR (or \f(CW\*(C`EVFLAG_AUTO\*(C'\fR).
248.Sp 403.Sp
249It supports the following flags: 404The following flags are supported:
250.RS 4 405.RS 4
251.ie n .IP """EVFLAG_AUTO""" 4 406.ie n .IP """EVFLAG_AUTO""" 4
252.el .IP "\f(CWEVFLAG_AUTO\fR" 4 407.el .IP "\f(CWEVFLAG_AUTO\fR" 4
253.IX Item "EVFLAG_AUTO" 408.IX Item "EVFLAG_AUTO"
254The default flags value. Use this if you have no clue (it's the right 409The default flags value. Use this if you have no clue (it's the right
… …
260or setgid) then libev will \fInot\fR look at the environment variable 415or setgid) then libev will \fInot\fR look at the environment variable
261\&\f(CW\*(C`LIBEV_FLAGS\*(C'\fR. Otherwise (the default), this environment variable will 416\&\f(CW\*(C`LIBEV_FLAGS\*(C'\fR. Otherwise (the default), this environment variable will
262override the flags completely if it is found in the environment. This is 417override the flags completely if it is found in the environment. This is
263useful to try out specific backends to test their performance, or to work 418useful to try out specific backends to test their performance, or to work
264around bugs. 419around bugs.
420.ie n .IP """EVFLAG_FORKCHECK""" 4
421.el .IP "\f(CWEVFLAG_FORKCHECK\fR" 4
422.IX Item "EVFLAG_FORKCHECK"
423Instead of calling \f(CW\*(C`ev_default_fork\*(C'\fR or \f(CW\*(C`ev_loop_fork\*(C'\fR manually after
424a fork, you can also make libev check for a fork in each iteration by
425enabling this flag.
426.Sp
427This works by calling \f(CW\*(C`getpid ()\*(C'\fR on every iteration of the loop,
428and thus this might slow down your event loop if you do a lot of loop
429iterations and little real work, but is usually not noticeable (on my
430Linux system for example, \f(CW\*(C`getpid\*(C'\fR is actually a simple 5\-insn sequence
431without a syscall and thus \fIvery\fR fast, but my Linux system also has
432\&\f(CW\*(C`pthread_atfork\*(C'\fR which is even faster).
433.Sp
434The big advantage of this flag is that you can forget about fork (and
435forget about forgetting to tell libev about forking) when you use this
436flag.
437.Sp
438This flag setting cannot be overriden or specified in the \f(CW\*(C`LIBEV_FLAGS\*(C'\fR
439environment variable.
265.ie n .IP """EVMETHOD_SELECT"" (portable select backend)" 4 440.ie n .IP """EVBACKEND_SELECT"" (value 1, portable select backend)" 4
266.el .IP "\f(CWEVMETHOD_SELECT\fR (portable select backend)" 4 441.el .IP "\f(CWEVBACKEND_SELECT\fR (value 1, portable select backend)" 4
267.IX Item "EVMETHOD_SELECT (portable select backend)" 442.IX Item "EVBACKEND_SELECT (value 1, portable select backend)"
268.PD 0 443This is your standard \fIselect\fR\|(2) backend. Not \fIcompletely\fR standard, as
444libev tries to roll its own fd_set with no limits on the number of fds,
445but if that fails, expect a fairly low limit on the number of fds when
446using this backend. It doesn't scale too well (O(highest_fd)), but its usually
447the fastest backend for a low number of fds.
269.ie n .IP """EVMETHOD_POLL"" (poll backend, available everywhere except on windows)" 4 448.ie n .IP """EVBACKEND_POLL"" (value 2, poll backend, available everywhere except on windows)" 4
270.el .IP "\f(CWEVMETHOD_POLL\fR (poll backend, available everywhere except on windows)" 4 449.el .IP "\f(CWEVBACKEND_POLL\fR (value 2, poll backend, available everywhere except on windows)" 4
271.IX Item "EVMETHOD_POLL (poll backend, available everywhere except on windows)" 450.IX Item "EVBACKEND_POLL (value 2, poll backend, available everywhere except on windows)"
451And this is your standard \fIpoll\fR\|(2) backend. It's more complicated than
452select, but handles sparse fds better and has no artificial limit on the
453number of fds you can use (except it will slow down considerably with a
454lot of inactive fds). It scales similarly to select, i.e. O(total_fds).
272.ie n .IP """EVMETHOD_EPOLL"" (linux only)" 4 455.ie n .IP """EVBACKEND_EPOLL"" (value 4, Linux)" 4
273.el .IP "\f(CWEVMETHOD_EPOLL\fR (linux only)" 4 456.el .IP "\f(CWEVBACKEND_EPOLL\fR (value 4, Linux)" 4
274.IX Item "EVMETHOD_EPOLL (linux only)" 457.IX Item "EVBACKEND_EPOLL (value 4, Linux)"
275.ie n .IP """EVMETHOD_KQUEUE"" (some bsds only)" 4 458For few fds, this backend is a bit little slower than poll and select,
276.el .IP "\f(CWEVMETHOD_KQUEUE\fR (some bsds only)" 4 459but it scales phenomenally better. While poll and select usually scale like
277.IX Item "EVMETHOD_KQUEUE (some bsds only)" 460O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
461either O(1) or O(active_fds).
462.Sp
463While stopping and starting an I/O watcher in the same iteration will
464result in some caching, there is still a syscall per such incident
465(because the fd could point to a different file description now), so its
466best to avoid that. Also, \fIdup()\fRed file descriptors might not work very
467well if you register events for both fds.
468.Sp
469Please note that epoll sometimes generates spurious notifications, so you
470need to use non-blocking I/O or other means to avoid blocking when no data
471(or space) is available.
472.ie n .IP """EVBACKEND_KQUEUE"" (value 8, most \s-1BSD\s0 clones)" 4
473.el .IP "\f(CWEVBACKEND_KQUEUE\fR (value 8, most \s-1BSD\s0 clones)" 4
474.IX Item "EVBACKEND_KQUEUE (value 8, most BSD clones)"
475Kqueue deserves special mention, as at the time of this writing, it
476was broken on all BSDs except NetBSD (usually it doesn't work with
477anything but sockets and pipes, except on Darwin, where of course its
478completely useless). For this reason its not being \*(L"autodetected\*(R"
479unless you explicitly specify it explicitly in the flags (i.e. using
480\&\f(CW\*(C`EVBACKEND_KQUEUE\*(C'\fR).
481.Sp
482It scales in the same way as the epoll backend, but the interface to the
483kernel is more efficient (which says nothing about its actual speed, of
484course). While starting and stopping an I/O watcher does not cause an
485extra syscall as with epoll, it still adds up to four event changes per
486incident, so its best to avoid that.
278.ie n .IP """EVMETHOD_DEVPOLL"" (solaris 8 only)" 4 487.ie n .IP """EVBACKEND_DEVPOLL"" (value 16, Solaris 8)" 4
279.el .IP "\f(CWEVMETHOD_DEVPOLL\fR (solaris 8 only)" 4 488.el .IP "\f(CWEVBACKEND_DEVPOLL\fR (value 16, Solaris 8)" 4
280.IX Item "EVMETHOD_DEVPOLL (solaris 8 only)" 489.IX Item "EVBACKEND_DEVPOLL (value 16, Solaris 8)"
490This is not implemented yet (and might never be).
281.ie n .IP """EVMETHOD_PORT"" (solaris 10 only)" 4 491.ie n .IP """EVBACKEND_PORT"" (value 32, Solaris 10)" 4
282.el .IP "\f(CWEVMETHOD_PORT\fR (solaris 10 only)" 4 492.el .IP "\f(CWEVBACKEND_PORT\fR (value 32, Solaris 10)" 4
283.IX Item "EVMETHOD_PORT (solaris 10 only)" 493.IX Item "EVBACKEND_PORT (value 32, Solaris 10)"
284.PD 494This uses the Solaris 10 port mechanism. As with everything on Solaris,
285If one or more of these are ored into the flags value, then only these 495it's really slow, but it still scales very well (O(active_fds)).
286backends will be tried (in the reverse order as given here). If one are 496.Sp
287specified, any backend will do. 497Please note that solaris ports can result in a lot of spurious
498notifications, so you need to use non-blocking I/O or other means to avoid
499blocking when no data (or space) is available.
500.ie n .IP """EVBACKEND_ALL""" 4
501.el .IP "\f(CWEVBACKEND_ALL\fR" 4
502.IX Item "EVBACKEND_ALL"
503Try all backends (even potentially broken ones that wouldn't be tried
504with \f(CW\*(C`EVFLAG_AUTO\*(C'\fR). Since this is a mask, you can do stuff such as
505\&\f(CW\*(C`EVBACKEND_ALL & ~EVBACKEND_KQUEUE\*(C'\fR.
288.RE 506.RE
289.RS 4 507.RS 4
508.Sp
509If one or more of these are ored into the flags value, then only these
510backends will be tried (in the reverse order as given here). If none are
511specified, most compiled-in backend will be tried, usually in reverse
512order of their flag values :)
513.Sp
514The most typical usage is like this:
515.Sp
516.Vb 2
517\& if (!ev_default_loop (0))
518\& fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
519.Ve
520.Sp
521Restrict libev to the select and poll backends, and do not allow
522environment settings to be taken into account:
523.Sp
524.Vb 1
525\& ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
526.Ve
527.Sp
528Use whatever libev has to offer, but make sure that kqueue is used if
529available (warning, breaks stuff, best use only with your own private
530event loop and only if you know the \s-1OS\s0 supports your types of fds):
531.Sp
532.Vb 1
533\& ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
534.Ve
290.RE 535.RE
291.IP "struct ev_loop *ev_loop_new (unsigned int flags)" 4 536.IP "struct ev_loop *ev_loop_new (unsigned int flags)" 4
292.IX Item "struct ev_loop *ev_loop_new (unsigned int flags)" 537.IX Item "struct ev_loop *ev_loop_new (unsigned int flags)"
293Similar to \f(CW\*(C`ev_default_loop\*(C'\fR, but always creates a new event loop that is 538Similar to \f(CW\*(C`ev_default_loop\*(C'\fR, but always creates a new event loop that is
294always distinct from the default loop. Unlike the default loop, it cannot 539always distinct from the default loop. Unlike the default loop, it cannot
295handle signal and child watchers, and attempts to do so will be greeted by 540handle signal and child watchers, and attempts to do so will be greeted by
296undefined behaviour (or a failed assertion if assertions are enabled). 541undefined behaviour (or a failed assertion if assertions are enabled).
542.Sp
543Example: Try to create a event loop that uses epoll and nothing else.
544.Sp
545.Vb 3
546\& struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
547\& if (!epoller)
548\& fatal ("no epoll found here, maybe it hides under your chair");
549.Ve
297.IP "ev_default_destroy ()" 4 550.IP "ev_default_destroy ()" 4
298.IX Item "ev_default_destroy ()" 551.IX Item "ev_default_destroy ()"
299Destroys the default loop again (frees all memory and kernel state 552Destroys the default loop again (frees all memory and kernel state
300etc.). This stops all registered event watchers (by not touching them in 553etc.). None of the active event watchers will be stopped in the normal
301any way whatsoever, although you cannot rely on this :). 554sense, so e.g. \f(CW\*(C`ev_is_active\*(C'\fR might still return true. It is your
555responsibility to either stop all watchers cleanly yoursef \fIbefore\fR
556calling this function, or cope with the fact afterwards (which is usually
557the easiest thing, youc na just ignore the watchers and/or \f(CW\*(C`free ()\*(C'\fR them
558for example).
302.IP "ev_loop_destroy (loop)" 4 559.IP "ev_loop_destroy (loop)" 4
303.IX Item "ev_loop_destroy (loop)" 560.IX Item "ev_loop_destroy (loop)"
304Like \f(CW\*(C`ev_default_destroy\*(C'\fR, but destroys an event loop created by an 561Like \f(CW\*(C`ev_default_destroy\*(C'\fR, but destroys an event loop created by an
305earlier call to \f(CW\*(C`ev_loop_new\*(C'\fR. 562earlier call to \f(CW\*(C`ev_loop_new\*(C'\fR.
306.IP "ev_default_fork ()" 4 563.IP "ev_default_fork ()" 4
… …
308This function reinitialises the kernel state for backends that have 565This function reinitialises the kernel state for backends that have
309one. Despite the name, you can call it anytime, but it makes most sense 566one. Despite the name, you can call it anytime, but it makes most sense
310after forking, in either the parent or child process (or both, but that 567after forking, in either the parent or child process (or both, but that
311again makes little sense). 568again makes little sense).
312.Sp 569.Sp
313You \fImust\fR call this function after forking if and only if you want to 570You \fImust\fR call this function in the child process after forking if and
314use the event library in both processes. If you just fork+exec, you don't 571only if you want to use the event library in both processes. If you just
315have to call it. 572fork+exec, you don't have to call it.
316.Sp 573.Sp
317The function itself is quite fast and it's usually not a problem to call 574The function itself is quite fast and it's usually not a problem to call
318it just in case after a fork. To make this easy, the function will fit in 575it just in case after a fork. To make this easy, the function will fit in
319quite nicely into a call to \f(CW\*(C`pthread_atfork\*(C'\fR: 576quite nicely into a call to \f(CW\*(C`pthread_atfork\*(C'\fR:
320.Sp 577.Sp
321.Vb 1 578.Vb 1
322\& pthread_atfork (0, 0, ev_default_fork); 579\& pthread_atfork (0, 0, ev_default_fork);
323.Ve 580.Ve
581.Sp
582At the moment, \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR and \f(CW\*(C`EVBACKEND_POLL\*(C'\fR are safe to use
583without calling this function, so if you force one of those backends you
584do not need to care.
324.IP "ev_loop_fork (loop)" 4 585.IP "ev_loop_fork (loop)" 4
325.IX Item "ev_loop_fork (loop)" 586.IX Item "ev_loop_fork (loop)"
326Like \f(CW\*(C`ev_default_fork\*(C'\fR, but acts on an event loop created by 587Like \f(CW\*(C`ev_default_fork\*(C'\fR, but acts on an event loop created by
327\&\f(CW\*(C`ev_loop_new\*(C'\fR. Yes, you have to call this on every allocated event loop 588\&\f(CW\*(C`ev_loop_new\*(C'\fR. Yes, you have to call this on every allocated event loop
328after fork, and how you do this is entirely your own problem. 589after fork, and how you do this is entirely your own problem.
590.IP "unsigned int ev_loop_count (loop)" 4
591.IX Item "unsigned int ev_loop_count (loop)"
592Returns the count of loop iterations for the loop, which is identical to
593the number of times libev did poll for new events. It starts at \f(CW0\fR and
594happily wraps around with enough iterations.
595.Sp
596This value can sometimes be useful as a generation counter of sorts (it
597\&\*(L"ticks\*(R" the number of loop iterations), as it roughly corresponds with
598\&\f(CW\*(C`ev_prepare\*(C'\fR and \f(CW\*(C`ev_check\*(C'\fR calls.
329.IP "unsigned int ev_method (loop)" 4 599.IP "unsigned int ev_backend (loop)" 4
330.IX Item "unsigned int ev_method (loop)" 600.IX Item "unsigned int ev_backend (loop)"
331Returns one of the \f(CW\*(C`EVMETHOD_*\*(C'\fR flags indicating the event backend in 601Returns one of the \f(CW\*(C`EVBACKEND_*\*(C'\fR flags indicating the event backend in
332use. 602use.
333.IP "ev_tstamp ev_now (loop)" 4 603.IP "ev_tstamp ev_now (loop)" 4
334.IX Item "ev_tstamp ev_now (loop)" 604.IX Item "ev_tstamp ev_now (loop)"
335Returns the current \*(L"event loop time\*(R", which is the time the event loop 605Returns the current \*(L"event loop time\*(R", which is the time the event loop
336got events and started processing them. This timestamp does not change 606received events and started processing them. This timestamp does not
337as long as callbacks are being processed, and this is also the base time 607change as long as callbacks are being processed, and this is also the base
338used for relative timers. You can treat it as the timestamp of the event 608time used for relative timers. You can treat it as the timestamp of the
339occuring (or more correctly, the mainloop finding out about it). 609event occuring (or more correctly, libev finding out about it).
340.IP "ev_loop (loop, int flags)" 4 610.IP "ev_loop (loop, int flags)" 4
341.IX Item "ev_loop (loop, int flags)" 611.IX Item "ev_loop (loop, int flags)"
342Finally, this is it, the event handler. This function usually is called 612Finally, this is it, the event handler. This function usually is called
343after you initialised all your watchers and you want to start handling 613after you initialised all your watchers and you want to start handling
344events. 614events.
345.Sp 615.Sp
346If the flags argument is specified as 0, it will not return until either 616If the flags argument is specified as \f(CW0\fR, it will not return until
347no event watchers are active anymore or \f(CW\*(C`ev_unloop\*(C'\fR was called. 617either no event watchers are active anymore or \f(CW\*(C`ev_unloop\*(C'\fR was called.
618.Sp
619Please note that an explicit \f(CW\*(C`ev_unloop\*(C'\fR is usually better than
620relying on all watchers to be stopped when deciding when a program has
621finished (especially in interactive programs), but having a program that
622automatically loops as long as it has to and no longer by virtue of
623relying on its watchers stopping correctly is a thing of beauty.
348.Sp 624.Sp
349A flags value of \f(CW\*(C`EVLOOP_NONBLOCK\*(C'\fR will look for new events, will handle 625A flags value of \f(CW\*(C`EVLOOP_NONBLOCK\*(C'\fR will look for new events, will handle
350those events and any outstanding ones, but will not block your process in 626those events and any outstanding ones, but will not block your process in
351case there are no events and will return after one iteration of the loop. 627case there are no events and will return after one iteration of the loop.
352.Sp 628.Sp
353A flags value of \f(CW\*(C`EVLOOP_ONESHOT\*(C'\fR will look for new events (waiting if 629A flags value of \f(CW\*(C`EVLOOP_ONESHOT\*(C'\fR will look for new events (waiting if
354neccessary) and will handle those and any outstanding ones. It will block 630neccessary) and will handle those and any outstanding ones. It will block
355your process until at least one new event arrives, and will return after 631your process until at least one new event arrives, and will return after
356one iteration of the loop. 632one iteration of the loop. This is useful if you are waiting for some
633external event in conjunction with something not expressible using other
634libev watchers. However, a pair of \f(CW\*(C`ev_prepare\*(C'\fR/\f(CW\*(C`ev_check\*(C'\fR watchers is
635usually a better approach for this kind of thing.
357.Sp 636.Sp
358This flags value could be used to implement alternative looping
359constructs, but the \f(CW\*(C`prepare\*(C'\fR and \f(CW\*(C`check\*(C'\fR watchers provide a better and
360more generic mechanism.
361.Sp
362Here are the gory details of what ev_loop does: 637Here are the gory details of what \f(CW\*(C`ev_loop\*(C'\fR does:
363.Sp 638.Sp
364.Vb 15 639.Vb 19
640\& - Before the first iteration, call any pending watchers.
365\& 1. If there are no active watchers (reference count is zero), return. 641\& * If there are no active watchers (reference count is zero), return.
366\& 2. Queue and immediately call all prepare watchers. 642\& - Queue all prepare watchers and then call all outstanding watchers.
367\& 3. If we have been forked, recreate the kernel state. 643\& - If we have been forked, recreate the kernel state.
368\& 4. Update the kernel state with all outstanding changes. 644\& - Update the kernel state with all outstanding changes.
369\& 5. Update the "event loop time". 645\& - Update the "event loop time".
370\& 6. Calculate for how long to block. 646\& - Calculate for how long to block.
371\& 7. Block the process, waiting for events. 647\& - Block the process, waiting for any events.
648\& - Queue all outstanding I/O (fd) events.
372\& 8. Update the "event loop time" and do time jump handling. 649\& - Update the "event loop time" and do time jump handling.
373\& 9. Queue all outstanding timers. 650\& - Queue all outstanding timers.
374\& 10. Queue all outstanding periodics. 651\& - Queue all outstanding periodics.
375\& 11. If no events are pending now, queue all idle watchers. 652\& - If no events are pending now, queue all idle watchers.
376\& 12. Queue all check watchers. 653\& - Queue all check watchers.
377\& 13. Call all queued watchers in reverse order (i.e. check watchers first). 654\& - Call all queued watchers in reverse order (i.e. check watchers first).
655\& Signals and child watchers are implemented as I/O watchers, and will
656\& be handled here by queueing them when their watcher gets executed.
378\& 14. If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK 657\& - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
379\& was used, return, otherwise continue with step #1. 658\& were used, return, otherwise continue with step *.
659.Ve
660.Sp
661Example: Queue some jobs and then loop until no events are outsanding
662anymore.
663.Sp
664.Vb 4
665\& ... queue jobs here, make sure they register event watchers as long
666\& ... as they still have work to do (even an idle watcher will do..)
667\& ev_loop (my_loop, 0);
668\& ... jobs done. yeah!
380.Ve 669.Ve
381.IP "ev_unloop (loop, how)" 4 670.IP "ev_unloop (loop, how)" 4
382.IX Item "ev_unloop (loop, how)" 671.IX Item "ev_unloop (loop, how)"
383Can be used to make a call to \f(CW\*(C`ev_loop\*(C'\fR return early (but only after it 672Can be used to make a call to \f(CW\*(C`ev_loop\*(C'\fR return early (but only after it
384has processed all outstanding events). The \f(CW\*(C`how\*(C'\fR argument must be either 673has processed all outstanding events). The \f(CW\*(C`how\*(C'\fR argument must be either
… …
398example, libev itself uses this for its internal signal pipe: It is not 687example, libev itself uses this for its internal signal pipe: It is not
399visible to the libev user and should not keep \f(CW\*(C`ev_loop\*(C'\fR from exiting if 688visible to the libev user and should not keep \f(CW\*(C`ev_loop\*(C'\fR from exiting if
400no event watchers registered by it are active. It is also an excellent 689no event watchers registered by it are active. It is also an excellent
401way to do this for generic recurring timers or from within third-party 690way to do this for generic recurring timers or from within third-party
402libraries. Just remember to \fIunref after start\fR and \fIref before stop\fR. 691libraries. Just remember to \fIunref after start\fR and \fIref before stop\fR.
692.Sp
693Example: Create a signal watcher, but keep it from keeping \f(CW\*(C`ev_loop\*(C'\fR
694running when nothing else is active.
695.Sp
696.Vb 4
697\& struct ev_signal exitsig;
698\& ev_signal_init (&exitsig, sig_cb, SIGINT);
699\& ev_signal_start (loop, &exitsig);
700\& evf_unref (loop);
701.Ve
702.Sp
703Example: For some weird reason, unregister the above signal handler again.
704.Sp
705.Vb 2
706\& ev_ref (loop);
707\& ev_signal_stop (loop, &exitsig);
708.Ve
403.SH "ANATOMY OF A WATCHER" 709.SH "ANATOMY OF A WATCHER"
404.IX Header "ANATOMY OF A WATCHER" 710.IX Header "ANATOMY OF A WATCHER"
405A watcher is a structure that you create and register to record your 711A watcher is a structure that you create and register to record your
406interest in some event. For instance, if you want to wait for \s-1STDIN\s0 to 712interest in some event. For instance, if you want to wait for \s-1STDIN\s0 to
407become readable, you would create an \f(CW\*(C`ev_io\*(C'\fR watcher for that: 713become readable, you would create an \f(CW\*(C`ev_io\*(C'\fR watcher for that:
… …
443*)\*(C'\fR), and you can stop watching for events at any time by calling the 749*)\*(C'\fR), and you can stop watching for events at any time by calling the
444corresponding stop function (\f(CW\*(C`ev_<type>_stop (loop, watcher *)\*(C'\fR. 750corresponding stop function (\f(CW\*(C`ev_<type>_stop (loop, watcher *)\*(C'\fR.
445.PP 751.PP
446As long as your watcher is active (has been started but not stopped) you 752As long as your watcher is active (has been started but not stopped) you
447must not touch the values stored in it. Most specifically you must never 753must not touch the values stored in it. Most specifically you must never
448reinitialise it or call its set method. 754reinitialise it or call its \f(CW\*(C`set\*(C'\fR macro.
449.PP
450You can check whether an event is active by calling the \f(CW\*(C`ev_is_active
451(watcher *)\*(C'\fR macro. To see whether an event is outstanding (but the
452callback for it has not been called yet) you can use the \f(CW\*(C`ev_is_pending
453(watcher *)\*(C'\fR macro.
454.PP 755.PP
455Each and every callback receives the event loop pointer as first, the 756Each and every callback receives the event loop pointer as first, the
456registered watcher structure as second, and a bitset of received events as 757registered watcher structure as second, and a bitset of received events as
457third argument. 758third argument.
458.PP 759.PP
… …
483The signal specified in the \f(CW\*(C`ev_signal\*(C'\fR watcher has been received by a thread. 784The signal specified in the \f(CW\*(C`ev_signal\*(C'\fR watcher has been received by a thread.
484.ie n .IP """EV_CHILD""" 4 785.ie n .IP """EV_CHILD""" 4
485.el .IP "\f(CWEV_CHILD\fR" 4 786.el .IP "\f(CWEV_CHILD\fR" 4
486.IX Item "EV_CHILD" 787.IX Item "EV_CHILD"
487The pid specified in the \f(CW\*(C`ev_child\*(C'\fR watcher has received a status change. 788The pid specified in the \f(CW\*(C`ev_child\*(C'\fR watcher has received a status change.
789.ie n .IP """EV_STAT""" 4
790.el .IP "\f(CWEV_STAT\fR" 4
791.IX Item "EV_STAT"
792The path specified in the \f(CW\*(C`ev_stat\*(C'\fR watcher changed its attributes somehow.
488.ie n .IP """EV_IDLE""" 4 793.ie n .IP """EV_IDLE""" 4
489.el .IP "\f(CWEV_IDLE\fR" 4 794.el .IP "\f(CWEV_IDLE\fR" 4
490.IX Item "EV_IDLE" 795.IX Item "EV_IDLE"
491The \f(CW\*(C`ev_idle\*(C'\fR watcher has determined that you have nothing better to do. 796The \f(CW\*(C`ev_idle\*(C'\fR watcher has determined that you have nothing better to do.
492.ie n .IP """EV_PREPARE""" 4 797.ie n .IP """EV_PREPARE""" 4
… …
502\&\f(CW\*(C`ev_loop\*(C'\fR has gathered them, but before it invokes any callbacks for any 807\&\f(CW\*(C`ev_loop\*(C'\fR has gathered them, but before it invokes any callbacks for any
503received events. Callbacks of both watcher types can start and stop as 808received events. Callbacks of both watcher types can start and stop as
504many watchers as they want, and all of them will be taken into account 809many watchers as they want, and all of them will be taken into account
505(for example, a \f(CW\*(C`ev_prepare\*(C'\fR watcher might start an idle watcher to keep 810(for example, a \f(CW\*(C`ev_prepare\*(C'\fR watcher might start an idle watcher to keep
506\&\f(CW\*(C`ev_loop\*(C'\fR from blocking). 811\&\f(CW\*(C`ev_loop\*(C'\fR from blocking).
812.ie n .IP """EV_EMBED""" 4
813.el .IP "\f(CWEV_EMBED\fR" 4
814.IX Item "EV_EMBED"
815The embedded event loop specified in the \f(CW\*(C`ev_embed\*(C'\fR watcher needs attention.
816.ie n .IP """EV_FORK""" 4
817.el .IP "\f(CWEV_FORK\fR" 4
818.IX Item "EV_FORK"
819The event loop has been resumed in the child process after fork (see
820\&\f(CW\*(C`ev_fork\*(C'\fR).
507.ie n .IP """EV_ERROR""" 4 821.ie n .IP """EV_ERROR""" 4
508.el .IP "\f(CWEV_ERROR\fR" 4 822.el .IP "\f(CWEV_ERROR\fR" 4
509.IX Item "EV_ERROR" 823.IX Item "EV_ERROR"
510An unspecified error has occured, the watcher has been stopped. This might 824An unspecified error has occured, the watcher has been stopped. This might
511happen because the watcher could not be properly started because libev 825happen because the watcher could not be properly started because libev
… …
516Libev will usually signal a few \*(L"dummy\*(R" events together with an error, 830Libev will usually signal a few \*(L"dummy\*(R" events together with an error,
517for example it might indicate that a fd is readable or writable, and if 831for example it might indicate that a fd is readable or writable, and if
518your callbacks is well-written it can just attempt the operation and cope 832your callbacks is well-written it can just attempt the operation and cope
519with the error from \fIread()\fR or \fIwrite()\fR. This will not work in multithreaded 833with the error from \fIread()\fR or \fIwrite()\fR. This will not work in multithreaded
520programs, though, so beware. 834programs, though, so beware.
835.Sh "\s-1GENERIC\s0 \s-1WATCHER\s0 \s-1FUNCTIONS\s0"
836.IX Subsection "GENERIC WATCHER FUNCTIONS"
837In the following description, \f(CW\*(C`TYPE\*(C'\fR stands for the watcher type,
838e.g. \f(CW\*(C`timer\*(C'\fR for \f(CW\*(C`ev_timer\*(C'\fR watchers and \f(CW\*(C`io\*(C'\fR for \f(CW\*(C`ev_io\*(C'\fR watchers.
839.ie n .IP """ev_init"" (ev_TYPE *watcher, callback)" 4
840.el .IP "\f(CWev_init\fR (ev_TYPE *watcher, callback)" 4
841.IX Item "ev_init (ev_TYPE *watcher, callback)"
842This macro initialises the generic portion of a watcher. The contents
843of the watcher object can be arbitrary (so \f(CW\*(C`malloc\*(C'\fR will do). Only
844the generic parts of the watcher are initialised, you \fIneed\fR to call
845the type-specific \f(CW\*(C`ev_TYPE_set\*(C'\fR macro afterwards to initialise the
846type-specific parts. For each type there is also a \f(CW\*(C`ev_TYPE_init\*(C'\fR macro
847which rolls both calls into one.
848.Sp
849You can reinitialise a watcher at any time as long as it has been stopped
850(or never started) and there are no pending events outstanding.
851.Sp
852The callback is always of type \f(CW\*(C`void (*)(ev_loop *loop, ev_TYPE *watcher,
853int revents)\*(C'\fR.
854.ie n .IP """ev_TYPE_set"" (ev_TYPE *, [args])" 4
855.el .IP "\f(CWev_TYPE_set\fR (ev_TYPE *, [args])" 4
856.IX Item "ev_TYPE_set (ev_TYPE *, [args])"
857This macro initialises the type-specific parts of a watcher. You need to
858call \f(CW\*(C`ev_init\*(C'\fR at least once before you call this macro, but you can
859call \f(CW\*(C`ev_TYPE_set\*(C'\fR any number of times. You must not, however, call this
860macro on a watcher that is active (it can be pending, however, which is a
861difference to the \f(CW\*(C`ev_init\*(C'\fR macro).
862.Sp
863Although some watcher types do not have type-specific arguments
864(e.g. \f(CW\*(C`ev_prepare\*(C'\fR) you still need to call its \f(CW\*(C`set\*(C'\fR macro.
865.ie n .IP """ev_TYPE_init"" (ev_TYPE *watcher, callback, [args])" 4
866.el .IP "\f(CWev_TYPE_init\fR (ev_TYPE *watcher, callback, [args])" 4
867.IX Item "ev_TYPE_init (ev_TYPE *watcher, callback, [args])"
868This convinience macro rolls both \f(CW\*(C`ev_init\*(C'\fR and \f(CW\*(C`ev_TYPE_set\*(C'\fR macro
869calls into a single call. This is the most convinient method to initialise
870a watcher. The same limitations apply, of course.
871.ie n .IP """ev_TYPE_start"" (loop *, ev_TYPE *watcher)" 4
872.el .IP "\f(CWev_TYPE_start\fR (loop *, ev_TYPE *watcher)" 4
873.IX Item "ev_TYPE_start (loop *, ev_TYPE *watcher)"
874Starts (activates) the given watcher. Only active watchers will receive
875events. If the watcher is already active nothing will happen.
876.ie n .IP """ev_TYPE_stop"" (loop *, ev_TYPE *watcher)" 4
877.el .IP "\f(CWev_TYPE_stop\fR (loop *, ev_TYPE *watcher)" 4
878.IX Item "ev_TYPE_stop (loop *, ev_TYPE *watcher)"
879Stops the given watcher again (if active) and clears the pending
880status. It is possible that stopped watchers are pending (for example,
881non-repeating timers are being stopped when they become pending), but
882\&\f(CW\*(C`ev_TYPE_stop\*(C'\fR ensures that the watcher is neither active nor pending. If
883you want to free or reuse the memory used by the watcher it is therefore a
884good idea to always call its \f(CW\*(C`ev_TYPE_stop\*(C'\fR function.
885.IP "bool ev_is_active (ev_TYPE *watcher)" 4
886.IX Item "bool ev_is_active (ev_TYPE *watcher)"
887Returns a true value iff the watcher is active (i.e. it has been started
888and not yet been stopped). As long as a watcher is active you must not modify
889it.
890.IP "bool ev_is_pending (ev_TYPE *watcher)" 4
891.IX Item "bool ev_is_pending (ev_TYPE *watcher)"
892Returns a true value iff the watcher is pending, (i.e. it has outstanding
893events but its callback has not yet been invoked). As long as a watcher
894is pending (but not active) you must not call an init function on it (but
895\&\f(CW\*(C`ev_TYPE_set\*(C'\fR is safe), you must not change its priority, and you must
896make sure the watcher is available to libev (e.g. you cannot \f(CW\*(C`free ()\*(C'\fR
897it).
898.IP "callback ev_cb (ev_TYPE *watcher)" 4
899.IX Item "callback ev_cb (ev_TYPE *watcher)"
900Returns the callback currently set on the watcher.
901.IP "ev_cb_set (ev_TYPE *watcher, callback)" 4
902.IX Item "ev_cb_set (ev_TYPE *watcher, callback)"
903Change the callback. You can change the callback at virtually any time
904(modulo threads).
905.IP "ev_set_priority (ev_TYPE *watcher, priority)" 4
906.IX Item "ev_set_priority (ev_TYPE *watcher, priority)"
907.PD 0
908.IP "int ev_priority (ev_TYPE *watcher)" 4
909.IX Item "int ev_priority (ev_TYPE *watcher)"
910.PD
911Set and query the priority of the watcher. The priority is a small
912integer between \f(CW\*(C`EV_MAXPRI\*(C'\fR (default: \f(CW2\fR) and \f(CW\*(C`EV_MINPRI\*(C'\fR
913(default: \f(CW\*(C`\-2\*(C'\fR). Pending watchers with higher priority will be invoked
914before watchers with lower priority, but priority will not keep watchers
915from being executed (except for \f(CW\*(C`ev_idle\*(C'\fR watchers).
916.Sp
917This means that priorities are \fIonly\fR used for ordering callback
918invocation after new events have been received. This is useful, for
919example, to reduce latency after idling, or more often, to bind two
920watchers on the same event and make sure one is called first.
921.Sp
922If you need to suppress invocation when higher priority events are pending
923you need to look at \f(CW\*(C`ev_idle\*(C'\fR watchers, which provide this functionality.
924.Sp
925You \fImust not\fR change the priority of a watcher as long as it is active or
926pending.
927.Sp
928The default priority used by watchers when no priority has been set is
929always \f(CW0\fR, which is supposed to not be too high and not be too low :).
930.Sp
931Setting a priority outside the range of \f(CW\*(C`EV_MINPRI\*(C'\fR to \f(CW\*(C`EV_MAXPRI\*(C'\fR is
932fine, as long as you do not mind that the priority value you query might
933or might not have been adjusted to be within valid range.
934.IP "ev_invoke (loop, ev_TYPE *watcher, int revents)" 4
935.IX Item "ev_invoke (loop, ev_TYPE *watcher, int revents)"
936Invoke the \f(CW\*(C`watcher\*(C'\fR with the given \f(CW\*(C`loop\*(C'\fR and \f(CW\*(C`revents\*(C'\fR. Neither
937\&\f(CW\*(C`loop\*(C'\fR nor \f(CW\*(C`revents\*(C'\fR need to be valid as long as the watcher callback
938can deal with that fact.
939.IP "int ev_clear_pending (loop, ev_TYPE *watcher)" 4
940.IX Item "int ev_clear_pending (loop, ev_TYPE *watcher)"
941If the watcher is pending, this function returns clears its pending status
942and returns its \f(CW\*(C`revents\*(C'\fR bitset (as if its callback was invoked). If the
943watcher isn't pending it does nothing and returns \f(CW0\fR.
521.Sh "\s-1ASSOCIATING\s0 \s-1CUSTOM\s0 \s-1DATA\s0 \s-1WITH\s0 A \s-1WATCHER\s0" 944.Sh "\s-1ASSOCIATING\s0 \s-1CUSTOM\s0 \s-1DATA\s0 \s-1WITH\s0 A \s-1WATCHER\s0"
522.IX Subsection "ASSOCIATING CUSTOM DATA WITH A WATCHER" 945.IX Subsection "ASSOCIATING CUSTOM DATA WITH A WATCHER"
523Each watcher has, by default, a member \f(CW\*(C`void *data\*(C'\fR that you can change 946Each watcher has, by default, a member \f(CW\*(C`void *data\*(C'\fR that you can change
524and read at any time, libev will completely ignore it. This can be used 947and read at any time, libev will completely ignore it. This can be used
525to associate arbitrary data with your watcher. If you need more data and 948to associate arbitrary data with your watcher. If you need more data and
… …
546\& struct my_io *w = (struct my_io *)w_; 969\& struct my_io *w = (struct my_io *)w_;
547\& ... 970\& ...
548\& } 971\& }
549.Ve 972.Ve
550.PP 973.PP
551More interesting and less C\-conformant ways of catsing your callback type 974More interesting and less C\-conformant ways of casting your callback type
552have been omitted.... 975instead have been omitted.
976.PP
977Another common scenario is having some data structure with multiple
978watchers:
979.PP
980.Vb 6
981\& struct my_biggy
982\& {
983\& int some_data;
984\& ev_timer t1;
985\& ev_timer t2;
986\& }
987.Ve
988.PP
989In this case getting the pointer to \f(CW\*(C`my_biggy\*(C'\fR is a bit more complicated,
990you need to use \f(CW\*(C`offsetof\*(C'\fR:
991.PP
992.Vb 1
993\& #include <stddef.h>
994.Ve
995.PP
996.Vb 6
997\& static void
998\& t1_cb (EV_P_ struct ev_timer *w, int revents)
999\& {
1000\& struct my_biggy big = (struct my_biggy *
1001\& (((char *)w) - offsetof (struct my_biggy, t1));
1002\& }
1003.Ve
1004.PP
1005.Vb 6
1006\& static void
1007\& t2_cb (EV_P_ struct ev_timer *w, int revents)
1008\& {
1009\& struct my_biggy big = (struct my_biggy *
1010\& (((char *)w) - offsetof (struct my_biggy, t2));
1011\& }
1012.Ve
553.SH "WATCHER TYPES" 1013.SH "WATCHER TYPES"
554.IX Header "WATCHER TYPES" 1014.IX Header "WATCHER TYPES"
555This section describes each watcher in detail, but will not repeat 1015This section describes each watcher in detail, but will not repeat
556information given in the last section. 1016information given in the last section. Any initialisation/set macros,
1017functions and members specific to the watcher type are explained.
1018.PP
1019Members are additionally marked with either \fI[read\-only]\fR, meaning that,
1020while the watcher is active, you can look at the member and expect some
1021sensible content, but you must not modify it (you can modify it while the
1022watcher is stopped to your hearts content), or \fI[read\-write]\fR, which
1023means you can expect it to have some sensible content while the watcher
1024is active, but you can also modify it. Modifying it may not do something
1025sensible or take immediate effect (or do anything at all), but libev will
1026not crash or malfunction in any way.
557.ie n .Sh """ev_io"" \- is this file descriptor readable or writable" 1027.ie n .Sh """ev_io"" \- is this file descriptor readable or writable?"
558.el .Sh "\f(CWev_io\fP \- is this file descriptor readable or writable" 1028.el .Sh "\f(CWev_io\fP \- is this file descriptor readable or writable?"
559.IX Subsection "ev_io - is this file descriptor readable or writable" 1029.IX Subsection "ev_io - is this file descriptor readable or writable?"
560I/O watchers check whether a file descriptor is readable or writable 1030I/O watchers check whether a file descriptor is readable or writable
561in each iteration of the event loop (This behaviour is called 1031in each iteration of the event loop, or, more precisely, when reading
562level-triggering because you keep receiving events as long as the 1032would not block the process and writing would at least be able to write
563condition persists. Remember you can stop the watcher if you don't want to 1033some data. This behaviour is called level-triggering because you keep
564act on the event and neither want to receive future events). 1034receiving events as long as the condition persists. Remember you can stop
1035the watcher if you don't want to act on the event and neither want to
1036receive future events.
565.PP 1037.PP
566In general you can register as many read and/or write event watchers per 1038In general you can register as many read and/or write event watchers per
567fd as you want (as long as you don't confuse yourself). Setting all file 1039fd as you want (as long as you don't confuse yourself). Setting all file
568descriptors to non-blocking mode is also usually a good idea (but not 1040descriptors to non-blocking mode is also usually a good idea (but not
569required if you know what you are doing). 1041required if you know what you are doing).
570.PP 1042.PP
571You have to be careful with dup'ed file descriptors, though. Some backends 1043You have to be careful with dup'ed file descriptors, though. Some backends
572(the linux epoll backend is a notable example) cannot handle dup'ed file 1044(the linux epoll backend is a notable example) cannot handle dup'ed file
573descriptors correctly if you register interest in two or more fds pointing 1045descriptors correctly if you register interest in two or more fds pointing
574to the same underlying file/socket etc. description (that is, they share 1046to the same underlying file/socket/etc. description (that is, they share
575the same underlying \*(L"file open\*(R"). 1047the same underlying \*(L"file open\*(R").
576.PP 1048.PP
577If you must do this, then force the use of a known-to-be-good backend 1049If you must do this, then force the use of a known-to-be-good backend
578(at the time of this writing, this includes only \s-1EVMETHOD_SELECT\s0 and 1050(at the time of this writing, this includes only \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR and
579\&\s-1EVMETHOD_POLL\s0). 1051\&\f(CW\*(C`EVBACKEND_POLL\*(C'\fR).
1052.PP
1053Another thing you have to watch out for is that it is quite easy to
1054receive \*(L"spurious\*(R" readyness notifications, that is your callback might
1055be called with \f(CW\*(C`EV_READ\*(C'\fR but a subsequent \f(CW\*(C`read\*(C'\fR(2) will actually block
1056because there is no data. Not only are some backends known to create a
1057lot of those (for example solaris ports), it is very easy to get into
1058this situation even with a relatively standard program structure. Thus
1059it is best to always use non-blocking I/O: An extra \f(CW\*(C`read\*(C'\fR(2) returning
1060\&\f(CW\*(C`EAGAIN\*(C'\fR is far preferable to a program hanging until some data arrives.
1061.PP
1062If you cannot run the fd in non-blocking mode (for example you should not
1063play around with an Xlib connection), then you have to seperately re-test
1064whether a file descriptor is really ready with a known-to-be good interface
1065such as poll (fortunately in our Xlib example, Xlib already does this on
1066its own, so its quite safe to use).
580.IP "ev_io_init (ev_io *, callback, int fd, int events)" 4 1067.IP "ev_io_init (ev_io *, callback, int fd, int events)" 4
581.IX Item "ev_io_init (ev_io *, callback, int fd, int events)" 1068.IX Item "ev_io_init (ev_io *, callback, int fd, int events)"
582.PD 0 1069.PD 0
583.IP "ev_io_set (ev_io *, int fd, int events)" 4 1070.IP "ev_io_set (ev_io *, int fd, int events)" 4
584.IX Item "ev_io_set (ev_io *, int fd, int events)" 1071.IX Item "ev_io_set (ev_io *, int fd, int events)"
585.PD 1072.PD
586Configures an \f(CW\*(C`ev_io\*(C'\fR watcher. The fd is the file descriptor to rceeive 1073Configures an \f(CW\*(C`ev_io\*(C'\fR watcher. The \f(CW\*(C`fd\*(C'\fR is the file descriptor to
587events for and events is either \f(CW\*(C`EV_READ\*(C'\fR, \f(CW\*(C`EV_WRITE\*(C'\fR or \f(CW\*(C`EV_READ | 1074rceeive events for and events is either \f(CW\*(C`EV_READ\*(C'\fR, \f(CW\*(C`EV_WRITE\*(C'\fR or
588EV_WRITE\*(C'\fR to receive the given events. 1075\&\f(CW\*(C`EV_READ | EV_WRITE\*(C'\fR to receive the given events.
1076.IP "int fd [read\-only]" 4
1077.IX Item "int fd [read-only]"
1078The file descriptor being watched.
1079.IP "int events [read\-only]" 4
1080.IX Item "int events [read-only]"
1081The events being watched.
1082.PP
1083Example: Call \f(CW\*(C`stdin_readable_cb\*(C'\fR when \s-1STDIN_FILENO\s0 has become, well
1084readable, but only once. Since it is likely line\-buffered, you could
1085attempt to read a whole line in the callback.
1086.PP
1087.Vb 6
1088\& static void
1089\& stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1090\& {
1091\& ev_io_stop (loop, w);
1092\& .. read from stdin here (or from w->fd) and haqndle any I/O errors
1093\& }
1094.Ve
1095.PP
1096.Vb 6
1097\& ...
1098\& struct ev_loop *loop = ev_default_init (0);
1099\& struct ev_io stdin_readable;
1100\& ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
1101\& ev_io_start (loop, &stdin_readable);
1102\& ev_loop (loop, 0);
1103.Ve
589.ie n .Sh """ev_timer"" \- relative and optionally recurring timeouts" 1104.ie n .Sh """ev_timer"" \- relative and optionally repeating timeouts"
590.el .Sh "\f(CWev_timer\fP \- relative and optionally recurring timeouts" 1105.el .Sh "\f(CWev_timer\fP \- relative and optionally repeating timeouts"
591.IX Subsection "ev_timer - relative and optionally recurring timeouts" 1106.IX Subsection "ev_timer - relative and optionally repeating timeouts"
592Timer watchers are simple relative timers that generate an event after a 1107Timer watchers are simple relative timers that generate an event after a
593given time, and optionally repeating in regular intervals after that. 1108given time, and optionally repeating in regular intervals after that.
594.PP 1109.PP
595The timers are based on real time, that is, if you register an event that 1110The timers are based on real time, that is, if you register an event that
596times out after an hour and you reset your system clock to last years 1111times out after an hour and you reset your system clock to last years
… …
630.IP "ev_timer_again (loop)" 4 1145.IP "ev_timer_again (loop)" 4
631.IX Item "ev_timer_again (loop)" 1146.IX Item "ev_timer_again (loop)"
632This will act as if the timer timed out and restart it again if it is 1147This will act as if the timer timed out and restart it again if it is
633repeating. The exact semantics are: 1148repeating. The exact semantics are:
634.Sp 1149.Sp
1150If the timer is pending, its pending status is cleared.
1151.Sp
635If the timer is started but nonrepeating, stop it. 1152If the timer is started but nonrepeating, stop it (as if it timed out).
636.Sp 1153.Sp
637If the timer is repeating, either start it if necessary (with the repeat 1154If the timer is repeating, either start it if necessary (with the
638value), or reset the running timer to the repeat value. 1155\&\f(CW\*(C`repeat\*(C'\fR value), or reset the running timer to the \f(CW\*(C`repeat\*(C'\fR value.
639.Sp 1156.Sp
640This sounds a bit complicated, but here is a useful and typical 1157This sounds a bit complicated, but here is a useful and typical
641example: Imagine you have a tcp connection and you want a so-called idle 1158example: Imagine you have a tcp connection and you want a so-called idle
642timeout, that is, you want to be called when there have been, say, 60 1159timeout, that is, you want to be called when there have been, say, 60
643seconds of inactivity on the socket. The easiest way to do this is to 1160seconds of inactivity on the socket. The easiest way to do this is to
644configure an \f(CW\*(C`ev_timer\*(C'\fR with after=repeat=60 and calling ev_timer_again each 1161configure an \f(CW\*(C`ev_timer\*(C'\fR with a \f(CW\*(C`repeat\*(C'\fR value of \f(CW60\fR and then call
645time you successfully read or write some data. If you go into an idle 1162\&\f(CW\*(C`ev_timer_again\*(C'\fR each time you successfully read or write some data. If
646state where you do not expect data to travel on the socket, you can stop 1163you go into an idle state where you do not expect data to travel on the
647the timer, and again will automatically restart it if need be. 1164socket, you can \f(CW\*(C`ev_timer_stop\*(C'\fR the timer, and \f(CW\*(C`ev_timer_again\*(C'\fR will
1165automatically restart it if need be.
1166.Sp
1167That means you can ignore the \f(CW\*(C`after\*(C'\fR value and \f(CW\*(C`ev_timer_start\*(C'\fR
1168altogether and only ever use the \f(CW\*(C`repeat\*(C'\fR value and \f(CW\*(C`ev_timer_again\*(C'\fR:
1169.Sp
1170.Vb 8
1171\& ev_timer_init (timer, callback, 0., 5.);
1172\& ev_timer_again (loop, timer);
1173\& ...
1174\& timer->again = 17.;
1175\& ev_timer_again (loop, timer);
1176\& ...
1177\& timer->again = 10.;
1178\& ev_timer_again (loop, timer);
1179.Ve
1180.Sp
1181This is more slightly efficient then stopping/starting the timer each time
1182you want to modify its timeout value.
1183.IP "ev_tstamp repeat [read\-write]" 4
1184.IX Item "ev_tstamp repeat [read-write]"
1185The current \f(CW\*(C`repeat\*(C'\fR value. Will be used each time the watcher times out
1186or \f(CW\*(C`ev_timer_again\*(C'\fR is called and determines the next timeout (if any),
1187which is also when any modifications are taken into account.
1188.PP
1189Example: Create a timer that fires after 60 seconds.
1190.PP
1191.Vb 5
1192\& static void
1193\& one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1194\& {
1195\& .. one minute over, w is actually stopped right here
1196\& }
1197.Ve
1198.PP
1199.Vb 3
1200\& struct ev_timer mytimer;
1201\& ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1202\& ev_timer_start (loop, &mytimer);
1203.Ve
1204.PP
1205Example: Create a timeout timer that times out after 10 seconds of
1206inactivity.
1207.PP
1208.Vb 5
1209\& static void
1210\& timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1211\& {
1212\& .. ten seconds without any activity
1213\& }
1214.Ve
1215.PP
1216.Vb 4
1217\& struct ev_timer mytimer;
1218\& ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1219\& ev_timer_again (&mytimer); /* start timer */
1220\& ev_loop (loop, 0);
1221.Ve
1222.PP
1223.Vb 3
1224\& // and in some piece of code that gets executed on any "activity":
1225\& // reset the timeout to start ticking again at 10 seconds
1226\& ev_timer_again (&mytimer);
1227.Ve
648.ie n .Sh """ev_periodic"" \- to cron or not to cron" 1228.ie n .Sh """ev_periodic"" \- to cron or not to cron?"
649.el .Sh "\f(CWev_periodic\fP \- to cron or not to cron" 1229.el .Sh "\f(CWev_periodic\fP \- to cron or not to cron?"
650.IX Subsection "ev_periodic - to cron or not to cron" 1230.IX Subsection "ev_periodic - to cron or not to cron?"
651Periodic watchers are also timers of a kind, but they are very versatile 1231Periodic watchers are also timers of a kind, but they are very versatile
652(and unfortunately a bit complex). 1232(and unfortunately a bit complex).
653.PP 1233.PP
654Unlike \f(CW\*(C`ev_timer\*(C'\fR's, they are not based on real time (or relative time) 1234Unlike \f(CW\*(C`ev_timer\*(C'\fR's, they are not based on real time (or relative time)
655but on wallclock time (absolute time). You can tell a periodic watcher 1235but on wallclock time (absolute time). You can tell a periodic watcher
656to trigger \*(L"at\*(R" some specific point in time. For example, if you tell a 1236to trigger \*(L"at\*(R" some specific point in time. For example, if you tell a
657periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now () 1237periodic watcher to trigger in 10 seconds (by specifiying e.g. \f(CW\*(C`ev_now ()
658+ 10.>) and then reset your system clock to the last year, then it will 1238+ 10.\*(C'\fR) and then reset your system clock to the last year, then it will
659take a year to trigger the event (unlike an \f(CW\*(C`ev_timer\*(C'\fR, which would trigger 1239take a year to trigger the event (unlike an \f(CW\*(C`ev_timer\*(C'\fR, which would trigger
660roughly 10 seconds later and of course not if you reset your system time 1240roughly 10 seconds later).
661again).
662.PP 1241.PP
663They can also be used to implement vastly more complex timers, such as 1242They can also be used to implement vastly more complex timers, such as
664triggering an event on eahc midnight, local time. 1243triggering an event on each midnight, local time or other, complicated,
1244rules.
665.PP 1245.PP
666As with timers, the callback is guarenteed to be invoked only when the 1246As with timers, the callback is guarenteed to be invoked only when the
667time (\f(CW\*(C`at\*(C'\fR) has been passed, but if multiple periodic timers become ready 1247time (\f(CW\*(C`at\*(C'\fR) has been passed, but if multiple periodic timers become ready
668during the same loop iteration then order of execution is undefined. 1248during the same loop iteration then order of execution is undefined.
669.IP "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 4 1249.IP "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 4
… …
673.IX Item "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)" 1253.IX Item "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)"
674.PD 1254.PD
675Lots of arguments, lets sort it out... There are basically three modes of 1255Lots of arguments, lets sort it out... There are basically three modes of
676operation, and we will explain them from simplest to complex: 1256operation, and we will explain them from simplest to complex:
677.RS 4 1257.RS 4
678.IP "* absolute timer (interval = reschedule_cb = 0)" 4 1258.IP "* absolute timer (at = time, interval = reschedule_cb = 0)" 4
679.IX Item "absolute timer (interval = reschedule_cb = 0)" 1259.IX Item "absolute timer (at = time, interval = reschedule_cb = 0)"
680In this configuration the watcher triggers an event at the wallclock time 1260In this configuration the watcher triggers an event at the wallclock time
681\&\f(CW\*(C`at\*(C'\fR and doesn't repeat. It will not adjust when a time jump occurs, 1261\&\f(CW\*(C`at\*(C'\fR and doesn't repeat. It will not adjust when a time jump occurs,
682that is, if it is to be run at January 1st 2011 then it will run when the 1262that is, if it is to be run at January 1st 2011 then it will run when the
683system time reaches or surpasses this time. 1263system time reaches or surpasses this time.
684.IP "* non-repeating interval timer (interval > 0, reschedule_cb = 0)" 4 1264.IP "* non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)" 4
685.IX Item "non-repeating interval timer (interval > 0, reschedule_cb = 0)" 1265.IX Item "non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)"
686In 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
687\&\f(CW\*(C`at + N * interval\*(C'\fR time (for some integer N) and then repeat, regardless 1267\&\f(CW\*(C`at + N * interval\*(C'\fR time (for some integer N, which can also be negative)
688of any time jumps. 1268and then repeat, regardless of any time jumps.
689.Sp 1269.Sp
690This 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
691time: 1271time:
692.Sp 1272.Sp
693.Vb 1 1273.Vb 1
… …
700by 3600. 1280by 3600.
701.Sp 1281.Sp
702Another way to think about it (for the mathematically inclined) is that 1282Another way to think about it (for the mathematically inclined) is that
703\&\f(CW\*(C`ev_periodic\*(C'\fR will try to run the callback in this mode at the next possible 1283\&\f(CW\*(C`ev_periodic\*(C'\fR will try to run the callback in this mode at the next possible
704time where \f(CW\*(C`time = at (mod interval)\*(C'\fR, regardless of any time jumps. 1284time where \f(CW\*(C`time = at (mod interval)\*(C'\fR, regardless of any time jumps.
1285.Sp
1286For numerical stability it is preferable that the \f(CW\*(C`at\*(C'\fR value is near
1287\&\f(CW\*(C`ev_now ()\*(C'\fR (the current time), but there is no range requirement for
1288this value.
705.IP "* manual reschedule mode (reschedule_cb = callback)" 4 1289.IP "* manual reschedule mode (at and interval ignored, reschedule_cb = callback)" 4
706.IX Item "manual reschedule mode (reschedule_cb = callback)" 1290.IX Item "manual reschedule mode (at and interval ignored, reschedule_cb = callback)"
707In this mode the values for \f(CW\*(C`interval\*(C'\fR and \f(CW\*(C`at\*(C'\fR are both being 1291In this mode the values for \f(CW\*(C`interval\*(C'\fR and \f(CW\*(C`at\*(C'\fR are both being
708ignored. Instead, each time the periodic watcher gets scheduled, the 1292ignored. Instead, each time the periodic watcher gets scheduled, the
709reschedule callback will be called with the watcher as first, and the 1293reschedule callback will be called with the watcher as first, and the
710current time as second argument. 1294current time as second argument.
711.Sp 1295.Sp
712\&\s-1NOTE:\s0 \fIThis callback \s-1MUST\s0 \s-1NOT\s0 stop or destroy any periodic watcher, 1296\&\s-1NOTE:\s0 \fIThis callback \s-1MUST\s0 \s-1NOT\s0 stop or destroy any periodic watcher,
713ever, or make any event loop modifications\fR. If you need to stop it, 1297ever, or make any event loop modifications\fR. If you need to stop it,
714return \f(CW\*(C`now + 1e30\*(C'\fR (or so, fudge fudge) and stop it afterwards (e.g. by 1298return \f(CW\*(C`now + 1e30\*(C'\fR (or so, fudge fudge) and stop it afterwards (e.g. by
715starting a prepare watcher). 1299starting an \f(CW\*(C`ev_prepare\*(C'\fR watcher, which is legal).
716.Sp 1300.Sp
717Its prototype is \f(CW\*(C`ev_tstamp (*reschedule_cb)(struct ev_periodic *w, 1301Its prototype is \f(CW\*(C`ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
718ev_tstamp now)\*(C'\fR, e.g.: 1302ev_tstamp now)\*(C'\fR, e.g.:
719.Sp 1303.Sp
720.Vb 4 1304.Vb 4
… …
744.IX Item "ev_periodic_again (loop, ev_periodic *)" 1328.IX Item "ev_periodic_again (loop, ev_periodic *)"
745Simply stops and restarts the periodic watcher again. This is only useful 1329Simply stops and restarts the periodic watcher again. This is only useful
746when you changed some parameters or the reschedule callback would return 1330when you changed some parameters or the reschedule callback would return
747a different time than the last time it was called (e.g. in a crond like 1331a different time than the last time it was called (e.g. in a crond like
748program when the crontabs have changed). 1332program when the crontabs have changed).
1333.IP "ev_tstamp offset [read\-write]" 4
1334.IX Item "ev_tstamp offset [read-write]"
1335When repeating, this contains the offset value, otherwise this is the
1336absolute point in time (the \f(CW\*(C`at\*(C'\fR value passed to \f(CW\*(C`ev_periodic_set\*(C'\fR).
1337.Sp
1338Can be modified any time, but changes only take effect when the periodic
1339timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being called.
1340.IP "ev_tstamp interval [read\-write]" 4
1341.IX Item "ev_tstamp interval [read-write]"
1342The current interval value. Can be modified any time, but changes only
1343take effect when the periodic timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being
1344called.
1345.IP "ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read\-write]" 4
1346.IX Item "ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]"
1347The current reschedule callback, or \f(CW0\fR, if this functionality is
1348switched off. Can be changed any time, but changes only take effect when
1349the periodic timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being called.
1350.PP
1351Example: Call a callback every hour, or, more precisely, whenever the
1352system clock is divisible by 3600. The callback invocation times have
1353potentially a lot of jittering, but good long-term stability.
1354.PP
1355.Vb 5
1356\& static void
1357\& clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1358\& {
1359\& ... its now a full hour (UTC, or TAI or whatever your clock follows)
1360\& }
1361.Ve
1362.PP
1363.Vb 3
1364\& struct ev_periodic hourly_tick;
1365\& ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1366\& ev_periodic_start (loop, &hourly_tick);
1367.Ve
1368.PP
1369Example: The same as above, but use a reschedule callback to do it:
1370.PP
1371.Vb 1
1372\& #include <math.h>
1373.Ve
1374.PP
1375.Vb 5
1376\& static ev_tstamp
1377\& my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1378\& {
1379\& return fmod (now, 3600.) + 3600.;
1380\& }
1381.Ve
1382.PP
1383.Vb 1
1384\& ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1385.Ve
1386.PP
1387Example: Call a callback every hour, starting now:
1388.PP
1389.Vb 4
1390\& struct ev_periodic hourly_tick;
1391\& ev_periodic_init (&hourly_tick, clock_cb,
1392\& fmod (ev_now (loop), 3600.), 3600., 0);
1393\& ev_periodic_start (loop, &hourly_tick);
1394.Ve
749.ie n .Sh """ev_signal"" \- signal me when a signal gets signalled" 1395.ie n .Sh """ev_signal"" \- signal me when a signal gets signalled!"
750.el .Sh "\f(CWev_signal\fP \- signal me when a signal gets signalled" 1396.el .Sh "\f(CWev_signal\fP \- signal me when a signal gets signalled!"
751.IX Subsection "ev_signal - signal me when a signal gets signalled" 1397.IX Subsection "ev_signal - signal me when a signal gets signalled!"
752Signal watchers will trigger an event when the process receives a specific 1398Signal watchers will trigger an event when the process receives a specific
753signal one or more times. Even though signals are very asynchronous, libev 1399signal one or more times. Even though signals are very asynchronous, libev
754will try it's best to deliver signals synchronously, i.e. as part of the 1400will try it's best to deliver signals synchronously, i.e. as part of the
755normal event processing, like any other event. 1401normal event processing, like any other event.
756.PP 1402.PP
… …
766.IP "ev_signal_set (ev_signal *, int signum)" 4 1412.IP "ev_signal_set (ev_signal *, int signum)" 4
767.IX Item "ev_signal_set (ev_signal *, int signum)" 1413.IX Item "ev_signal_set (ev_signal *, int signum)"
768.PD 1414.PD
769Configures the watcher to trigger on the given signal number (usually one 1415Configures the watcher to trigger on the given signal number (usually one
770of the \f(CW\*(C`SIGxxx\*(C'\fR constants). 1416of the \f(CW\*(C`SIGxxx\*(C'\fR constants).
1417.IP "int signum [read\-only]" 4
1418.IX Item "int signum [read-only]"
1419The signal the watcher watches out for.
771.ie n .Sh """ev_child"" \- wait for pid status changes" 1420.ie n .Sh """ev_child"" \- watch out for process status changes"
772.el .Sh "\f(CWev_child\fP \- wait for pid status changes" 1421.el .Sh "\f(CWev_child\fP \- watch out for process status changes"
773.IX Subsection "ev_child - wait for pid status changes" 1422.IX Subsection "ev_child - watch out for process status changes"
774Child watchers trigger when your process receives a \s-1SIGCHLD\s0 in response to 1423Child watchers trigger when your process receives a \s-1SIGCHLD\s0 in response to
775some child status changes (most typically when a child of yours dies). 1424some child status changes (most typically when a child of yours dies).
776.IP "ev_child_init (ev_child *, callback, int pid)" 4 1425.IP "ev_child_init (ev_child *, callback, int pid)" 4
777.IX Item "ev_child_init (ev_child *, callback, int pid)" 1426.IX Item "ev_child_init (ev_child *, callback, int pid)"
778.PD 0 1427.PD 0
… …
783\&\fIany\fR process if \f(CW\*(C`pid\*(C'\fR is specified as \f(CW0\fR). The callback can look 1432\&\fIany\fR process if \f(CW\*(C`pid\*(C'\fR is specified as \f(CW0\fR). The callback can look
784at the \f(CW\*(C`rstatus\*(C'\fR member of the \f(CW\*(C`ev_child\*(C'\fR watcher structure to see 1433at the \f(CW\*(C`rstatus\*(C'\fR member of the \f(CW\*(C`ev_child\*(C'\fR watcher structure to see
785the status word (use the macros from \f(CW\*(C`sys/wait.h\*(C'\fR and see your systems 1434the status word (use the macros from \f(CW\*(C`sys/wait.h\*(C'\fR and see your systems
786\&\f(CW\*(C`waitpid\*(C'\fR documentation). The \f(CW\*(C`rpid\*(C'\fR member contains the pid of the 1435\&\f(CW\*(C`waitpid\*(C'\fR documentation). The \f(CW\*(C`rpid\*(C'\fR member contains the pid of the
787process causing the status change. 1436process causing the status change.
1437.IP "int pid [read\-only]" 4
1438.IX Item "int pid [read-only]"
1439The process id this watcher watches out for, or \f(CW0\fR, meaning any process id.
1440.IP "int rpid [read\-write]" 4
1441.IX Item "int rpid [read-write]"
1442The process id that detected a status change.
1443.IP "int rstatus [read\-write]" 4
1444.IX Item "int rstatus [read-write]"
1445The process exit/trace status caused by \f(CW\*(C`rpid\*(C'\fR (see your systems
1446\&\f(CW\*(C`waitpid\*(C'\fR and \f(CW\*(C`sys/wait.h\*(C'\fR documentation for details).
1447.PP
1448Example: Try to exit cleanly on \s-1SIGINT\s0 and \s-1SIGTERM\s0.
1449.PP
1450.Vb 5
1451\& static void
1452\& sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1453\& {
1454\& ev_unloop (loop, EVUNLOOP_ALL);
1455\& }
1456.Ve
1457.PP
1458.Vb 3
1459\& struct ev_signal signal_watcher;
1460\& ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1461\& ev_signal_start (loop, &sigint_cb);
1462.Ve
1463.ie n .Sh """ev_stat"" \- did the file attributes just change?"
1464.el .Sh "\f(CWev_stat\fP \- did the file attributes just change?"
1465.IX Subsection "ev_stat - did the file attributes just change?"
1466This watches a filesystem path for attribute changes. That is, it calls
1467\&\f(CW\*(C`stat\*(C'\fR regularly (or when the \s-1OS\s0 says it changed) and sees if it changed
1468compared to the last time, invoking the callback if it did.
1469.PP
1470The path does not need to exist: changing from \*(L"path exists\*(R" to \*(L"path does
1471not exist\*(R" is a status change like any other. The condition \*(L"path does
1472not exist\*(R" is signified by the \f(CW\*(C`st_nlink\*(C'\fR field being zero (which is
1473otherwise always forced to be at least one) and all the other fields of
1474the stat buffer having unspecified contents.
1475.PP
1476The path \fIshould\fR be absolute and \fImust not\fR end in a slash. If it is
1477relative and your working directory changes, the behaviour is undefined.
1478.PP
1479Since there is no standard to do this, the portable implementation simply
1480calls \f(CW\*(C`stat (2)\*(C'\fR regularly on the path to see if it changed somehow. You
1481can specify a recommended polling interval for this case. If you specify
1482a polling interval of \f(CW0\fR (highly recommended!) then a \fIsuitable,
1483unspecified default\fR value will be used (which you can expect to be around
1484five seconds, although this might change dynamically). Libev will also
1485impose a minimum interval which is currently around \f(CW0.1\fR, but thats
1486usually overkill.
1487.PP
1488This watcher type is not meant for massive numbers of stat watchers,
1489as even with OS-supported change notifications, this can be
1490resource\-intensive.
1491.PP
1492At the time of this writing, only the Linux inotify interface is
1493implemented (implementing kqueue support is left as an exercise for the
1494reader). Inotify will be used to give hints only and should not change the
1495semantics of \f(CW\*(C`ev_stat\*(C'\fR watchers, which means that libev sometimes needs
1496to fall back to regular polling again even with inotify, but changes are
1497usually detected immediately, and if the file exists there will be no
1498polling.
1499.IP "ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)" 4
1500.IX Item "ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)"
1501.PD 0
1502.IP "ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)" 4
1503.IX Item "ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)"
1504.PD
1505Configures the watcher to wait for status changes of the given
1506\&\f(CW\*(C`path\*(C'\fR. The \f(CW\*(C`interval\*(C'\fR is a hint on how quickly a change is expected to
1507be detected and should normally be specified as \f(CW0\fR to let libev choose
1508a suitable value. The memory pointed to by \f(CW\*(C`path\*(C'\fR must point to the same
1509path for as long as the watcher is active.
1510.Sp
1511The callback will be receive \f(CW\*(C`EV_STAT\*(C'\fR when a change was detected,
1512relative to the attributes at the time the watcher was started (or the
1513last change was detected).
1514.IP "ev_stat_stat (ev_stat *)" 4
1515.IX Item "ev_stat_stat (ev_stat *)"
1516Updates the stat buffer immediately with new values. If you change the
1517watched path in your callback, you could call this fucntion to avoid
1518detecting this change (while introducing a race condition). Can also be
1519useful simply to find out the new values.
1520.IP "ev_statdata attr [read\-only]" 4
1521.IX Item "ev_statdata attr [read-only]"
1522The most-recently detected attributes of the file. Although the type is of
1523\&\f(CW\*(C`ev_statdata\*(C'\fR, this is usually the (or one of the) \f(CW\*(C`struct stat\*(C'\fR types
1524suitable for your system. If the \f(CW\*(C`st_nlink\*(C'\fR member is \f(CW0\fR, then there
1525was some error while \f(CW\*(C`stat\*(C'\fRing the file.
1526.IP "ev_statdata prev [read\-only]" 4
1527.IX Item "ev_statdata prev [read-only]"
1528The previous attributes of the file. The callback gets invoked whenever
1529\&\f(CW\*(C`prev\*(C'\fR != \f(CW\*(C`attr\*(C'\fR.
1530.IP "ev_tstamp interval [read\-only]" 4
1531.IX Item "ev_tstamp interval [read-only]"
1532The specified interval.
1533.IP "const char *path [read\-only]" 4
1534.IX Item "const char *path [read-only]"
1535The filesystem path that is being watched.
1536.PP
1537Example: Watch \f(CW\*(C`/etc/passwd\*(C'\fR for attribute changes.
1538.PP
1539.Vb 15
1540\& static void
1541\& passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1542\& {
1543\& /* /etc/passwd changed in some way */
1544\& if (w->attr.st_nlink)
1545\& {
1546\& printf ("passwd current size %ld\en", (long)w->attr.st_size);
1547\& printf ("passwd current atime %ld\en", (long)w->attr.st_mtime);
1548\& printf ("passwd current mtime %ld\en", (long)w->attr.st_mtime);
1549\& }
1550\& else
1551\& /* you shalt not abuse printf for puts */
1552\& puts ("wow, /etc/passwd is not there, expect problems. "
1553\& "if this is windows, they already arrived\en");
1554\& }
1555.Ve
1556.PP
1557.Vb 2
1558\& ...
1559\& ev_stat passwd;
1560.Ve
1561.PP
1562.Vb 2
1563\& ev_stat_init (&passwd, passwd_cb, "/etc/passwd");
1564\& ev_stat_start (loop, &passwd);
1565.Ve
788.ie n .Sh """ev_idle"" \- when you've got nothing better to do" 1566.ie n .Sh """ev_idle"" \- when you've got nothing better to do..."
789.el .Sh "\f(CWev_idle\fP \- when you've got nothing better to do" 1567.el .Sh "\f(CWev_idle\fP \- when you've got nothing better to do..."
790.IX Subsection "ev_idle - when you've got nothing better to do" 1568.IX Subsection "ev_idle - when you've got nothing better to do..."
791Idle watchers trigger events when there are no other events are pending 1569Idle watchers trigger events when no other events of the same or higher
792(prepare, check and other idle watchers do not count). That is, as long 1570priority are pending (prepare, check and other idle watchers do not
793as your process is busy handling sockets or timeouts (or even signals, 1571count).
794imagine) it will not be triggered. But when your process is idle all idle 1572.PP
795watchers are being called again and again, once per event loop iteration \- 1573That is, as long as your process is busy handling sockets or timeouts
1574(or even signals, imagine) of the same or higher priority it will not be
1575triggered. But when your process is idle (or only lower-priority watchers
1576are pending), the idle watchers are being called once per event loop
796until stopped, that is, or your process receives more events and becomes 1577iteration \- until stopped, that is, or your process receives more events
797busy. 1578and becomes busy again with higher priority stuff.
798.PP 1579.PP
799The most noteworthy effect is that as long as any idle watchers are 1580The most noteworthy effect is that as long as any idle watchers are
800active, the process will not block when waiting for new events. 1581active, the process will not block when waiting for new events.
801.PP 1582.PP
802Apart from keeping your process non-blocking (which is a useful 1583Apart from keeping your process non-blocking (which is a useful
… …
806.IP "ev_idle_init (ev_signal *, callback)" 4 1587.IP "ev_idle_init (ev_signal *, callback)" 4
807.IX Item "ev_idle_init (ev_signal *, callback)" 1588.IX Item "ev_idle_init (ev_signal *, callback)"
808Initialises and configures the idle watcher \- it has no parameters of any 1589Initialises and configures the idle watcher \- it has no parameters of any
809kind. There is a \f(CW\*(C`ev_idle_set\*(C'\fR macro, but using it is utterly pointless, 1590kind. There is a \f(CW\*(C`ev_idle_set\*(C'\fR macro, but using it is utterly pointless,
810believe me. 1591believe me.
1592.PP
1593Example: Dynamically allocate an \f(CW\*(C`ev_idle\*(C'\fR watcher, start it, and in the
1594callback, free it. Also, use no error checking, as usual.
1595.PP
1596.Vb 7
1597\& static void
1598\& idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1599\& {
1600\& free (w);
1601\& // now do something you wanted to do when the program has
1602\& // no longer asnything immediate to do.
1603\& }
1604.Ve
1605.PP
1606.Vb 3
1607\& struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1608\& ev_idle_init (idle_watcher, idle_cb);
1609\& ev_idle_start (loop, idle_cb);
1610.Ve
811.ie n .Sh """ev_prepare""\fP and \f(CW""ev_check"" \- customise your event loop" 1611.ie n .Sh """ev_prepare""\fP and \f(CW""ev_check"" \- customise your event loop!"
812.el .Sh "\f(CWev_prepare\fP and \f(CWev_check\fP \- customise your event loop" 1612.el .Sh "\f(CWev_prepare\fP and \f(CWev_check\fP \- customise your event loop!"
813.IX Subsection "ev_prepare and ev_check - customise your event loop" 1613.IX Subsection "ev_prepare and ev_check - customise your event loop!"
814Prepare and check watchers are usually (but not always) used in tandem: 1614Prepare and check watchers are usually (but not always) used in tandem:
815prepare watchers get invoked before the process blocks and check watchers 1615prepare watchers get invoked before the process blocks and check watchers
816afterwards. 1616afterwards.
817.PP 1617.PP
1618You \fImust not\fR call \f(CW\*(C`ev_loop\*(C'\fR or similar functions that enter
1619the current event loop from either \f(CW\*(C`ev_prepare\*(C'\fR or \f(CW\*(C`ev_check\*(C'\fR
1620watchers. Other loops than the current one are fine, however. The
1621rationale behind this is that you do not need to check for recursion in
1622those watchers, i.e. the sequence will always be \f(CW\*(C`ev_prepare\*(C'\fR, blocking,
1623\&\f(CW\*(C`ev_check\*(C'\fR so if you have one watcher of each kind they will always be
1624called in pairs bracketing the blocking call.
1625.PP
818Their main purpose is to integrate other event mechanisms into libev. This 1626Their main purpose is to integrate other event mechanisms into libev and
819could be used, for example, to track variable changes, implement your own 1627their use is somewhat advanced. This could be used, for example, to track
820watchers, integrate net-snmp or a coroutine library and lots more. 1628variable changes, implement your own watchers, integrate net-snmp or a
1629coroutine library and lots more. They are also occasionally useful if
1630you cache some data and want to flush it before blocking (for example,
1631in X programs you might want to do an \f(CW\*(C`XFlush ()\*(C'\fR in an \f(CW\*(C`ev_prepare\*(C'\fR
1632watcher).
821.PP 1633.PP
822This is done by examining in each prepare call which file descriptors need 1634This is done by examining in each prepare call which file descriptors need
823to be watched by the other library, registering \f(CW\*(C`ev_io\*(C'\fR watchers for 1635to be watched by the other library, registering \f(CW\*(C`ev_io\*(C'\fR watchers for
824them and starting an \f(CW\*(C`ev_timer\*(C'\fR watcher for any timeouts (many libraries 1636them and starting an \f(CW\*(C`ev_timer\*(C'\fR watcher for any timeouts (many libraries
825provide just this functionality). Then, in the check watcher you check for 1637provide just this functionality). Then, in the check watcher you check for
… …
834are ready to run (it's actually more complicated: it only runs coroutines 1646are ready to run (it's actually more complicated: it only runs coroutines
835with priority higher than or equal to the event loop and one coroutine 1647with priority higher than or equal to the event loop and one coroutine
836of lower priority, but only once, using idle watchers to keep the event 1648of lower priority, but only once, using idle watchers to keep the event
837loop from blocking if lower-priority coroutines are active, thus mapping 1649loop from blocking if lower-priority coroutines are active, thus mapping
838low-priority coroutines to idle/background tasks). 1650low-priority coroutines to idle/background tasks).
1651.PP
1652It is recommended to give \f(CW\*(C`ev_check\*(C'\fR watchers highest (\f(CW\*(C`EV_MAXPRI\*(C'\fR)
1653priority, to ensure that they are being run before any other watchers
1654after the poll. Also, \f(CW\*(C`ev_check\*(C'\fR watchers (and \f(CW\*(C`ev_prepare\*(C'\fR watchers,
1655too) should not activate (\*(L"feed\*(R") events into libev. While libev fully
1656supports this, they will be called before other \f(CW\*(C`ev_check\*(C'\fR watchers did
1657their job. As \f(CW\*(C`ev_check\*(C'\fR watchers are often used to embed other event
1658loops those other event loops might be in an unusable state until their
1659\&\f(CW\*(C`ev_check\*(C'\fR watcher ran (always remind yourself to coexist peacefully with
1660others).
839.IP "ev_prepare_init (ev_prepare *, callback)" 4 1661.IP "ev_prepare_init (ev_prepare *, callback)" 4
840.IX Item "ev_prepare_init (ev_prepare *, callback)" 1662.IX Item "ev_prepare_init (ev_prepare *, callback)"
841.PD 0 1663.PD 0
842.IP "ev_check_init (ev_check *, callback)" 4 1664.IP "ev_check_init (ev_check *, callback)" 4
843.IX Item "ev_check_init (ev_check *, callback)" 1665.IX Item "ev_check_init (ev_check *, callback)"
844.PD 1666.PD
845Initialises and configures the prepare or check watcher \- they have no 1667Initialises and configures the prepare or check watcher \- they have no
846parameters of any kind. There are \f(CW\*(C`ev_prepare_set\*(C'\fR and \f(CW\*(C`ev_check_set\*(C'\fR 1668parameters of any kind. There are \f(CW\*(C`ev_prepare_set\*(C'\fR and \f(CW\*(C`ev_check_set\*(C'\fR
847macros, but using them is utterly, utterly and completely pointless. 1669macros, but using them is utterly, utterly and completely pointless.
1670.PP
1671There are a number of principal ways to embed other event loops or modules
1672into libev. Here are some ideas on how to include libadns into libev
1673(there is a Perl module named \f(CW\*(C`EV::ADNS\*(C'\fR that does this, which you could
1674use for an actually working example. Another Perl module named \f(CW\*(C`EV::Glib\*(C'\fR
1675embeds a Glib main context into libev, and finally, \f(CW\*(C`Glib::EV\*(C'\fR embeds \s-1EV\s0
1676into the Glib event loop).
1677.PP
1678Method 1: Add \s-1IO\s0 watchers and a timeout watcher in a prepare handler,
1679and in a check watcher, destroy them and call into libadns. What follows
1680is pseudo-code only of course. This requires you to either use a low
1681priority for the check watcher or use \f(CW\*(C`ev_clear_pending\*(C'\fR explicitly, as
1682the callbacks for the IO/timeout watchers might not have been called yet.
1683.PP
1684.Vb 2
1685\& static ev_io iow [nfd];
1686\& static ev_timer tw;
1687.Ve
1688.PP
1689.Vb 4
1690\& static void
1691\& io_cb (ev_loop *loop, ev_io *w, int revents)
1692\& {
1693\& }
1694.Ve
1695.PP
1696.Vb 8
1697\& // create io watchers for each fd and a timer before blocking
1698\& static void
1699\& adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1700\& {
1701\& int timeout = 3600000;
1702\& struct pollfd fds [nfd];
1703\& // actual code will need to loop here and realloc etc.
1704\& adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1705.Ve
1706.PP
1707.Vb 3
1708\& /* the callback is illegal, but won't be called as we stop during check */
1709\& ev_timer_init (&tw, 0, timeout * 1e-3);
1710\& ev_timer_start (loop, &tw);
1711.Ve
1712.PP
1713.Vb 6
1714\& // create one ev_io per pollfd
1715\& for (int i = 0; i < nfd; ++i)
1716\& {
1717\& ev_io_init (iow + i, io_cb, fds [i].fd,
1718\& ((fds [i].events & POLLIN ? EV_READ : 0)
1719\& | (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1720.Ve
1721.PP
1722.Vb 4
1723\& fds [i].revents = 0;
1724\& ev_io_start (loop, iow + i);
1725\& }
1726\& }
1727.Ve
1728.PP
1729.Vb 5
1730\& // stop all watchers after blocking
1731\& static void
1732\& adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1733\& {
1734\& ev_timer_stop (loop, &tw);
1735.Ve
1736.PP
1737.Vb 8
1738\& for (int i = 0; i < nfd; ++i)
1739\& {
1740\& // set the relevant poll flags
1741\& // could also call adns_processreadable etc. here
1742\& struct pollfd *fd = fds + i;
1743\& int revents = ev_clear_pending (iow + i);
1744\& if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1745\& if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1746.Ve
1747.PP
1748.Vb 3
1749\& // now stop the watcher
1750\& ev_io_stop (loop, iow + i);
1751\& }
1752.Ve
1753.PP
1754.Vb 2
1755\& adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1756\& }
1757.Ve
1758.PP
1759Method 2: This would be just like method 1, but you run \f(CW\*(C`adns_afterpoll\*(C'\fR
1760in the prepare watcher and would dispose of the check watcher.
1761.PP
1762Method 3: If the module to be embedded supports explicit event
1763notification (adns does), you can also make use of the actual watcher
1764callbacks, and only destroy/create the watchers in the prepare watcher.
1765.PP
1766.Vb 5
1767\& static void
1768\& timer_cb (EV_P_ ev_timer *w, int revents)
1769\& {
1770\& adns_state ads = (adns_state)w->data;
1771\& update_now (EV_A);
1772.Ve
1773.PP
1774.Vb 2
1775\& adns_processtimeouts (ads, &tv_now);
1776\& }
1777.Ve
1778.PP
1779.Vb 5
1780\& static void
1781\& io_cb (EV_P_ ev_io *w, int revents)
1782\& {
1783\& adns_state ads = (adns_state)w->data;
1784\& update_now (EV_A);
1785.Ve
1786.PP
1787.Vb 3
1788\& if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
1789\& if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
1790\& }
1791.Ve
1792.PP
1793.Vb 1
1794\& // do not ever call adns_afterpoll
1795.Ve
1796.PP
1797Method 4: Do not use a prepare or check watcher because the module you
1798want to embed is too inflexible to support it. Instead, youc na override
1799their poll function. The drawback with this solution is that the main
1800loop is now no longer controllable by \s-1EV\s0. The \f(CW\*(C`Glib::EV\*(C'\fR module does
1801this.
1802.PP
1803.Vb 4
1804\& static gint
1805\& event_poll_func (GPollFD *fds, guint nfds, gint timeout)
1806\& {
1807\& int got_events = 0;
1808.Ve
1809.PP
1810.Vb 2
1811\& for (n = 0; n < nfds; ++n)
1812\& // create/start io watcher that sets the relevant bits in fds[n] and increment got_events
1813.Ve
1814.PP
1815.Vb 2
1816\& if (timeout >= 0)
1817\& // create/start timer
1818.Ve
1819.PP
1820.Vb 2
1821\& // poll
1822\& ev_loop (EV_A_ 0);
1823.Ve
1824.PP
1825.Vb 3
1826\& // stop timer again
1827\& if (timeout >= 0)
1828\& ev_timer_stop (EV_A_ &to);
1829.Ve
1830.PP
1831.Vb 3
1832\& // stop io watchers again - their callbacks should have set
1833\& for (n = 0; n < nfds; ++n)
1834\& ev_io_stop (EV_A_ iow [n]);
1835.Ve
1836.PP
1837.Vb 2
1838\& return got_events;
1839\& }
1840.Ve
1841.ie n .Sh """ev_embed"" \- when one backend isn't enough..."
1842.el .Sh "\f(CWev_embed\fP \- when one backend isn't enough..."
1843.IX Subsection "ev_embed - when one backend isn't enough..."
1844This is a rather advanced watcher type that lets you embed one event loop
1845into another (currently only \f(CW\*(C`ev_io\*(C'\fR events are supported in the embedded
1846loop, other types of watchers might be handled in a delayed or incorrect
1847fashion and must not be used).
1848.PP
1849There are primarily two reasons you would want that: work around bugs and
1850prioritise I/O.
1851.PP
1852As an example for a bug workaround, the kqueue backend might only support
1853sockets on some platform, so it is unusable as generic backend, but you
1854still want to make use of it because you have many sockets and it scales
1855so nicely. In this case, you would create a kqueue-based loop and embed it
1856into your default loop (which might use e.g. poll). Overall operation will
1857be a bit slower because first libev has to poll and then call kevent, but
1858at least you can use both at what they are best.
1859.PP
1860As for prioritising I/O: rarely you have the case where some fds have
1861to be watched and handled very quickly (with low latency), and even
1862priorities and idle watchers might have too much overhead. In this case
1863you would put all the high priority stuff in one loop and all the rest in
1864a second one, and embed the second one in the first.
1865.PP
1866As long as the watcher is active, the callback will be invoked every time
1867there might be events pending in the embedded loop. The callback must then
1868call \f(CW\*(C`ev_embed_sweep (mainloop, watcher)\*(C'\fR to make a single sweep and invoke
1869their callbacks (you could also start an idle watcher to give the embedded
1870loop strictly lower priority for example). You can also set the callback
1871to \f(CW0\fR, in which case the embed watcher will automatically execute the
1872embedded loop sweep.
1873.PP
1874As long as the watcher is started it will automatically handle events. The
1875callback will be invoked whenever some events have been handled. You can
1876set the callback to \f(CW0\fR to avoid having to specify one if you are not
1877interested in that.
1878.PP
1879Also, there have not currently been made special provisions for forking:
1880when you fork, you not only have to call \f(CW\*(C`ev_loop_fork\*(C'\fR on both loops,
1881but you will also have to stop and restart any \f(CW\*(C`ev_embed\*(C'\fR watchers
1882yourself.
1883.PP
1884Unfortunately, not all backends are embeddable, only the ones returned by
1885\&\f(CW\*(C`ev_embeddable_backends\*(C'\fR are, which, unfortunately, does not include any
1886portable one.
1887.PP
1888So when you want to use this feature you will always have to be prepared
1889that you cannot get an embeddable loop. The recommended way to get around
1890this is to have a separate variables for your embeddable loop, try to
1891create it, and if that fails, use the normal loop for everything:
1892.PP
1893.Vb 3
1894\& struct ev_loop *loop_hi = ev_default_init (0);
1895\& struct ev_loop *loop_lo = 0;
1896\& struct ev_embed embed;
1897.Ve
1898.PP
1899.Vb 5
1900\& // see if there is a chance of getting one that works
1901\& // (remember that a flags value of 0 means autodetection)
1902\& loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1903\& ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1904\& : 0;
1905.Ve
1906.PP
1907.Vb 8
1908\& // if we got one, then embed it, otherwise default to loop_hi
1909\& if (loop_lo)
1910\& {
1911\& ev_embed_init (&embed, 0, loop_lo);
1912\& ev_embed_start (loop_hi, &embed);
1913\& }
1914\& else
1915\& loop_lo = loop_hi;
1916.Ve
1917.IP "ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)" 4
1918.IX Item "ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)"
1919.PD 0
1920.IP "ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)" 4
1921.IX Item "ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)"
1922.PD
1923Configures the watcher to embed the given loop, which must be
1924embeddable. If the callback is \f(CW0\fR, then \f(CW\*(C`ev_embed_sweep\*(C'\fR will be
1925invoked automatically, otherwise it is the responsibility of the callback
1926to invoke it (it will continue to be called until the sweep has been done,
1927if you do not want thta, you need to temporarily stop the embed watcher).
1928.IP "ev_embed_sweep (loop, ev_embed *)" 4
1929.IX Item "ev_embed_sweep (loop, ev_embed *)"
1930Make a single, non-blocking sweep over the embedded loop. This works
1931similarly to \f(CW\*(C`ev_loop (embedded_loop, EVLOOP_NONBLOCK)\*(C'\fR, but in the most
1932apropriate way for embedded loops.
1933.IP "struct ev_loop *loop [read\-only]" 4
1934.IX Item "struct ev_loop *loop [read-only]"
1935The embedded event loop.
1936.ie n .Sh """ev_fork"" \- the audacity to resume the event loop after a fork"
1937.el .Sh "\f(CWev_fork\fP \- the audacity to resume the event loop after a fork"
1938.IX Subsection "ev_fork - the audacity to resume the event loop after a fork"
1939Fork watchers are called when a \f(CW\*(C`fork ()\*(C'\fR was detected (usually because
1940whoever is a good citizen cared to tell libev about it by calling
1941\&\f(CW\*(C`ev_default_fork\*(C'\fR or \f(CW\*(C`ev_loop_fork\*(C'\fR). The invocation is done before the
1942event loop blocks next and before \f(CW\*(C`ev_check\*(C'\fR watchers are being called,
1943and only in the child after the fork. If whoever good citizen calling
1944\&\f(CW\*(C`ev_default_fork\*(C'\fR cheats and calls it in the wrong process, the fork
1945handlers will be invoked, too, of course.
1946.IP "ev_fork_init (ev_signal *, callback)" 4
1947.IX Item "ev_fork_init (ev_signal *, callback)"
1948Initialises and configures the fork watcher \- it has no parameters of any
1949kind. There is a \f(CW\*(C`ev_fork_set\*(C'\fR macro, but using it is utterly pointless,
1950believe me.
848.SH "OTHER FUNCTIONS" 1951.SH "OTHER FUNCTIONS"
849.IX Header "OTHER FUNCTIONS" 1952.IX Header "OTHER FUNCTIONS"
850There are some other functions of possible interest. Described. Here. Now. 1953There are some other functions of possible interest. Described. Here. Now.
851.IP "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 4 1954.IP "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 4
852.IX Item "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 1955.IX Item "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)"
… …
881.Ve 1984.Ve
882.Sp 1985.Sp
883.Vb 1 1986.Vb 1
884\& ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0); 1987\& ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
885.Ve 1988.Ve
886.IP "ev_feed_event (loop, watcher, int events)" 4 1989.IP "ev_feed_event (ev_loop *, watcher *, int revents)" 4
887.IX Item "ev_feed_event (loop, watcher, int events)" 1990.IX Item "ev_feed_event (ev_loop *, watcher *, int revents)"
888Feeds the given event set into the event loop, as if the specified event 1991Feeds the given event set into the event loop, as if the specified event
889had happened for the specified watcher (which must be a pointer to an 1992had happened for the specified watcher (which must be a pointer to an
890initialised but not necessarily started event watcher). 1993initialised but not necessarily started event watcher).
891.IP "ev_feed_fd_event (loop, int fd, int revents)" 4 1994.IP "ev_feed_fd_event (ev_loop *, int fd, int revents)" 4
892.IX Item "ev_feed_fd_event (loop, int fd, int revents)" 1995.IX Item "ev_feed_fd_event (ev_loop *, int fd, int revents)"
893Feed an event on the given fd, as if a file descriptor backend detected 1996Feed an event on the given fd, as if a file descriptor backend detected
894the given events it. 1997the given events it.
895.IP "ev_feed_signal_event (loop, int signum)" 4 1998.IP "ev_feed_signal_event (ev_loop *loop, int signum)" 4
896.IX Item "ev_feed_signal_event (loop, int signum)" 1999.IX Item "ev_feed_signal_event (ev_loop *loop, int signum)"
897Feed an event as if the given signal occured (loop must be the default loop!). 2000Feed an event as if the given signal occured (\f(CW\*(C`loop\*(C'\fR must be the default
2001loop!).
898.SH "LIBEVENT EMULATION" 2002.SH "LIBEVENT EMULATION"
899.IX Header "LIBEVENT EMULATION" 2003.IX Header "LIBEVENT EMULATION"
900Libev offers a compatibility emulation layer for libevent. It cannot 2004Libev offers a compatibility emulation layer for libevent. It cannot
901emulate the internals of libevent, so here are some usage hints: 2005emulate the internals of libevent, so here are some usage hints:
902.IP "* Use it by including <event.h>, as usual." 4 2006.IP "* Use it by including <event.h>, as usual." 4
… …
913.IP "* The libev emulation is \fInot\fR \s-1ABI\s0 compatible to libevent, you need to use the libev header file and library." 4 2017.IP "* The libev emulation is \fInot\fR \s-1ABI\s0 compatible to libevent, you need to use the libev header file and library." 4
914.IX Item "The libev emulation is not ABI compatible to libevent, you need to use the libev header file and library." 2018.IX Item "The libev emulation is not ABI compatible to libevent, you need to use the libev header file and library."
915.PD 2019.PD
916.SH "\*(C+ SUPPORT" 2020.SH "\*(C+ SUPPORT"
917.IX Header " SUPPORT" 2021.IX Header " SUPPORT"
918\&\s-1TBD\s0. 2022Libev comes with some simplistic wrapper classes for \*(C+ that mainly allow
2023you to use some convinience methods to start/stop watchers and also change
2024the callback model to a model using method callbacks on objects.
2025.PP
2026To use it,
2027.PP
2028.Vb 1
2029\& #include <ev++.h>
2030.Ve
2031.PP
2032This automatically includes \fIev.h\fR and puts all of its definitions (many
2033of them macros) into the global namespace. All \*(C+ specific things are
2034put into the \f(CW\*(C`ev\*(C'\fR namespace. It should support all the same embedding
2035options as \fIev.h\fR, most notably \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR.
2036.PP
2037Care has been taken to keep the overhead low. The only data member the \*(C+
2038classes add (compared to plain C\-style watchers) is the event loop pointer
2039that the watcher is associated with (or no additional members at all if
2040you disable \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR when embedding libev).
2041.PP
2042Currently, functions, and static and non-static member functions can be
2043used as callbacks. Other types should be easy to add as long as they only
2044need one additional pointer for context. If you need support for other
2045types of functors please contact the author (preferably after implementing
2046it).
2047.PP
2048Here is a list of things available in the \f(CW\*(C`ev\*(C'\fR namespace:
2049.ie n .IP """ev::READ""\fR, \f(CW""ev::WRITE"" etc." 4
2050.el .IP "\f(CWev::READ\fR, \f(CWev::WRITE\fR etc." 4
2051.IX Item "ev::READ, ev::WRITE etc."
2052These are just enum values with the same values as the \f(CW\*(C`EV_READ\*(C'\fR etc.
2053macros from \fIev.h\fR.
2054.ie n .IP """ev::tstamp""\fR, \f(CW""ev::now""" 4
2055.el .IP "\f(CWev::tstamp\fR, \f(CWev::now\fR" 4
2056.IX Item "ev::tstamp, ev::now"
2057Aliases to the same types/functions as with the \f(CW\*(C`ev_\*(C'\fR prefix.
2058.ie n .IP """ev::io""\fR, \f(CW""ev::timer""\fR, \f(CW""ev::periodic""\fR, \f(CW""ev::idle""\fR, \f(CW""ev::sig"" etc." 4
2059.el .IP "\f(CWev::io\fR, \f(CWev::timer\fR, \f(CWev::periodic\fR, \f(CWev::idle\fR, \f(CWev::sig\fR etc." 4
2060.IX Item "ev::io, ev::timer, ev::periodic, ev::idle, ev::sig etc."
2061For each \f(CW\*(C`ev_TYPE\*(C'\fR watcher in \fIev.h\fR there is a corresponding class of
2062the same name in the \f(CW\*(C`ev\*(C'\fR namespace, with the exception of \f(CW\*(C`ev_signal\*(C'\fR
2063which is called \f(CW\*(C`ev::sig\*(C'\fR to avoid clashes with the \f(CW\*(C`signal\*(C'\fR macro
2064defines by many implementations.
2065.Sp
2066All of those classes have these methods:
2067.RS 4
2068.IP "ev::TYPE::TYPE ()" 4
2069.IX Item "ev::TYPE::TYPE ()"
2070.PD 0
2071.IP "ev::TYPE::TYPE (struct ev_loop *)" 4
2072.IX Item "ev::TYPE::TYPE (struct ev_loop *)"
2073.IP "ev::TYPE::~TYPE" 4
2074.IX Item "ev::TYPE::~TYPE"
2075.PD
2076The constructor (optionally) takes an event loop to associate the watcher
2077with. If it is omitted, it will use \f(CW\*(C`EV_DEFAULT\*(C'\fR.
2078.Sp
2079The constructor calls \f(CW\*(C`ev_init\*(C'\fR for you, which means you have to call the
2080\&\f(CW\*(C`set\*(C'\fR method before starting it.
2081.Sp
2082It will not set a callback, however: You have to call the templated \f(CW\*(C`set\*(C'\fR
2083method to set a callback before you can start the watcher.
2084.Sp
2085(The reason why you have to use a method is a limitation in \*(C+ which does
2086not allow explicit template arguments for constructors).
2087.Sp
2088The destructor automatically stops the watcher if it is active.
2089.IP "w\->set<class, &class::method> (object *)" 4
2090.IX Item "w->set<class, &class::method> (object *)"
2091This method sets the callback method to call. The method has to have a
2092signature of \f(CW\*(C`void (*)(ev_TYPE &, int)\*(C'\fR, it receives the watcher as
2093first argument and the \f(CW\*(C`revents\*(C'\fR as second. The object must be given as
2094parameter and is stored in the \f(CW\*(C`data\*(C'\fR member of the watcher.
2095.Sp
2096This method synthesizes efficient thunking code to call your method from
2097the C callback that libev requires. If your compiler can inline your
2098callback (i.e. it is visible to it at the place of the \f(CW\*(C`set\*(C'\fR call and
2099your compiler is good :), then the method will be fully inlined into the
2100thunking function, making it as fast as a direct C callback.
2101.Sp
2102Example: simple class declaration and watcher initialisation
2103.Sp
2104.Vb 4
2105\& struct myclass
2106\& {
2107\& void io_cb (ev::io &w, int revents) { }
2108\& }
2109.Ve
2110.Sp
2111.Vb 3
2112\& myclass obj;
2113\& ev::io iow;
2114\& iow.set <myclass, &myclass::io_cb> (&obj);
2115.Ve
2116.IP "w\->set<function> (void *data = 0)" 4
2117.IX Item "w->set<function> (void *data = 0)"
2118Also sets a callback, but uses a static method or plain function as
2119callback. The optional \f(CW\*(C`data\*(C'\fR argument will be stored in the watcher's
2120\&\f(CW\*(C`data\*(C'\fR member and is free for you to use.
2121.Sp
2122The prototype of the \f(CW\*(C`function\*(C'\fR must be \f(CW\*(C`void (*)(ev::TYPE &w, int)\*(C'\fR.
2123.Sp
2124See the method\-\f(CW\*(C`set\*(C'\fR above for more details.
2125.Sp
2126Example:
2127.Sp
2128.Vb 2
2129\& static void io_cb (ev::io &w, int revents) { }
2130\& iow.set <io_cb> ();
2131.Ve
2132.IP "w\->set (struct ev_loop *)" 4
2133.IX Item "w->set (struct ev_loop *)"
2134Associates a different \f(CW\*(C`struct ev_loop\*(C'\fR with this watcher. You can only
2135do this when the watcher is inactive (and not pending either).
2136.IP "w\->set ([args])" 4
2137.IX Item "w->set ([args])"
2138Basically the same as \f(CW\*(C`ev_TYPE_set\*(C'\fR, with the same args. Must be
2139called at least once. Unlike the C counterpart, an active watcher gets
2140automatically stopped and restarted when reconfiguring it with this
2141method.
2142.IP "w\->start ()" 4
2143.IX Item "w->start ()"
2144Starts the watcher. Note that there is no \f(CW\*(C`loop\*(C'\fR argument, as the
2145constructor already stores the event loop.
2146.IP "w\->stop ()" 4
2147.IX Item "w->stop ()"
2148Stops the watcher if it is active. Again, no \f(CW\*(C`loop\*(C'\fR argument.
2149.ie n .IP "w\->again () ""ev::timer""\fR, \f(CW""ev::periodic"" only" 4
2150.el .IP "w\->again () \f(CWev::timer\fR, \f(CWev::periodic\fR only" 4
2151.IX Item "w->again () ev::timer, ev::periodic only"
2152For \f(CW\*(C`ev::timer\*(C'\fR and \f(CW\*(C`ev::periodic\*(C'\fR, this invokes the corresponding
2153\&\f(CW\*(C`ev_TYPE_again\*(C'\fR function.
2154.ie n .IP "w\->sweep () ""ev::embed"" only" 4
2155.el .IP "w\->sweep () \f(CWev::embed\fR only" 4
2156.IX Item "w->sweep () ev::embed only"
2157Invokes \f(CW\*(C`ev_embed_sweep\*(C'\fR.
2158.ie n .IP "w\->update () ""ev::stat"" only" 4
2159.el .IP "w\->update () \f(CWev::stat\fR only" 4
2160.IX Item "w->update () ev::stat only"
2161Invokes \f(CW\*(C`ev_stat_stat\*(C'\fR.
2162.RE
2163.RS 4
2164.RE
2165.PP
2166Example: Define a class with an \s-1IO\s0 and idle watcher, start one of them in
2167the constructor.
2168.PP
2169.Vb 4
2170\& class myclass
2171\& {
2172\& ev_io io; void io_cb (ev::io &w, int revents);
2173\& ev_idle idle void idle_cb (ev::idle &w, int revents);
2174.Ve
2175.PP
2176.Vb 2
2177\& myclass ();
2178\& }
2179.Ve
2180.PP
2181.Vb 4
2182\& myclass::myclass (int fd)
2183\& {
2184\& io .set <myclass, &myclass::io_cb > (this);
2185\& idle.set <myclass, &myclass::idle_cb> (this);
2186.Ve
2187.PP
2188.Vb 2
2189\& io.start (fd, ev::READ);
2190\& }
2191.Ve
2192.SH "MACRO MAGIC"
2193.IX Header "MACRO MAGIC"
2194Libev can be compiled with a variety of options, the most fundemantal is
2195\&\f(CW\*(C`EV_MULTIPLICITY\*(C'\fR. This option determines whether (most) functions and
2196callbacks have an initial \f(CW\*(C`struct ev_loop *\*(C'\fR argument.
2197.PP
2198To make it easier to write programs that cope with either variant, the
2199following macros are defined:
2200.ie n .IP """EV_A""\fR, \f(CW""EV_A_""" 4
2201.el .IP "\f(CWEV_A\fR, \f(CWEV_A_\fR" 4
2202.IX Item "EV_A, EV_A_"
2203This provides the loop \fIargument\fR for functions, if one is required (\*(L"ev
2204loop argument\*(R"). The \f(CW\*(C`EV_A\*(C'\fR form is used when this is the sole argument,
2205\&\f(CW\*(C`EV_A_\*(C'\fR is used when other arguments are following. Example:
2206.Sp
2207.Vb 3
2208\& ev_unref (EV_A);
2209\& ev_timer_add (EV_A_ watcher);
2210\& ev_loop (EV_A_ 0);
2211.Ve
2212.Sp
2213It assumes the variable \f(CW\*(C`loop\*(C'\fR of type \f(CW\*(C`struct ev_loop *\*(C'\fR is in scope,
2214which is often provided by the following macro.
2215.ie n .IP """EV_P""\fR, \f(CW""EV_P_""" 4
2216.el .IP "\f(CWEV_P\fR, \f(CWEV_P_\fR" 4
2217.IX Item "EV_P, EV_P_"
2218This provides the loop \fIparameter\fR for functions, if one is required (\*(L"ev
2219loop parameter\*(R"). The \f(CW\*(C`EV_P\*(C'\fR form is used when this is the sole parameter,
2220\&\f(CW\*(C`EV_P_\*(C'\fR is used when other parameters are following. Example:
2221.Sp
2222.Vb 2
2223\& // this is how ev_unref is being declared
2224\& static void ev_unref (EV_P);
2225.Ve
2226.Sp
2227.Vb 2
2228\& // this is how you can declare your typical callback
2229\& static void cb (EV_P_ ev_timer *w, int revents)
2230.Ve
2231.Sp
2232It declares a parameter \f(CW\*(C`loop\*(C'\fR of type \f(CW\*(C`struct ev_loop *\*(C'\fR, quite
2233suitable for use with \f(CW\*(C`EV_A\*(C'\fR.
2234.ie n .IP """EV_DEFAULT""\fR, \f(CW""EV_DEFAULT_""" 4
2235.el .IP "\f(CWEV_DEFAULT\fR, \f(CWEV_DEFAULT_\fR" 4
2236.IX Item "EV_DEFAULT, EV_DEFAULT_"
2237Similar to the other two macros, this gives you the value of the default
2238loop, if multiple loops are supported (\*(L"ev loop default\*(R").
2239.PP
2240Example: Declare and initialise a check watcher, utilising the above
2241macros so it will work regardless of whether multiple loops are supported
2242or not.
2243.PP
2244.Vb 5
2245\& static void
2246\& check_cb (EV_P_ ev_timer *w, int revents)
2247\& {
2248\& ev_check_stop (EV_A_ w);
2249\& }
2250.Ve
2251.PP
2252.Vb 4
2253\& ev_check check;
2254\& ev_check_init (&check, check_cb);
2255\& ev_check_start (EV_DEFAULT_ &check);
2256\& ev_loop (EV_DEFAULT_ 0);
2257.Ve
2258.SH "EMBEDDING"
2259.IX Header "EMBEDDING"
2260Libev can (and often is) directly embedded into host
2261applications. Examples of applications that embed it include the Deliantra
2262Game Server, the \s-1EV\s0 perl module, the \s-1GNU\s0 Virtual Private Ethernet (gvpe)
2263and rxvt\-unicode.
2264.PP
2265The goal is to enable you to just copy the neecssary files into your
2266source directory without having to change even a single line in them, so
2267you can easily upgrade by simply copying (or having a checked-out copy of
2268libev somewhere in your source tree).
2269.Sh "\s-1FILESETS\s0"
2270.IX Subsection "FILESETS"
2271Depending on what features you need you need to include one or more sets of files
2272in your app.
2273.PP
2274\fI\s-1CORE\s0 \s-1EVENT\s0 \s-1LOOP\s0\fR
2275.IX Subsection "CORE EVENT LOOP"
2276.PP
2277To include only the libev core (all the \f(CW\*(C`ev_*\*(C'\fR functions), with manual
2278configuration (no autoconf):
2279.PP
2280.Vb 2
2281\& #define EV_STANDALONE 1
2282\& #include "ev.c"
2283.Ve
2284.PP
2285This will automatically include \fIev.h\fR, too, and should be done in a
2286single C source file only to provide the function implementations. To use
2287it, do the same for \fIev.h\fR in all files wishing to use this \s-1API\s0 (best
2288done by writing a wrapper around \fIev.h\fR that you can include instead and
2289where you can put other configuration options):
2290.PP
2291.Vb 2
2292\& #define EV_STANDALONE 1
2293\& #include "ev.h"
2294.Ve
2295.PP
2296Both header files and implementation files can be compiled with a \*(C+
2297compiler (at least, thats a stated goal, and breakage will be treated
2298as a bug).
2299.PP
2300You need the following files in your source tree, or in a directory
2301in your include path (e.g. in libev/ when using \-Ilibev):
2302.PP
2303.Vb 4
2304\& ev.h
2305\& ev.c
2306\& ev_vars.h
2307\& ev_wrap.h
2308.Ve
2309.PP
2310.Vb 1
2311\& ev_win32.c required on win32 platforms only
2312.Ve
2313.PP
2314.Vb 5
2315\& ev_select.c only when select backend is enabled (which is enabled by default)
2316\& ev_poll.c only when poll backend is enabled (disabled by default)
2317\& ev_epoll.c only when the epoll backend is enabled (disabled by default)
2318\& ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2319\& ev_port.c only when the solaris port backend is enabled (disabled by default)
2320.Ve
2321.PP
2322\&\fIev.c\fR includes the backend files directly when enabled, so you only need
2323to compile this single file.
2324.PP
2325\fI\s-1LIBEVENT\s0 \s-1COMPATIBILITY\s0 \s-1API\s0\fR
2326.IX Subsection "LIBEVENT COMPATIBILITY API"
2327.PP
2328To include the libevent compatibility \s-1API\s0, also include:
2329.PP
2330.Vb 1
2331\& #include "event.c"
2332.Ve
2333.PP
2334in the file including \fIev.c\fR, and:
2335.PP
2336.Vb 1
2337\& #include "event.h"
2338.Ve
2339.PP
2340in the files that want to use the libevent \s-1API\s0. This also includes \fIev.h\fR.
2341.PP
2342You need the following additional files for this:
2343.PP
2344.Vb 2
2345\& event.h
2346\& event.c
2347.Ve
2348.PP
2349\fI\s-1AUTOCONF\s0 \s-1SUPPORT\s0\fR
2350.IX Subsection "AUTOCONF SUPPORT"
2351.PP
2352Instead of using \f(CW\*(C`EV_STANDALONE=1\*(C'\fR and providing your config in
2353whatever way you want, you can also \f(CW\*(C`m4_include([libev.m4])\*(C'\fR in your
2354\&\fIconfigure.ac\fR and leave \f(CW\*(C`EV_STANDALONE\*(C'\fR undefined. \fIev.c\fR will then
2355include \fIconfig.h\fR and configure itself accordingly.
2356.PP
2357For this of course you need the m4 file:
2358.PP
2359.Vb 1
2360\& libev.m4
2361.Ve
2362.Sh "\s-1PREPROCESSOR\s0 \s-1SYMBOLS/MACROS\s0"
2363.IX Subsection "PREPROCESSOR SYMBOLS/MACROS"
2364Libev can be configured via a variety of preprocessor symbols you have to define
2365before including any of its files. The default is not to build for multiplicity
2366and only include the select backend.
2367.IP "\s-1EV_STANDALONE\s0" 4
2368.IX Item "EV_STANDALONE"
2369Must always be \f(CW1\fR if you do not use autoconf configuration, which
2370keeps libev from including \fIconfig.h\fR, and it also defines dummy
2371implementations for some libevent functions (such as logging, which is not
2372supported). It will also not define any of the structs usually found in
2373\&\fIevent.h\fR that are not directly supported by the libev core alone.
2374.IP "\s-1EV_USE_MONOTONIC\s0" 4
2375.IX Item "EV_USE_MONOTONIC"
2376If defined to be \f(CW1\fR, libev will try to detect the availability of the
2377monotonic clock option at both compiletime and runtime. Otherwise no use
2378of the monotonic clock option will be attempted. If you enable this, you
2379usually have to link against librt or something similar. Enabling it when
2380the functionality isn't available is safe, though, althoguh you have
2381to make sure you link against any libraries where the \f(CW\*(C`clock_gettime\*(C'\fR
2382function is hiding in (often \fI\-lrt\fR).
2383.IP "\s-1EV_USE_REALTIME\s0" 4
2384.IX Item "EV_USE_REALTIME"
2385If defined to be \f(CW1\fR, libev will try to detect the availability of the
2386realtime clock option at compiletime (and assume its availability at
2387runtime if successful). Otherwise no use of the realtime clock option will
2388be attempted. This effectively replaces \f(CW\*(C`gettimeofday\*(C'\fR by \f(CW\*(C`clock_get
2389(CLOCK_REALTIME, ...)\*(C'\fR and will not normally affect correctness. See tzhe note about libraries
2390in the description of \f(CW\*(C`EV_USE_MONOTONIC\*(C'\fR, though.
2391.IP "\s-1EV_USE_SELECT\s0" 4
2392.IX Item "EV_USE_SELECT"
2393If undefined or defined to be \f(CW1\fR, libev will compile in support for the
2394\&\f(CW\*(C`select\*(C'\fR(2) backend. No attempt at autodetection will be done: if no
2395other method takes over, select will be it. Otherwise the select backend
2396will not be compiled in.
2397.IP "\s-1EV_SELECT_USE_FD_SET\s0" 4
2398.IX Item "EV_SELECT_USE_FD_SET"
2399If defined to \f(CW1\fR, then the select backend will use the system \f(CW\*(C`fd_set\*(C'\fR
2400structure. This is useful if libev doesn't compile due to a missing
2401\&\f(CW\*(C`NFDBITS\*(C'\fR or \f(CW\*(C`fd_mask\*(C'\fR definition or it misguesses the bitset layout on
2402exotic systems. This usually limits the range of file descriptors to some
2403low limit such as 1024 or might have other limitations (winsocket only
2404allows 64 sockets). The \f(CW\*(C`FD_SETSIZE\*(C'\fR macro, set before compilation, might
2405influence the size of the \f(CW\*(C`fd_set\*(C'\fR used.
2406.IP "\s-1EV_SELECT_IS_WINSOCKET\s0" 4
2407.IX Item "EV_SELECT_IS_WINSOCKET"
2408When defined to \f(CW1\fR, the select backend will assume that
2409select/socket/connect etc. don't understand file descriptors but
2410wants osf handles on win32 (this is the case when the select to
2411be used is the winsock select). This means that it will call
2412\&\f(CW\*(C`_get_osfhandle\*(C'\fR on the fd to convert it to an \s-1OS\s0 handle. Otherwise,
2413it is assumed that all these functions actually work on fds, even
2414on win32. Should not be defined on non\-win32 platforms.
2415.IP "\s-1EV_USE_POLL\s0" 4
2416.IX Item "EV_USE_POLL"
2417If defined to be \f(CW1\fR, libev will compile in support for the \f(CW\*(C`poll\*(C'\fR(2)
2418backend. Otherwise it will be enabled on non\-win32 platforms. It
2419takes precedence over select.
2420.IP "\s-1EV_USE_EPOLL\s0" 4
2421.IX Item "EV_USE_EPOLL"
2422If defined to be \f(CW1\fR, libev will compile in support for the Linux
2423\&\f(CW\*(C`epoll\*(C'\fR(7) backend. Its availability will be detected at runtime,
2424otherwise another method will be used as fallback. This is the
2425preferred backend for GNU/Linux systems.
2426.IP "\s-1EV_USE_KQUEUE\s0" 4
2427.IX Item "EV_USE_KQUEUE"
2428If defined to be \f(CW1\fR, libev will compile in support for the \s-1BSD\s0 style
2429\&\f(CW\*(C`kqueue\*(C'\fR(2) backend. Its actual availability will be detected at runtime,
2430otherwise another method will be used as fallback. This is the preferred
2431backend for \s-1BSD\s0 and BSD-like systems, although on most BSDs kqueue only
2432supports some types of fds correctly (the only platform we found that
2433supports ptys for example was NetBSD), so kqueue might be compiled in, but
2434not be used unless explicitly requested. The best way to use it is to find
2435out whether kqueue supports your type of fd properly and use an embedded
2436kqueue loop.
2437.IP "\s-1EV_USE_PORT\s0" 4
2438.IX Item "EV_USE_PORT"
2439If defined to be \f(CW1\fR, libev will compile in support for the Solaris
244010 port style backend. Its availability will be detected at runtime,
2441otherwise another method will be used as fallback. This is the preferred
2442backend for Solaris 10 systems.
2443.IP "\s-1EV_USE_DEVPOLL\s0" 4
2444.IX Item "EV_USE_DEVPOLL"
2445reserved for future expansion, works like the \s-1USE\s0 symbols above.
2446.IP "\s-1EV_USE_INOTIFY\s0" 4
2447.IX Item "EV_USE_INOTIFY"
2448If defined to be \f(CW1\fR, libev will compile in support for the Linux inotify
2449interface to speed up \f(CW\*(C`ev_stat\*(C'\fR watchers. Its actual availability will
2450be detected at runtime.
2451.IP "\s-1EV_H\s0" 4
2452.IX Item "EV_H"
2453The name of the \fIev.h\fR header file used to include it. The default if
2454undefined is \f(CW\*(C`<ev.h>\*(C'\fR in \fIevent.h\fR and \f(CW"ev.h"\fR in \fIev.c\fR. This
2455can be used to virtually rename the \fIev.h\fR header file in case of conflicts.
2456.IP "\s-1EV_CONFIG_H\s0" 4
2457.IX Item "EV_CONFIG_H"
2458If \f(CW\*(C`EV_STANDALONE\*(C'\fR isn't \f(CW1\fR, this variable can be used to override
2459\&\fIev.c\fR's idea of where to find the \fIconfig.h\fR file, similarly to
2460\&\f(CW\*(C`EV_H\*(C'\fR, above.
2461.IP "\s-1EV_EVENT_H\s0" 4
2462.IX Item "EV_EVENT_H"
2463Similarly to \f(CW\*(C`EV_H\*(C'\fR, this macro can be used to override \fIevent.c\fR's idea
2464of how the \fIevent.h\fR header can be found.
2465.IP "\s-1EV_PROTOTYPES\s0" 4
2466.IX Item "EV_PROTOTYPES"
2467If defined to be \f(CW0\fR, then \fIev.h\fR will not define any function
2468prototypes, but still define all the structs and other symbols. This is
2469occasionally useful if you want to provide your own wrapper functions
2470around libev functions.
2471.IP "\s-1EV_MULTIPLICITY\s0" 4
2472.IX Item "EV_MULTIPLICITY"
2473If undefined or defined to \f(CW1\fR, then all event-loop-specific functions
2474will have the \f(CW\*(C`struct ev_loop *\*(C'\fR as first argument, and you can create
2475additional independent event loops. Otherwise there will be no support
2476for multiple event loops and there is no first event loop pointer
2477argument. Instead, all functions act on the single default loop.
2478.IP "\s-1EV_MINPRI\s0" 4
2479.IX Item "EV_MINPRI"
2480.PD 0
2481.IP "\s-1EV_MAXPRI\s0" 4
2482.IX Item "EV_MAXPRI"
2483.PD
2484The range of allowed priorities. \f(CW\*(C`EV_MINPRI\*(C'\fR must be smaller or equal to
2485\&\f(CW\*(C`EV_MAXPRI\*(C'\fR, but otherwise there are no non-obvious limitations. You can
2486provide for more priorities by overriding those symbols (usually defined
2487to be \f(CW\*(C`\-2\*(C'\fR and \f(CW2\fR, respectively).
2488.Sp
2489When doing priority-based operations, libev usually has to linearly search
2490all the priorities, so having many of them (hundreds) uses a lot of space
2491and time, so using the defaults of five priorities (\-2 .. +2) is usually
2492fine.
2493.Sp
2494If your embedding app does not need any priorities, defining these both to
2495\&\f(CW0\fR will save some memory and cpu.
2496.IP "\s-1EV_PERIODIC_ENABLE\s0" 4
2497.IX Item "EV_PERIODIC_ENABLE"
2498If undefined or defined to be \f(CW1\fR, then periodic timers are supported. If
2499defined to be \f(CW0\fR, then they are not. Disabling them saves a few kB of
2500code.
2501.IP "\s-1EV_IDLE_ENABLE\s0" 4
2502.IX Item "EV_IDLE_ENABLE"
2503If undefined or defined to be \f(CW1\fR, then idle watchers are supported. If
2504defined to be \f(CW0\fR, then they are not. Disabling them saves a few kB of
2505code.
2506.IP "\s-1EV_EMBED_ENABLE\s0" 4
2507.IX Item "EV_EMBED_ENABLE"
2508If undefined or defined to be \f(CW1\fR, then embed watchers are supported. If
2509defined to be \f(CW0\fR, then they are not.
2510.IP "\s-1EV_STAT_ENABLE\s0" 4
2511.IX Item "EV_STAT_ENABLE"
2512If undefined or defined to be \f(CW1\fR, then stat watchers are supported. If
2513defined to be \f(CW0\fR, then they are not.
2514.IP "\s-1EV_FORK_ENABLE\s0" 4
2515.IX Item "EV_FORK_ENABLE"
2516If undefined or defined to be \f(CW1\fR, then fork watchers are supported. If
2517defined to be \f(CW0\fR, then they are not.
2518.IP "\s-1EV_MINIMAL\s0" 4
2519.IX Item "EV_MINIMAL"
2520If you need to shave off some kilobytes of code at the expense of some
2521speed, define this symbol to \f(CW1\fR. Currently only used for gcc to override
2522some inlining decisions, saves roughly 30% codesize of amd64.
2523.IP "\s-1EV_PID_HASHSIZE\s0" 4
2524.IX Item "EV_PID_HASHSIZE"
2525\&\f(CW\*(C`ev_child\*(C'\fR watchers use a small hash table to distribute workload by
2526pid. The default size is \f(CW16\fR (or \f(CW1\fR with \f(CW\*(C`EV_MINIMAL\*(C'\fR), usually more
2527than enough. If you need to manage thousands of children you might want to
2528increase this value (\fImust\fR be a power of two).
2529.IP "\s-1EV_INOTIFY_HASHSIZE\s0" 4
2530.IX Item "EV_INOTIFY_HASHSIZE"
2531\&\f(CW\*(C`ev_staz\*(C'\fR watchers use a small hash table to distribute workload by
2532inotify watch id. The default size is \f(CW16\fR (or \f(CW1\fR with \f(CW\*(C`EV_MINIMAL\*(C'\fR),
2533usually more than enough. If you need to manage thousands of \f(CW\*(C`ev_stat\*(C'\fR
2534watchers you might want to increase this value (\fImust\fR be a power of
2535two).
2536.IP "\s-1EV_COMMON\s0" 4
2537.IX Item "EV_COMMON"
2538By default, all watchers have a \f(CW\*(C`void *data\*(C'\fR member. By redefining
2539this macro to a something else you can include more and other types of
2540members. You have to define it each time you include one of the files,
2541though, and it must be identical each time.
2542.Sp
2543For example, the perl \s-1EV\s0 module uses something like this:
2544.Sp
2545.Vb 3
2546\& #define EV_COMMON \e
2547\& SV *self; /* contains this struct */ \e
2548\& SV *cb_sv, *fh /* note no trailing ";" */
2549.Ve
2550.IP "\s-1EV_CB_DECLARE\s0 (type)" 4
2551.IX Item "EV_CB_DECLARE (type)"
2552.PD 0
2553.IP "\s-1EV_CB_INVOKE\s0 (watcher, revents)" 4
2554.IX Item "EV_CB_INVOKE (watcher, revents)"
2555.IP "ev_set_cb (ev, cb)" 4
2556.IX Item "ev_set_cb (ev, cb)"
2557.PD
2558Can be used to change the callback member declaration in each watcher,
2559and the way callbacks are invoked and set. Must expand to a struct member
2560definition and a statement, respectively. See the \fIev.v\fR header file for
2561their default definitions. One possible use for overriding these is to
2562avoid the \f(CW\*(C`struct ev_loop *\*(C'\fR as first argument in all cases, or to use
2563method calls instead of plain function calls in \*(C+.
2564.Sh "\s-1EXAMPLES\s0"
2565.IX Subsection "EXAMPLES"
2566For a real-world example of a program the includes libev
2567verbatim, you can have a look at the \s-1EV\s0 perl module
2568(<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
2569the \fIlibev/\fR subdirectory and includes them in the \fI\s-1EV/EVAPI\s0.h\fR (public
2570interface) and \fI\s-1EV\s0.xs\fR (implementation) files. Only the \fI\s-1EV\s0.xs\fR file
2571will be compiled. It is pretty complex because it provides its own header
2572file.
2573.Sp
2574The usage in rxvt-unicode is simpler. It has a \fIev_cpp.h\fR header file
2575that everybody includes and which overrides some configure choices:
2576.Sp
2577.Vb 9
2578\& #define EV_MINIMAL 1
2579\& #define EV_USE_POLL 0
2580\& #define EV_MULTIPLICITY 0
2581\& #define EV_PERIODIC_ENABLE 0
2582\& #define EV_STAT_ENABLE 0
2583\& #define EV_FORK_ENABLE 0
2584\& #define EV_CONFIG_H <config.h>
2585\& #define EV_MINPRI 0
2586\& #define EV_MAXPRI 0
2587.Ve
2588.Sp
2589.Vb 1
2590\& #include "ev++.h"
2591.Ve
2592.Sp
2593And a \fIev_cpp.C\fR implementation file that contains libev proper and is compiled:
2594.Sp
2595.Vb 2
2596\& #include "ev_cpp.h"
2597\& #include "ev.c"
2598.Ve
2599.SH "COMPLEXITIES"
2600.IX Header "COMPLEXITIES"
2601In this section the complexities of (many of) the algorithms used inside
2602libev will be explained. For complexity discussions about backends see the
2603documentation for \f(CW\*(C`ev_default_init\*(C'\fR.
2604.Sp
2605All of the following are about amortised time: If an array needs to be
2606extended, libev needs to realloc and move the whole array, but this
2607happens asymptotically never with higher number of elements, so O(1) might
2608mean it might do a lengthy realloc operation in rare cases, but on average
2609it is much faster and asymptotically approaches constant time.
2610.RS 4
2611.IP "Starting and stopping timer/periodic watchers: O(log skipped_other_timers)" 4
2612.IX Item "Starting and stopping timer/periodic watchers: O(log skipped_other_timers)"
2613This means that, when you have a watcher that triggers in one hour and
2614there are 100 watchers that would trigger before that then inserting will
2615have to skip those 100 watchers.
2616.IP "Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)" 4
2617.IX Item "Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)"
2618That means that for changing a timer costs less than removing/adding them
2619as only the relative motion in the event queue has to be paid for.
2620.IP "Starting io/check/prepare/idle/signal/child watchers: O(1)" 4
2621.IX Item "Starting io/check/prepare/idle/signal/child watchers: O(1)"
2622These just add the watcher into an array or at the head of a list.
2623=item Stopping check/prepare/idle watchers: O(1)
2624.IP "Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % \s-1EV_PID_HASHSIZE\s0))" 4
2625.IX Item "Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))"
2626These watchers are stored in lists then need to be walked to find the
2627correct watcher to remove. The lists are usually short (you don't usually
2628have many watchers waiting for the same fd or signal).
2629.IP "Finding the next timer per loop iteration: O(1)" 4
2630.IX Item "Finding the next timer per loop iteration: O(1)"
2631.PD 0
2632.IP "Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)" 4
2633.IX Item "Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)"
2634.PD
2635A change means an I/O watcher gets started or stopped, which requires
2636libev to recalculate its status (and possibly tell the kernel).
2637.IP "Activating one watcher: O(1)" 4
2638.IX Item "Activating one watcher: O(1)"
2639.PD 0
2640.IP "Priority handling: O(number_of_priorities)" 4
2641.IX Item "Priority handling: O(number_of_priorities)"
2642.PD
2643Priorities are implemented by allocating some space for each
2644priority. When doing priority-based operations, libev usually has to
2645linearly search all the priorities.
2646.RE
2647.RS 4
919.SH "AUTHOR" 2648.SH "AUTHOR"
920.IX Header "AUTHOR" 2649.IX Header "AUTHOR"
921Marc Lehmann <libev@schmorp.de>. 2650Marc Lehmann <libev@schmorp.de>.

Diff Legend

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