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

Comparing libev/ev.3 (file contents):
Revision 1.6 by root, Fri Nov 23 05:14:58 2007 UTC vs.
Revision 1.57 by root, Sat Dec 22 11:49:17 2007 UTC

… …
126. ds Ae AE 126. ds Ae AE
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 "EV 1"
132.TH "<STANDARD INPUT>" 1 "2007-11-23" "perl v5.8.8" "User Contributed Perl Documentation" 132.TH EV 1 "2007-12-22" "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 occurring), 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
147(or thread) by executing the \fIevent loop\fR handler, and will then 210(or thread) by executing the \fIevent loop\fR handler, and will then
148communicate events via a callback mechanism. 211communicate events via a callback mechanism.
… …
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 some floatingpoint value. Unlike the name
249component \f(CW\*(C`stamp\*(C'\fR might indicate, it is also used for time differences
250throughout libev.
179.SH "GLOBAL FUNCTIONS" 251.SH "GLOBAL FUNCTIONS"
180.IX Header "GLOBAL FUNCTIONS" 252.IX Header "GLOBAL FUNCTIONS"
181These functions can be called anytime, even before initialising the 253These functions can be called anytime, even before initialising the
182library in any way. 254library in any way.
183.IP "ev_tstamp ev_time ()" 4 255.IP "ev_tstamp ev_time ()" 4
184.IX Item "ev_tstamp ev_time ()" 256.IX Item "ev_tstamp ev_time ()"
185Returns the current time as libev would use it. Please note that the 257Returns the current time as libev would use it. Please note that the
186\&\f(CW\*(C`ev_now\*(C'\fR function is usually faster and also often returns the timestamp 258\&\f(CW\*(C`ev_now\*(C'\fR function is usually faster and also often returns the timestamp
187you actually want to know. 259you actually want to know.
260.IP "ev_sleep (ev_tstamp interval)" 4
261.IX Item "ev_sleep (ev_tstamp interval)"
262Sleep for the given interval: The current thread will be blocked until
263either it is interrupted or the given time interval has passed. Basically
264this is a subsecond-resolution \f(CW\*(C`sleep ()\*(C'\fR.
188.IP "int ev_version_major ()" 4 265.IP "int ev_version_major ()" 4
189.IX Item "int ev_version_major ()" 266.IX Item "int ev_version_major ()"
190.PD 0 267.PD 0
191.IP "int ev_version_minor ()" 4 268.IP "int ev_version_minor ()" 4
192.IX Item "int ev_version_minor ()" 269.IX Item "int ev_version_minor ()"
193.PD 270.PD
194You can find out the major and minor version numbers of the library 271You can find out the major and minor \s-1ABI\s0 version numbers of the library
195you linked against by calling the functions \f(CW\*(C`ev_version_major\*(C'\fR and 272you linked against by calling the functions \f(CW\*(C`ev_version_major\*(C'\fR and
196\&\f(CW\*(C`ev_version_minor\*(C'\fR. If you want, you can compare against the global 273\&\f(CW\*(C`ev_version_minor\*(C'\fR. If you want, you can compare against the global
197symbols \f(CW\*(C`EV_VERSION_MAJOR\*(C'\fR and \f(CW\*(C`EV_VERSION_MINOR\*(C'\fR, which specify the 274symbols \f(CW\*(C`EV_VERSION_MAJOR\*(C'\fR and \f(CW\*(C`EV_VERSION_MINOR\*(C'\fR, which specify the
198version of the library your program was compiled against. 275version of the library your program was compiled against.
199.Sp 276.Sp
277These version numbers refer to the \s-1ABI\s0 version of the library, not the
278release version.
279.Sp
200Usually, it's a good idea to terminate if the major versions mismatch, 280Usually, it's a good idea to terminate if the major versions mismatch,
201as this indicates an incompatible change. Minor versions are usually 281as this indicates an incompatible change. Minor versions are usually
202compatible to older versions, so a larger minor version alone is usually 282compatible to older versions, so a larger minor version alone is usually
203not a problem. 283not a problem.
284.Sp
285Example: Make sure we haven't accidentally been linked against the wrong
286version.
287.Sp
288.Vb 3
289\& assert (("libev version mismatch",
290\& ev_version_major () == EV_VERSION_MAJOR
291\& && ev_version_minor () >= EV_VERSION_MINOR));
292.Ve
204.IP "unsigned int ev_supported_backends ()" 4 293.IP "unsigned int ev_supported_backends ()" 4
205.IX Item "unsigned int ev_supported_backends ()" 294.IX Item "unsigned int ev_supported_backends ()"
206Return the set of all backends (i.e. their corresponding \f(CW\*(C`EV_BACKEND_*\*(C'\fR 295Return the set of all backends (i.e. their corresponding \f(CW\*(C`EV_BACKEND_*\*(C'\fR
207value) compiled into this binary of libev (independent of their 296value) compiled into this binary of libev (independent of their
208availability on the system you are running on). See \f(CW\*(C`ev_default_loop\*(C'\fR for 297availability on the system you are running on). See \f(CW\*(C`ev_default_loop\*(C'\fR for
209a description of the set values. 298a description of the set values.
299.Sp
300Example: make sure we have the epoll method, because yeah this is cool and
301a must have and can we have a torrent of it please!!!11
302.Sp
303.Vb 2
304\& assert (("sorry, no epoll, no sex",
305\& ev_supported_backends () & EVBACKEND_EPOLL));
306.Ve
210.IP "unsigned int ev_recommended_backends ()" 4 307.IP "unsigned int ev_recommended_backends ()" 4
211.IX Item "unsigned int ev_recommended_backends ()" 308.IX Item "unsigned int ev_recommended_backends ()"
212Return the set of all backends compiled into this binary of libev and also 309Return the set of all backends compiled into this binary of libev and also
213recommended for this platform. This set is often smaller than the one 310recommended for this platform. This set is often smaller than the one
214returned by \f(CW\*(C`ev_supported_backends\*(C'\fR, as for example kqueue is broken on 311returned by \f(CW\*(C`ev_supported_backends\*(C'\fR, as for example kqueue is broken on
215most BSDs and will not be autodetected unless you explicitly request it 312most BSDs and will not be autodetected unless you explicitly request it
216(assuming you know what you are doing). This is the set of backends that 313(assuming you know what you are doing). This is the set of backends that
217\&\f(CW\*(C`EVFLAG_AUTO\*(C'\fR will probe for. 314libev will probe for if you specify no backends explicitly.
315.IP "unsigned int ev_embeddable_backends ()" 4
316.IX Item "unsigned int ev_embeddable_backends ()"
317Returns the set of backends that are embeddable in other event loops. This
318is the theoretical, all\-platform, value. To find which backends
319might be supported on the current system, you would need to look at
320\&\f(CW\*(C`ev_embeddable_backends () & ev_supported_backends ()\*(C'\fR, likewise for
321recommended ones.
322.Sp
323See the description of \f(CW\*(C`ev_embed\*(C'\fR watchers for more info.
218.IP "ev_set_allocator (void *(*cb)(void *ptr, long size))" 4 324.IP "ev_set_allocator (void *(*cb)(void *ptr, long size))" 4
219.IX Item "ev_set_allocator (void *(*cb)(void *ptr, long size))" 325.IX Item "ev_set_allocator (void *(*cb)(void *ptr, long size))"
220Sets the allocation function to use (the prototype is similar to the 326Sets the allocation function to use (the prototype is similar \- the
221realloc C function, the semantics are identical). It is used to allocate 327semantics is identical \- to the realloc C function). It is used to
222and free memory (no surprises here). If it returns zero when memory 328allocate and free memory (no surprises here). If it returns zero when
223needs to be allocated, the library might abort or take some potentially 329memory needs to be allocated, the library might abort or take some
224destructive action. The default is your system realloc function. 330potentially destructive action. The default is your system realloc
331function.
225.Sp 332.Sp
226You could override this function in high-availability programs to, say, 333You could override this function in high-availability programs to, say,
227free some memory if it cannot allocate memory, to use a special allocator, 334free some memory if it cannot allocate memory, to use a special allocator,
228or even to sleep a while and retry until some memory is available. 335or even to sleep a while and retry until some memory is available.
336.Sp
337Example: Replace the libev allocator with one that waits a bit and then
338retries).
339.Sp
340.Vb 6
341\& static void *
342\& persistent_realloc (void *ptr, size_t size)
343\& {
344\& for (;;)
345\& {
346\& void *newptr = realloc (ptr, size);
347.Ve
348.Sp
349.Vb 2
350\& if (newptr)
351\& return newptr;
352.Ve
353.Sp
354.Vb 3
355\& sleep (60);
356\& }
357\& }
358.Ve
359.Sp
360.Vb 2
361\& ...
362\& ev_set_allocator (persistent_realloc);
363.Ve
229.IP "ev_set_syserr_cb (void (*cb)(const char *msg));" 4 364.IP "ev_set_syserr_cb (void (*cb)(const char *msg));" 4
230.IX Item "ev_set_syserr_cb (void (*cb)(const char *msg));" 365.IX Item "ev_set_syserr_cb (void (*cb)(const char *msg));"
231Set the callback function to call on a retryable syscall error (such 366Set the callback function to call on a retryable syscall error (such
232as failed select, poll, epoll_wait). The message is a printable string 367as failed select, poll, epoll_wait). The message is a printable string
233indicating the system call or subsystem causing the problem. If this 368indicating the system call or subsystem causing the problem. If this
234callback is set, then libev will expect it to remedy the sitution, no 369callback is set, then libev will expect it to remedy the sitution, no
235matter what, when it returns. That is, libev will generally retry the 370matter what, when it returns. That is, libev will generally retry the
236requested operation, or, if the condition doesn't go away, do bad stuff 371requested operation, or, if the condition doesn't go away, do bad stuff
237(such as abort). 372(such as abort).
373.Sp
374Example: This is basically the same thing that libev does internally, too.
375.Sp
376.Vb 6
377\& static void
378\& fatal_error (const char *msg)
379\& {
380\& perror (msg);
381\& abort ();
382\& }
383.Ve
384.Sp
385.Vb 2
386\& ...
387\& ev_set_syserr_cb (fatal_error);
388.Ve
238.SH "FUNCTIONS CONTROLLING THE EVENT LOOP" 389.SH "FUNCTIONS CONTROLLING THE EVENT LOOP"
239.IX Header "FUNCTIONS CONTROLLING THE EVENT LOOP" 390.IX Header "FUNCTIONS CONTROLLING THE EVENT LOOP"
240An event loop is described by a \f(CW\*(C`struct ev_loop *\*(C'\fR. The library knows two 391An event loop is described by a \f(CW\*(C`struct ev_loop *\*(C'\fR. The library knows two
241types of such loops, the \fIdefault\fR loop, which supports signals and child 392types of such loops, the \fIdefault\fR loop, which supports signals and child
242events, and dynamically created loops which do not. 393events, and dynamically created loops which do not.
… …
256.Sp 407.Sp
257If you don't know what event loop to use, use the one returned from this 408If you don't know what event loop to use, use the one returned from this
258function. 409function.
259.Sp 410.Sp
260The flags argument can be used to specify special behaviour or specific 411The flags argument can be used to specify special behaviour or specific
261backends to use, and is usually specified as \f(CW0\fR (or \s-1EVFLAG_AUTO\s0). 412backends to use, and is usually specified as \f(CW0\fR (or \f(CW\*(C`EVFLAG_AUTO\*(C'\fR).
262.Sp 413.Sp
263It supports the following flags: 414The following flags are supported:
264.RS 4 415.RS 4
265.ie n .IP """EVFLAG_AUTO""" 4 416.ie n .IP """EVFLAG_AUTO""" 4
266.el .IP "\f(CWEVFLAG_AUTO\fR" 4 417.el .IP "\f(CWEVFLAG_AUTO\fR" 4
267.IX Item "EVFLAG_AUTO" 418.IX Item "EVFLAG_AUTO"
268The default flags value. Use this if you have no clue (it's the right 419The default flags value. Use this if you have no clue (it's the right
… …
274or setgid) then libev will \fInot\fR look at the environment variable 425or setgid) then libev will \fInot\fR look at the environment variable
275\&\f(CW\*(C`LIBEV_FLAGS\*(C'\fR. Otherwise (the default), this environment variable will 426\&\f(CW\*(C`LIBEV_FLAGS\*(C'\fR. Otherwise (the default), this environment variable will
276override the flags completely if it is found in the environment. This is 427override the flags completely if it is found in the environment. This is
277useful to try out specific backends to test their performance, or to work 428useful to try out specific backends to test their performance, or to work
278around bugs. 429around bugs.
430.ie n .IP """EVFLAG_FORKCHECK""" 4
431.el .IP "\f(CWEVFLAG_FORKCHECK\fR" 4
432.IX Item "EVFLAG_FORKCHECK"
433Instead of calling \f(CW\*(C`ev_default_fork\*(C'\fR or \f(CW\*(C`ev_loop_fork\*(C'\fR manually after
434a fork, you can also make libev check for a fork in each iteration by
435enabling this flag.
436.Sp
437This works by calling \f(CW\*(C`getpid ()\*(C'\fR on every iteration of the loop,
438and thus this might slow down your event loop if you do a lot of loop
439iterations and little real work, but is usually not noticeable (on my
440Linux system for example, \f(CW\*(C`getpid\*(C'\fR is actually a simple 5\-insn sequence
441without a syscall and thus \fIvery\fR fast, but my Linux system also has
442\&\f(CW\*(C`pthread_atfork\*(C'\fR which is even faster).
443.Sp
444The big advantage of this flag is that you can forget about fork (and
445forget about forgetting to tell libev about forking) when you use this
446flag.
447.Sp
448This flag setting cannot be overriden or specified in the \f(CW\*(C`LIBEV_FLAGS\*(C'\fR
449environment variable.
279.ie n .IP """EVBACKEND_SELECT"" (value 1, portable select backend)" 4 450.ie n .IP """EVBACKEND_SELECT"" (value 1, portable select backend)" 4
280.el .IP "\f(CWEVBACKEND_SELECT\fR (value 1, portable select backend)" 4 451.el .IP "\f(CWEVBACKEND_SELECT\fR (value 1, portable select backend)" 4
281.IX Item "EVBACKEND_SELECT (value 1, portable select backend)" 452.IX Item "EVBACKEND_SELECT (value 1, portable select backend)"
282This is your standard \fIselect\fR\|(2) backend. Not \fIcompletely\fR standard, as 453This is your standard \fIselect\fR\|(2) backend. Not \fIcompletely\fR standard, as
283libev tries to roll its own fd_set with no limits on the number of fds, 454libev tries to roll its own fd_set with no limits on the number of fds,
… …
293lot of inactive fds). It scales similarly to select, i.e. O(total_fds). 464lot of inactive fds). It scales similarly to select, i.e. O(total_fds).
294.ie n .IP """EVBACKEND_EPOLL"" (value 4, Linux)" 4 465.ie n .IP """EVBACKEND_EPOLL"" (value 4, Linux)" 4
295.el .IP "\f(CWEVBACKEND_EPOLL\fR (value 4, Linux)" 4 466.el .IP "\f(CWEVBACKEND_EPOLL\fR (value 4, Linux)" 4
296.IX Item "EVBACKEND_EPOLL (value 4, Linux)" 467.IX Item "EVBACKEND_EPOLL (value 4, Linux)"
297For few fds, this backend is a bit little slower than poll and select, 468For few fds, this backend is a bit little slower than poll and select,
298but it scales phenomenally better. While poll and select usually scale like 469but it scales phenomenally better. While poll and select usually scale
299O(total_fds) where n is the total number of fds (or the highest fd), epoll scales 470like O(total_fds) where n is the total number of fds (or the highest fd),
300either O(1) or O(active_fds). 471epoll scales either O(1) or O(active_fds). The epoll design has a number
472of shortcomings, such as silently dropping events in some hard-to-detect
473cases and rewiring a syscall per fd change, no fork support and bad
474support for dup:
301.Sp 475.Sp
302While stopping and starting an I/O watcher in the same iteration will 476While stopping, setting and starting an I/O watcher in the same iteration
303result in some caching, there is still a syscall per such incident 477will result in some caching, there is still a syscall per such incident
304(because the fd could point to a different file description now), so its 478(because the fd could point to a different file description now), so its
305best to avoid that. Also, \fIdup()\fRed file descriptors might not work very 479best to avoid that. Also, \f(CW\*(C`dup ()\*(C'\fR'ed file descriptors might not work
306well if you register events for both fds. 480very well if you register events for both fds.
481.Sp
482Please note that epoll sometimes generates spurious notifications, so you
483need to use non-blocking I/O or other means to avoid blocking when no data
484(or space) is available.
307.ie n .IP """EVBACKEND_KQUEUE"" (value 8, most \s-1BSD\s0 clones)" 4 485.ie n .IP """EVBACKEND_KQUEUE"" (value 8, most \s-1BSD\s0 clones)" 4
308.el .IP "\f(CWEVBACKEND_KQUEUE\fR (value 8, most \s-1BSD\s0 clones)" 4 486.el .IP "\f(CWEVBACKEND_KQUEUE\fR (value 8, most \s-1BSD\s0 clones)" 4
309.IX Item "EVBACKEND_KQUEUE (value 8, most BSD clones)" 487.IX Item "EVBACKEND_KQUEUE (value 8, most BSD clones)"
310Kqueue deserves special mention, as at the time of this writing, it 488Kqueue deserves special mention, as at the time of this writing, it
311was broken on all BSDs except NetBSD (usually it doesn't work with 489was broken on all BSDs except NetBSD (usually it doesn't work reliably
312anything but sockets and pipes, except on Darwin, where of course its 490with anything but sockets and pipes, except on Darwin, where of course
313completely useless). For this reason its not being \*(L"autodetected\*(R" unless 491it's completely useless). For this reason it's not being \*(L"autodetected\*(R"
314you explicitly specify the flags (i.e. you don't use \s-1EVFLAG_AUTO\s0). 492unless you explicitly specify it explicitly in the flags (i.e. using
493\&\f(CW\*(C`EVBACKEND_KQUEUE\*(C'\fR) or libev was compiled on a known-to-be-good (\-enough)
494system like NetBSD.
495.Sp
496You still can embed kqueue into a normal poll or select backend and use it
497only for sockets (after having made sure that sockets work with kqueue on
498the target platform). See \f(CW\*(C`ev_embed\*(C'\fR watchers for more info.
315.Sp 499.Sp
316It scales in the same way as the epoll backend, but the interface to the 500It scales in the same way as the epoll backend, but the interface to the
317kernel is more efficient (which says nothing about its actual speed, of 501kernel is more efficient (which says nothing about its actual speed, of
318course). While starting and stopping an I/O watcher does not cause an 502course). While stopping, setting and starting an I/O watcher does never
319extra syscall as with epoll, it still adds up to four event changes per 503cause an extra syscall as with \f(CW\*(C`EVBACKEND_EPOLL\*(C'\fR, it still adds up to
320incident, so its best to avoid that. 504two event changes per incident, support for \f(CW\*(C`fork ()\*(C'\fR is very bad and it
505drops fds silently in similarly hard-to-detect cases.
321.ie n .IP """EVBACKEND_DEVPOLL"" (value 16, Solaris 8)" 4 506.ie n .IP """EVBACKEND_DEVPOLL"" (value 16, Solaris 8)" 4
322.el .IP "\f(CWEVBACKEND_DEVPOLL\fR (value 16, Solaris 8)" 4 507.el .IP "\f(CWEVBACKEND_DEVPOLL\fR (value 16, Solaris 8)" 4
323.IX Item "EVBACKEND_DEVPOLL (value 16, Solaris 8)" 508.IX Item "EVBACKEND_DEVPOLL (value 16, Solaris 8)"
324This is not implemented yet (and might never be). 509This is not implemented yet (and might never be).
325.ie n .IP """EVBACKEND_PORT"" (value 32, Solaris 10)" 4 510.ie n .IP """EVBACKEND_PORT"" (value 32, Solaris 10)" 4
326.el .IP "\f(CWEVBACKEND_PORT\fR (value 32, Solaris 10)" 4 511.el .IP "\f(CWEVBACKEND_PORT\fR (value 32, Solaris 10)" 4
327.IX Item "EVBACKEND_PORT (value 32, Solaris 10)" 512.IX Item "EVBACKEND_PORT (value 32, Solaris 10)"
328This uses the Solaris 10 port mechanism. As with everything on Solaris, 513This uses the Solaris 10 event port mechanism. As with everything on Solaris,
329it's really slow, but it still scales very well (O(active_fds)). 514it's really slow, but it still scales very well (O(active_fds)).
515.Sp
516Please note that solaris event ports can deliver a lot of spurious
517notifications, so you need to use non-blocking I/O or other means to avoid
518blocking when no data (or space) is available.
330.ie n .IP """EVBACKEND_ALL""" 4 519.ie n .IP """EVBACKEND_ALL""" 4
331.el .IP "\f(CWEVBACKEND_ALL\fR" 4 520.el .IP "\f(CWEVBACKEND_ALL\fR" 4
332.IX Item "EVBACKEND_ALL" 521.IX Item "EVBACKEND_ALL"
333Try all backends (even potentially broken ones that wouldn't be tried 522Try all backends (even potentially broken ones that wouldn't be tried
334with \f(CW\*(C`EVFLAG_AUTO\*(C'\fR). Since this is a mask, you can do stuff such as 523with \f(CW\*(C`EVFLAG_AUTO\*(C'\fR). Since this is a mask, you can do stuff such as
… …
338.Sp 527.Sp
339If one or more of these are ored into the flags value, then only these 528If one or more of these are ored into the flags value, then only these
340backends will be tried (in the reverse order as given here). If none are 529backends will be tried (in the reverse order as given here). If none are
341specified, most compiled-in backend will be tried, usually in reverse 530specified, most compiled-in backend will be tried, usually in reverse
342order of their flag values :) 531order of their flag values :)
532.Sp
533The most typical usage is like this:
534.Sp
535.Vb 2
536\& if (!ev_default_loop (0))
537\& fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
538.Ve
539.Sp
540Restrict libev to the select and poll backends, and do not allow
541environment settings to be taken into account:
542.Sp
543.Vb 1
544\& ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
545.Ve
546.Sp
547Use whatever libev has to offer, but make sure that kqueue is used if
548available (warning, breaks stuff, best use only with your own private
549event loop and only if you know the \s-1OS\s0 supports your types of fds):
550.Sp
551.Vb 1
552\& ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
553.Ve
343.RE 554.RE
344.IP "struct ev_loop *ev_loop_new (unsigned int flags)" 4 555.IP "struct ev_loop *ev_loop_new (unsigned int flags)" 4
345.IX Item "struct ev_loop *ev_loop_new (unsigned int flags)" 556.IX Item "struct ev_loop *ev_loop_new (unsigned int flags)"
346Similar to \f(CW\*(C`ev_default_loop\*(C'\fR, but always creates a new event loop that is 557Similar to \f(CW\*(C`ev_default_loop\*(C'\fR, but always creates a new event loop that is
347always distinct from the default loop. Unlike the default loop, it cannot 558always distinct from the default loop. Unlike the default loop, it cannot
348handle signal and child watchers, and attempts to do so will be greeted by 559handle signal and child watchers, and attempts to do so will be greeted by
349undefined behaviour (or a failed assertion if assertions are enabled). 560undefined behaviour (or a failed assertion if assertions are enabled).
561.Sp
562Example: Try to create a event loop that uses epoll and nothing else.
563.Sp
564.Vb 3
565\& struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
566\& if (!epoller)
567\& fatal ("no epoll found here, maybe it hides under your chair");
568.Ve
350.IP "ev_default_destroy ()" 4 569.IP "ev_default_destroy ()" 4
351.IX Item "ev_default_destroy ()" 570.IX Item "ev_default_destroy ()"
352Destroys the default loop again (frees all memory and kernel state 571Destroys the default loop again (frees all memory and kernel state
353etc.). This stops all registered event watchers (by not touching them in 572etc.). None of the active event watchers will be stopped in the normal
354any way whatsoever, although you cannot rely on this :). 573sense, so e.g. \f(CW\*(C`ev_is_active\*(C'\fR might still return true. It is your
574responsibility to either stop all watchers cleanly yoursef \fIbefore\fR
575calling this function, or cope with the fact afterwards (which is usually
576the easiest thing, you can just ignore the watchers and/or \f(CW\*(C`free ()\*(C'\fR them
577for example).
578.Sp
579Note that certain global state, such as signal state, will not be freed by
580this function, and related watchers (such as signal and child watchers)
581would need to be stopped manually.
582.Sp
583In general it is not advisable to call this function except in the
584rare occasion where you really need to free e.g. the signal handling
585pipe fds. If you need dynamically allocated loops it is better to use
586\&\f(CW\*(C`ev_loop_new\*(C'\fR and \f(CW\*(C`ev_loop_destroy\*(C'\fR).
355.IP "ev_loop_destroy (loop)" 4 587.IP "ev_loop_destroy (loop)" 4
356.IX Item "ev_loop_destroy (loop)" 588.IX Item "ev_loop_destroy (loop)"
357Like \f(CW\*(C`ev_default_destroy\*(C'\fR, but destroys an event loop created by an 589Like \f(CW\*(C`ev_default_destroy\*(C'\fR, but destroys an event loop created by an
358earlier call to \f(CW\*(C`ev_loop_new\*(C'\fR. 590earlier call to \f(CW\*(C`ev_loop_new\*(C'\fR.
359.IP "ev_default_fork ()" 4 591.IP "ev_default_fork ()" 4
… …
381.IP "ev_loop_fork (loop)" 4 613.IP "ev_loop_fork (loop)" 4
382.IX Item "ev_loop_fork (loop)" 614.IX Item "ev_loop_fork (loop)"
383Like \f(CW\*(C`ev_default_fork\*(C'\fR, but acts on an event loop created by 615Like \f(CW\*(C`ev_default_fork\*(C'\fR, but acts on an event loop created by
384\&\f(CW\*(C`ev_loop_new\*(C'\fR. Yes, you have to call this on every allocated event loop 616\&\f(CW\*(C`ev_loop_new\*(C'\fR. Yes, you have to call this on every allocated event loop
385after fork, and how you do this is entirely your own problem. 617after fork, and how you do this is entirely your own problem.
618.IP "unsigned int ev_loop_count (loop)" 4
619.IX Item "unsigned int ev_loop_count (loop)"
620Returns the count of loop iterations for the loop, which is identical to
621the number of times libev did poll for new events. It starts at \f(CW0\fR and
622happily wraps around with enough iterations.
623.Sp
624This value can sometimes be useful as a generation counter of sorts (it
625\&\*(L"ticks\*(R" the number of loop iterations), as it roughly corresponds with
626\&\f(CW\*(C`ev_prepare\*(C'\fR and \f(CW\*(C`ev_check\*(C'\fR calls.
386.IP "unsigned int ev_backend (loop)" 4 627.IP "unsigned int ev_backend (loop)" 4
387.IX Item "unsigned int ev_backend (loop)" 628.IX Item "unsigned int ev_backend (loop)"
388Returns one of the \f(CW\*(C`EVBACKEND_*\*(C'\fR flags indicating the event backend in 629Returns one of the \f(CW\*(C`EVBACKEND_*\*(C'\fR flags indicating the event backend in
389use. 630use.
390.IP "ev_tstamp ev_now (loop)" 4 631.IP "ev_tstamp ev_now (loop)" 4
391.IX Item "ev_tstamp ev_now (loop)" 632.IX Item "ev_tstamp ev_now (loop)"
392Returns the current \*(L"event loop time\*(R", which is the time the event loop 633Returns the current \*(L"event loop time\*(R", which is the time the event loop
393got events and started processing them. This timestamp does not change 634received events and started processing them. This timestamp does not
394as long as callbacks are being processed, and this is also the base time 635change as long as callbacks are being processed, and this is also the base
395used for relative timers. You can treat it as the timestamp of the event 636time used for relative timers. You can treat it as the timestamp of the
396occuring (or more correctly, the mainloop finding out about it). 637event occurring (or more correctly, libev finding out about it).
397.IP "ev_loop (loop, int flags)" 4 638.IP "ev_loop (loop, int flags)" 4
398.IX Item "ev_loop (loop, int flags)" 639.IX Item "ev_loop (loop, int flags)"
399Finally, this is it, the event handler. This function usually is called 640Finally, this is it, the event handler. This function usually is called
400after you initialised all your watchers and you want to start handling 641after you initialised all your watchers and you want to start handling
401events. 642events.
402.Sp 643.Sp
403If the flags argument is specified as 0, it will not return until either 644If the flags argument is specified as \f(CW0\fR, it will not return until
404no event watchers are active anymore or \f(CW\*(C`ev_unloop\*(C'\fR was called. 645either no event watchers are active anymore or \f(CW\*(C`ev_unloop\*(C'\fR was called.
646.Sp
647Please note that an explicit \f(CW\*(C`ev_unloop\*(C'\fR is usually better than
648relying on all watchers to be stopped when deciding when a program has
649finished (especially in interactive programs), but having a program that
650automatically loops as long as it has to and no longer by virtue of
651relying on its watchers stopping correctly is a thing of beauty.
405.Sp 652.Sp
406A flags value of \f(CW\*(C`EVLOOP_NONBLOCK\*(C'\fR will look for new events, will handle 653A flags value of \f(CW\*(C`EVLOOP_NONBLOCK\*(C'\fR will look for new events, will handle
407those events and any outstanding ones, but will not block your process in 654those events and any outstanding ones, but will not block your process in
408case there are no events and will return after one iteration of the loop. 655case there are no events and will return after one iteration of the loop.
409.Sp 656.Sp
410A flags value of \f(CW\*(C`EVLOOP_ONESHOT\*(C'\fR will look for new events (waiting if 657A flags value of \f(CW\*(C`EVLOOP_ONESHOT\*(C'\fR will look for new events (waiting if
411neccessary) and will handle those and any outstanding ones. It will block 658neccessary) and will handle those and any outstanding ones. It will block
412your process until at least one new event arrives, and will return after 659your process until at least one new event arrives, and will return after
413one iteration of the loop. 660one iteration of the loop. This is useful if you are waiting for some
661external event in conjunction with something not expressible using other
662libev watchers. However, a pair of \f(CW\*(C`ev_prepare\*(C'\fR/\f(CW\*(C`ev_check\*(C'\fR watchers is
663usually a better approach for this kind of thing.
414.Sp 664.Sp
415This flags value could be used to implement alternative looping
416constructs, but the \f(CW\*(C`prepare\*(C'\fR and \f(CW\*(C`check\*(C'\fR watchers provide a better and
417more generic mechanism.
418.Sp
419Here are the gory details of what ev_loop does: 665Here are the gory details of what \f(CW\*(C`ev_loop\*(C'\fR does:
420.Sp 666.Sp
421.Vb 15 667.Vb 19
668\& - Before the first iteration, call any pending watchers.
422\& 1. If there are no active watchers (reference count is zero), return. 669\& * If there are no active watchers (reference count is zero), return.
423\& 2. Queue and immediately call all prepare watchers. 670\& - Queue all prepare watchers and then call all outstanding watchers.
424\& 3. If we have been forked, recreate the kernel state. 671\& - If we have been forked, recreate the kernel state.
425\& 4. Update the kernel state with all outstanding changes. 672\& - Update the kernel state with all outstanding changes.
426\& 5. Update the "event loop time". 673\& - Update the "event loop time".
427\& 6. Calculate for how long to block. 674\& - Calculate for how long to block.
428\& 7. Block the process, waiting for events. 675\& - Block the process, waiting for any events.
676\& - Queue all outstanding I/O (fd) events.
429\& 8. Update the "event loop time" and do time jump handling. 677\& - Update the "event loop time" and do time jump handling.
430\& 9. Queue all outstanding timers. 678\& - Queue all outstanding timers.
431\& 10. Queue all outstanding periodics. 679\& - Queue all outstanding periodics.
432\& 11. If no events are pending now, queue all idle watchers. 680\& - If no events are pending now, queue all idle watchers.
433\& 12. Queue all check watchers. 681\& - Queue all check watchers.
434\& 13. Call all queued watchers in reverse order (i.e. check watchers first). 682\& - Call all queued watchers in reverse order (i.e. check watchers first).
683\& Signals and child watchers are implemented as I/O watchers, and will
684\& be handled here by queueing them when their watcher gets executed.
435\& 14. If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK 685\& - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
436\& was used, return, otherwise continue with step #1. 686\& were used, return, otherwise continue with step *.
687.Ve
688.Sp
689Example: Queue some jobs and then loop until no events are outsanding
690anymore.
691.Sp
692.Vb 4
693\& ... queue jobs here, make sure they register event watchers as long
694\& ... as they still have work to do (even an idle watcher will do..)
695\& ev_loop (my_loop, 0);
696\& ... jobs done. yeah!
437.Ve 697.Ve
438.IP "ev_unloop (loop, how)" 4 698.IP "ev_unloop (loop, how)" 4
439.IX Item "ev_unloop (loop, how)" 699.IX Item "ev_unloop (loop, how)"
440Can be used to make a call to \f(CW\*(C`ev_loop\*(C'\fR return early (but only after it 700Can be used to make a call to \f(CW\*(C`ev_loop\*(C'\fR return early (but only after it
441has processed all outstanding events). The \f(CW\*(C`how\*(C'\fR argument must be either 701has processed all outstanding events). The \f(CW\*(C`how\*(C'\fR argument must be either
… …
455example, libev itself uses this for its internal signal pipe: It is not 715example, libev itself uses this for its internal signal pipe: It is not
456visible to the libev user and should not keep \f(CW\*(C`ev_loop\*(C'\fR from exiting if 716visible to the libev user and should not keep \f(CW\*(C`ev_loop\*(C'\fR from exiting if
457no event watchers registered by it are active. It is also an excellent 717no event watchers registered by it are active. It is also an excellent
458way to do this for generic recurring timers or from within third-party 718way to do this for generic recurring timers or from within third-party
459libraries. Just remember to \fIunref after start\fR and \fIref before stop\fR. 719libraries. Just remember to \fIunref after start\fR and \fIref before stop\fR.
720.Sp
721Example: Create a signal watcher, but keep it from keeping \f(CW\*(C`ev_loop\*(C'\fR
722running when nothing else is active.
723.Sp
724.Vb 4
725\& struct ev_signal exitsig;
726\& ev_signal_init (&exitsig, sig_cb, SIGINT);
727\& ev_signal_start (loop, &exitsig);
728\& evf_unref (loop);
729.Ve
730.Sp
731Example: For some weird reason, unregister the above signal handler again.
732.Sp
733.Vb 2
734\& ev_ref (loop);
735\& ev_signal_stop (loop, &exitsig);
736.Ve
737.IP "ev_set_io_collect_interval (loop, ev_tstamp interval)" 4
738.IX Item "ev_set_io_collect_interval (loop, ev_tstamp interval)"
739.PD 0
740.IP "ev_set_timeout_collect_interval (loop, ev_tstamp interval)" 4
741.IX Item "ev_set_timeout_collect_interval (loop, ev_tstamp interval)"
742.PD
743These advanced functions influence the time that libev will spend waiting
744for events. Both are by default \f(CW0\fR, meaning that libev will try to
745invoke timer/periodic callbacks and I/O callbacks with minimum latency.
746.Sp
747Setting these to a higher value (the \f(CW\*(C`interval\*(C'\fR \fImust\fR be >= \f(CW0\fR)
748allows libev to delay invocation of I/O and timer/periodic callbacks to
749increase efficiency of loop iterations.
750.Sp
751The background is that sometimes your program runs just fast enough to
752handle one (or very few) event(s) per loop iteration. While this makes
753the program responsive, it also wastes a lot of \s-1CPU\s0 time to poll for new
754events, especially with backends like \f(CW\*(C`select ()\*(C'\fR which have a high
755overhead for the actual polling but can deliver many events at once.
756.Sp
757By setting a higher \fIio collect interval\fR you allow libev to spend more
758time collecting I/O events, so you can handle more events per iteration,
759at the cost of increasing latency. Timeouts (both \f(CW\*(C`ev_periodic\*(C'\fR and
760\&\f(CW\*(C`ev_timer\*(C'\fR) will be not affected. Setting this to a non-null bvalue will
761introduce an additional \f(CW\*(C`ev_sleep ()\*(C'\fR call into most loop iterations.
762.Sp
763Likewise, by setting a higher \fItimeout collect interval\fR you allow libev
764to spend more time collecting timeouts, at the expense of increased
765latency (the watcher callback will be called later). \f(CW\*(C`ev_io\*(C'\fR watchers
766will not be affected. Setting this to a non-null value will not introduce
767any overhead in libev.
768.Sp
769Many (busy) programs can usually benefit by setting the io collect
770interval to a value near \f(CW0.1\fR or so, which is often enough for
771interactive servers (of course not for games), likewise for timeouts. It
772usually doesn't make much sense to set it to a lower value than \f(CW0.01\fR,
773as this approsaches the timing granularity of most systems.
460.SH "ANATOMY OF A WATCHER" 774.SH "ANATOMY OF A WATCHER"
461.IX Header "ANATOMY OF A WATCHER" 775.IX Header "ANATOMY OF A WATCHER"
462A watcher is a structure that you create and register to record your 776A watcher is a structure that you create and register to record your
463interest in some event. For instance, if you want to wait for \s-1STDIN\s0 to 777interest in some event. For instance, if you want to wait for \s-1STDIN\s0 to
464become readable, you would create an \f(CW\*(C`ev_io\*(C'\fR watcher for that: 778become readable, you would create an \f(CW\*(C`ev_io\*(C'\fR watcher for that:
… …
500*)\*(C'\fR), and you can stop watching for events at any time by calling the 814*)\*(C'\fR), and you can stop watching for events at any time by calling the
501corresponding stop function (\f(CW\*(C`ev_<type>_stop (loop, watcher *)\*(C'\fR. 815corresponding stop function (\f(CW\*(C`ev_<type>_stop (loop, watcher *)\*(C'\fR.
502.PP 816.PP
503As long as your watcher is active (has been started but not stopped) you 817As long as your watcher is active (has been started but not stopped) you
504must not touch the values stored in it. Most specifically you must never 818must not touch the values stored in it. Most specifically you must never
505reinitialise it or call its set macro. 819reinitialise it or call its \f(CW\*(C`set\*(C'\fR macro.
506.PP
507You can check whether an event is active by calling the \f(CW\*(C`ev_is_active
508(watcher *)\*(C'\fR macro. To see whether an event is outstanding (but the
509callback for it has not been called yet) you can use the \f(CW\*(C`ev_is_pending
510(watcher *)\*(C'\fR macro.
511.PP 820.PP
512Each and every callback receives the event loop pointer as first, the 821Each and every callback receives the event loop pointer as first, the
513registered watcher structure as second, and a bitset of received events as 822registered watcher structure as second, and a bitset of received events as
514third argument. 823third argument.
515.PP 824.PP
… …
540The signal specified in the \f(CW\*(C`ev_signal\*(C'\fR watcher has been received by a thread. 849The signal specified in the \f(CW\*(C`ev_signal\*(C'\fR watcher has been received by a thread.
541.ie n .IP """EV_CHILD""" 4 850.ie n .IP """EV_CHILD""" 4
542.el .IP "\f(CWEV_CHILD\fR" 4 851.el .IP "\f(CWEV_CHILD\fR" 4
543.IX Item "EV_CHILD" 852.IX Item "EV_CHILD"
544The pid specified in the \f(CW\*(C`ev_child\*(C'\fR watcher has received a status change. 853The pid specified in the \f(CW\*(C`ev_child\*(C'\fR watcher has received a status change.
854.ie n .IP """EV_STAT""" 4
855.el .IP "\f(CWEV_STAT\fR" 4
856.IX Item "EV_STAT"
857The path specified in the \f(CW\*(C`ev_stat\*(C'\fR watcher changed its attributes somehow.
545.ie n .IP """EV_IDLE""" 4 858.ie n .IP """EV_IDLE""" 4
546.el .IP "\f(CWEV_IDLE\fR" 4 859.el .IP "\f(CWEV_IDLE\fR" 4
547.IX Item "EV_IDLE" 860.IX Item "EV_IDLE"
548The \f(CW\*(C`ev_idle\*(C'\fR watcher has determined that you have nothing better to do. 861The \f(CW\*(C`ev_idle\*(C'\fR watcher has determined that you have nothing better to do.
549.ie n .IP """EV_PREPARE""" 4 862.ie n .IP """EV_PREPARE""" 4
… …
559\&\f(CW\*(C`ev_loop\*(C'\fR has gathered them, but before it invokes any callbacks for any 872\&\f(CW\*(C`ev_loop\*(C'\fR has gathered them, but before it invokes any callbacks for any
560received events. Callbacks of both watcher types can start and stop as 873received events. Callbacks of both watcher types can start and stop as
561many watchers as they want, and all of them will be taken into account 874many watchers as they want, and all of them will be taken into account
562(for example, a \f(CW\*(C`ev_prepare\*(C'\fR watcher might start an idle watcher to keep 875(for example, a \f(CW\*(C`ev_prepare\*(C'\fR watcher might start an idle watcher to keep
563\&\f(CW\*(C`ev_loop\*(C'\fR from blocking). 876\&\f(CW\*(C`ev_loop\*(C'\fR from blocking).
877.ie n .IP """EV_EMBED""" 4
878.el .IP "\f(CWEV_EMBED\fR" 4
879.IX Item "EV_EMBED"
880The embedded event loop specified in the \f(CW\*(C`ev_embed\*(C'\fR watcher needs attention.
881.ie n .IP """EV_FORK""" 4
882.el .IP "\f(CWEV_FORK\fR" 4
883.IX Item "EV_FORK"
884The event loop has been resumed in the child process after fork (see
885\&\f(CW\*(C`ev_fork\*(C'\fR).
564.ie n .IP """EV_ERROR""" 4 886.ie n .IP """EV_ERROR""" 4
565.el .IP "\f(CWEV_ERROR\fR" 4 887.el .IP "\f(CWEV_ERROR\fR" 4
566.IX Item "EV_ERROR" 888.IX Item "EV_ERROR"
567An unspecified error has occured, the watcher has been stopped. This might 889An unspecified error has occured, the watcher has been stopped. This might
568happen because the watcher could not be properly started because libev 890happen because the watcher could not be properly started because libev
… …
573Libev will usually signal a few \*(L"dummy\*(R" events together with an error, 895Libev will usually signal a few \*(L"dummy\*(R" events together with an error,
574for example it might indicate that a fd is readable or writable, and if 896for example it might indicate that a fd is readable or writable, and if
575your callbacks is well-written it can just attempt the operation and cope 897your callbacks is well-written it can just attempt the operation and cope
576with the error from \fIread()\fR or \fIwrite()\fR. This will not work in multithreaded 898with the error from \fIread()\fR or \fIwrite()\fR. This will not work in multithreaded
577programs, though, so beware. 899programs, though, so beware.
900.Sh "\s-1GENERIC\s0 \s-1WATCHER\s0 \s-1FUNCTIONS\s0"
901.IX Subsection "GENERIC WATCHER FUNCTIONS"
902In the following description, \f(CW\*(C`TYPE\*(C'\fR stands for the watcher type,
903e.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.
904.ie n .IP """ev_init"" (ev_TYPE *watcher, callback)" 4
905.el .IP "\f(CWev_init\fR (ev_TYPE *watcher, callback)" 4
906.IX Item "ev_init (ev_TYPE *watcher, callback)"
907This macro initialises the generic portion of a watcher. The contents
908of the watcher object can be arbitrary (so \f(CW\*(C`malloc\*(C'\fR will do). Only
909the generic parts of the watcher are initialised, you \fIneed\fR to call
910the type-specific \f(CW\*(C`ev_TYPE_set\*(C'\fR macro afterwards to initialise the
911type-specific parts. For each type there is also a \f(CW\*(C`ev_TYPE_init\*(C'\fR macro
912which rolls both calls into one.
913.Sp
914You can reinitialise a watcher at any time as long as it has been stopped
915(or never started) and there are no pending events outstanding.
916.Sp
917The callback is always of type \f(CW\*(C`void (*)(ev_loop *loop, ev_TYPE *watcher,
918int revents)\*(C'\fR.
919.ie n .IP """ev_TYPE_set"" (ev_TYPE *, [args])" 4
920.el .IP "\f(CWev_TYPE_set\fR (ev_TYPE *, [args])" 4
921.IX Item "ev_TYPE_set (ev_TYPE *, [args])"
922This macro initialises the type-specific parts of a watcher. You need to
923call \f(CW\*(C`ev_init\*(C'\fR at least once before you call this macro, but you can
924call \f(CW\*(C`ev_TYPE_set\*(C'\fR any number of times. You must not, however, call this
925macro on a watcher that is active (it can be pending, however, which is a
926difference to the \f(CW\*(C`ev_init\*(C'\fR macro).
927.Sp
928Although some watcher types do not have type-specific arguments
929(e.g. \f(CW\*(C`ev_prepare\*(C'\fR) you still need to call its \f(CW\*(C`set\*(C'\fR macro.
930.ie n .IP """ev_TYPE_init"" (ev_TYPE *watcher, callback, [args])" 4
931.el .IP "\f(CWev_TYPE_init\fR (ev_TYPE *watcher, callback, [args])" 4
932.IX Item "ev_TYPE_init (ev_TYPE *watcher, callback, [args])"
933This convinience macro rolls both \f(CW\*(C`ev_init\*(C'\fR and \f(CW\*(C`ev_TYPE_set\*(C'\fR macro
934calls into a single call. This is the most convinient method to initialise
935a watcher. The same limitations apply, of course.
936.ie n .IP """ev_TYPE_start"" (loop *, ev_TYPE *watcher)" 4
937.el .IP "\f(CWev_TYPE_start\fR (loop *, ev_TYPE *watcher)" 4
938.IX Item "ev_TYPE_start (loop *, ev_TYPE *watcher)"
939Starts (activates) the given watcher. Only active watchers will receive
940events. If the watcher is already active nothing will happen.
941.ie n .IP """ev_TYPE_stop"" (loop *, ev_TYPE *watcher)" 4
942.el .IP "\f(CWev_TYPE_stop\fR (loop *, ev_TYPE *watcher)" 4
943.IX Item "ev_TYPE_stop (loop *, ev_TYPE *watcher)"
944Stops the given watcher again (if active) and clears the pending
945status. It is possible that stopped watchers are pending (for example,
946non-repeating timers are being stopped when they become pending), but
947\&\f(CW\*(C`ev_TYPE_stop\*(C'\fR ensures that the watcher is neither active nor pending. If
948you want to free or reuse the memory used by the watcher it is therefore a
949good idea to always call its \f(CW\*(C`ev_TYPE_stop\*(C'\fR function.
950.IP "bool ev_is_active (ev_TYPE *watcher)" 4
951.IX Item "bool ev_is_active (ev_TYPE *watcher)"
952Returns a true value iff the watcher is active (i.e. it has been started
953and not yet been stopped). As long as a watcher is active you must not modify
954it.
955.IP "bool ev_is_pending (ev_TYPE *watcher)" 4
956.IX Item "bool ev_is_pending (ev_TYPE *watcher)"
957Returns a true value iff the watcher is pending, (i.e. it has outstanding
958events but its callback has not yet been invoked). As long as a watcher
959is pending (but not active) you must not call an init function on it (but
960\&\f(CW\*(C`ev_TYPE_set\*(C'\fR is safe), you must not change its priority, and you must
961make sure the watcher is available to libev (e.g. you cannot \f(CW\*(C`free ()\*(C'\fR
962it).
963.IP "callback ev_cb (ev_TYPE *watcher)" 4
964.IX Item "callback ev_cb (ev_TYPE *watcher)"
965Returns the callback currently set on the watcher.
966.IP "ev_cb_set (ev_TYPE *watcher, callback)" 4
967.IX Item "ev_cb_set (ev_TYPE *watcher, callback)"
968Change the callback. You can change the callback at virtually any time
969(modulo threads).
970.IP "ev_set_priority (ev_TYPE *watcher, priority)" 4
971.IX Item "ev_set_priority (ev_TYPE *watcher, priority)"
972.PD 0
973.IP "int ev_priority (ev_TYPE *watcher)" 4
974.IX Item "int ev_priority (ev_TYPE *watcher)"
975.PD
976Set and query the priority of the watcher. The priority is a small
977integer between \f(CW\*(C`EV_MAXPRI\*(C'\fR (default: \f(CW2\fR) and \f(CW\*(C`EV_MINPRI\*(C'\fR
978(default: \f(CW\*(C`\-2\*(C'\fR). Pending watchers with higher priority will be invoked
979before watchers with lower priority, but priority will not keep watchers
980from being executed (except for \f(CW\*(C`ev_idle\*(C'\fR watchers).
981.Sp
982This means that priorities are \fIonly\fR used for ordering callback
983invocation after new events have been received. This is useful, for
984example, to reduce latency after idling, or more often, to bind two
985watchers on the same event and make sure one is called first.
986.Sp
987If you need to suppress invocation when higher priority events are pending
988you need to look at \f(CW\*(C`ev_idle\*(C'\fR watchers, which provide this functionality.
989.Sp
990You \fImust not\fR change the priority of a watcher as long as it is active or
991pending.
992.Sp
993The default priority used by watchers when no priority has been set is
994always \f(CW0\fR, which is supposed to not be too high and not be too low :).
995.Sp
996Setting a priority outside the range of \f(CW\*(C`EV_MINPRI\*(C'\fR to \f(CW\*(C`EV_MAXPRI\*(C'\fR is
997fine, as long as you do not mind that the priority value you query might
998or might not have been adjusted to be within valid range.
999.IP "ev_invoke (loop, ev_TYPE *watcher, int revents)" 4
1000.IX Item "ev_invoke (loop, ev_TYPE *watcher, int revents)"
1001Invoke the \f(CW\*(C`watcher\*(C'\fR with the given \f(CW\*(C`loop\*(C'\fR and \f(CW\*(C`revents\*(C'\fR. Neither
1002\&\f(CW\*(C`loop\*(C'\fR nor \f(CW\*(C`revents\*(C'\fR need to be valid as long as the watcher callback
1003can deal with that fact.
1004.IP "int ev_clear_pending (loop, ev_TYPE *watcher)" 4
1005.IX Item "int ev_clear_pending (loop, ev_TYPE *watcher)"
1006If the watcher is pending, this function returns clears its pending status
1007and returns its \f(CW\*(C`revents\*(C'\fR bitset (as if its callback was invoked). If the
1008watcher isn't pending it does nothing and returns \f(CW0\fR.
578.Sh "\s-1ASSOCIATING\s0 \s-1CUSTOM\s0 \s-1DATA\s0 \s-1WITH\s0 A \s-1WATCHER\s0" 1009.Sh "\s-1ASSOCIATING\s0 \s-1CUSTOM\s0 \s-1DATA\s0 \s-1WITH\s0 A \s-1WATCHER\s0"
579.IX Subsection "ASSOCIATING CUSTOM DATA WITH A WATCHER" 1010.IX Subsection "ASSOCIATING CUSTOM DATA WITH A WATCHER"
580Each watcher has, by default, a member \f(CW\*(C`void *data\*(C'\fR that you can change 1011Each watcher has, by default, a member \f(CW\*(C`void *data\*(C'\fR that you can change
581and read at any time, libev will completely ignore it. This can be used 1012and read at any time, libev will completely ignore it. This can be used
582to associate arbitrary data with your watcher. If you need more data and 1013to associate arbitrary data with your watcher. If you need more data and
… …
603\& struct my_io *w = (struct my_io *)w_; 1034\& struct my_io *w = (struct my_io *)w_;
604\& ... 1035\& ...
605\& } 1036\& }
606.Ve 1037.Ve
607.PP 1038.PP
608More interesting and less C\-conformant ways of catsing your callback type 1039More interesting and less C\-conformant ways of casting your callback type
609have been omitted.... 1040instead have been omitted.
1041.PP
1042Another common scenario is having some data structure with multiple
1043watchers:
1044.PP
1045.Vb 6
1046\& struct my_biggy
1047\& {
1048\& int some_data;
1049\& ev_timer t1;
1050\& ev_timer t2;
1051\& }
1052.Ve
1053.PP
1054In this case getting the pointer to \f(CW\*(C`my_biggy\*(C'\fR is a bit more complicated,
1055you need to use \f(CW\*(C`offsetof\*(C'\fR:
1056.PP
1057.Vb 1
1058\& #include <stddef.h>
1059.Ve
1060.PP
1061.Vb 6
1062\& static void
1063\& t1_cb (EV_P_ struct ev_timer *w, int revents)
1064\& {
1065\& struct my_biggy big = (struct my_biggy *
1066\& (((char *)w) - offsetof (struct my_biggy, t1));
1067\& }
1068.Ve
1069.PP
1070.Vb 6
1071\& static void
1072\& t2_cb (EV_P_ struct ev_timer *w, int revents)
1073\& {
1074\& struct my_biggy big = (struct my_biggy *
1075\& (((char *)w) - offsetof (struct my_biggy, t2));
1076\& }
1077.Ve
610.SH "WATCHER TYPES" 1078.SH "WATCHER TYPES"
611.IX Header "WATCHER TYPES" 1079.IX Header "WATCHER TYPES"
612This section describes each watcher in detail, but will not repeat 1080This section describes each watcher in detail, but will not repeat
613information given in the last section. 1081information given in the last section. Any initialisation/set macros,
1082functions and members specific to the watcher type are explained.
1083.PP
1084Members are additionally marked with either \fI[read\-only]\fR, meaning that,
1085while the watcher is active, you can look at the member and expect some
1086sensible content, but you must not modify it (you can modify it while the
1087watcher is stopped to your hearts content), or \fI[read\-write]\fR, which
1088means you can expect it to have some sensible content while the watcher
1089is active, but you can also modify it. Modifying it may not do something
1090sensible or take immediate effect (or do anything at all), but libev will
1091not crash or malfunction in any way.
614.ie n .Sh """ev_io"" \- is this file descriptor readable or writable" 1092.ie n .Sh """ev_io"" \- is this file descriptor readable or writable?"
615.el .Sh "\f(CWev_io\fP \- is this file descriptor readable or writable" 1093.el .Sh "\f(CWev_io\fP \- is this file descriptor readable or writable?"
616.IX Subsection "ev_io - is this file descriptor readable or writable" 1094.IX Subsection "ev_io - is this file descriptor readable or writable?"
617I/O watchers check whether a file descriptor is readable or writable 1095I/O watchers check whether a file descriptor is readable or writable
618in each iteration of the event loop (This behaviour is called 1096in each iteration of the event loop, or, more precisely, when reading
619level-triggering because you keep receiving events as long as the 1097would not block the process and writing would at least be able to write
620condition persists. Remember you can stop the watcher if you don't want to 1098some data. This behaviour is called level-triggering because you keep
621act on the event and neither want to receive future events). 1099receiving events as long as the condition persists. Remember you can stop
1100the watcher if you don't want to act on the event and neither want to
1101receive future events.
622.PP 1102.PP
623In general you can register as many read and/or write event watchers per 1103In general you can register as many read and/or write event watchers per
624fd as you want (as long as you don't confuse yourself). Setting all file 1104fd as you want (as long as you don't confuse yourself). Setting all file
625descriptors to non-blocking mode is also usually a good idea (but not 1105descriptors to non-blocking mode is also usually a good idea (but not
626required if you know what you are doing). 1106required if you know what you are doing).
627.PP 1107.PP
628You have to be careful with dup'ed file descriptors, though. Some backends 1108You have to be careful with dup'ed file descriptors, though. Some backends
629(the linux epoll backend is a notable example) cannot handle dup'ed file 1109(the linux epoll backend is a notable example) cannot handle dup'ed file
630descriptors correctly if you register interest in two or more fds pointing 1110descriptors correctly if you register interest in two or more fds pointing
631to the same underlying file/socket etc. description (that is, they share 1111to the same underlying file/socket/etc. description (that is, they share
632the same underlying \*(L"file open\*(R"). 1112the same underlying \*(L"file open\*(R").
633.PP 1113.PP
634If you must do this, then force the use of a known-to-be-good backend 1114If you must do this, then force the use of a known-to-be-good backend
635(at the time of this writing, this includes only \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR and 1115(at the time of this writing, this includes only \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR and
636\&\f(CW\*(C`EVBACKEND_POLL\*(C'\fR). 1116\&\f(CW\*(C`EVBACKEND_POLL\*(C'\fR).
1117.PP
1118Another thing you have to watch out for is that it is quite easy to
1119receive \*(L"spurious\*(R" readyness notifications, that is your callback might
1120be called with \f(CW\*(C`EV_READ\*(C'\fR but a subsequent \f(CW\*(C`read\*(C'\fR(2) will actually block
1121because there is no data. Not only are some backends known to create a
1122lot of those (for example solaris ports), it is very easy to get into
1123this situation even with a relatively standard program structure. Thus
1124it is best to always use non-blocking I/O: An extra \f(CW\*(C`read\*(C'\fR(2) returning
1125\&\f(CW\*(C`EAGAIN\*(C'\fR is far preferable to a program hanging until some data arrives.
1126.PP
1127If you cannot run the fd in non-blocking mode (for example you should not
1128play around with an Xlib connection), then you have to seperately re-test
1129whether a file descriptor is really ready with a known-to-be good interface
1130such as poll (fortunately in our Xlib example, Xlib already does this on
1131its own, so its quite safe to use).
1132.PP
1133\fIThe special problem of disappearing file descriptors\fR
1134.IX Subsection "The special problem of disappearing file descriptors"
1135.PP
1136Some backends (e.g. kqueue, epoll) need to be told about closing a file
1137descriptor (either by calling \f(CW\*(C`close\*(C'\fR explicitly or by any other means,
1138such as \f(CW\*(C`dup\*(C'\fR). The reason is that you register interest in some file
1139descriptor, but when it goes away, the operating system will silently drop
1140this interest. If another file descriptor with the same number then is
1141registered with libev, there is no efficient way to see that this is, in
1142fact, a different file descriptor.
1143.PP
1144To avoid having to explicitly tell libev about such cases, libev follows
1145the following policy: Each time \f(CW\*(C`ev_io_set\*(C'\fR is being called, libev
1146will assume that this is potentially a new file descriptor, otherwise
1147it is assumed that the file descriptor stays the same. That means that
1148you \fIhave\fR to call \f(CW\*(C`ev_io_set\*(C'\fR (or \f(CW\*(C`ev_io_init\*(C'\fR) when you change the
1149descriptor even if the file descriptor number itself did not change.
1150.PP
1151This is how one would do it normally anyway, the important point is that
1152the libev application should not optimise around libev but should leave
1153optimisations to libev.
1154.PP
1155\fIThe special problem of dup'ed file descriptors\fR
1156.IX Subsection "The special problem of dup'ed file descriptors"
1157.PP
1158Some backends (e.g. epoll), cannot register events for file descriptors,
1159but only events for the underlying file descriptions. That menas when you
1160have \f(CW\*(C`dup ()\*(C'\fR'ed file descriptors and register events for them, only one
1161file descriptor might actually receive events.
1162.PP
1163There is no workaorund possible except not registering events
1164for potentially \f(CW\*(C`dup ()\*(C'\fR'ed file descriptors or to resort to
1165\&\f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or \f(CW\*(C`EVBACKEND_POLL\*(C'\fR.
1166.PP
1167\fIThe special problem of fork\fR
1168.IX Subsection "The special problem of fork"
1169.PP
1170Some backends (epoll, kqueue) do not support \f(CW\*(C`fork ()\*(C'\fR at all or exhibit
1171useless behaviour. Libev fully supports fork, but needs to be told about
1172it in the child.
1173.PP
1174To support fork in your programs, you either have to call
1175\&\f(CW\*(C`ev_default_fork ()\*(C'\fR or \f(CW\*(C`ev_loop_fork ()\*(C'\fR after a fork in the child,
1176enable \f(CW\*(C`EVFLAG_FORKCHECK\*(C'\fR, or resort to \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or
1177\&\f(CW\*(C`EVBACKEND_POLL\*(C'\fR.
1178.PP
1179\fIWatcher-Specific Functions\fR
1180.IX Subsection "Watcher-Specific Functions"
637.IP "ev_io_init (ev_io *, callback, int fd, int events)" 4 1181.IP "ev_io_init (ev_io *, callback, int fd, int events)" 4
638.IX Item "ev_io_init (ev_io *, callback, int fd, int events)" 1182.IX Item "ev_io_init (ev_io *, callback, int fd, int events)"
639.PD 0 1183.PD 0
640.IP "ev_io_set (ev_io *, int fd, int events)" 4 1184.IP "ev_io_set (ev_io *, int fd, int events)" 4
641.IX Item "ev_io_set (ev_io *, int fd, int events)" 1185.IX Item "ev_io_set (ev_io *, int fd, int events)"
642.PD 1186.PD
643Configures an \f(CW\*(C`ev_io\*(C'\fR watcher. The fd is the file descriptor to rceeive 1187Configures an \f(CW\*(C`ev_io\*(C'\fR watcher. The \f(CW\*(C`fd\*(C'\fR is the file descriptor to
644events 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 | 1188rceeive events for and events is either \f(CW\*(C`EV_READ\*(C'\fR, \f(CW\*(C`EV_WRITE\*(C'\fR or
645EV_WRITE\*(C'\fR to receive the given events. 1189\&\f(CW\*(C`EV_READ | EV_WRITE\*(C'\fR to receive the given events.
1190.IP "int fd [read\-only]" 4
1191.IX Item "int fd [read-only]"
1192The file descriptor being watched.
1193.IP "int events [read\-only]" 4
1194.IX Item "int events [read-only]"
1195The events being watched.
1196.PP
1197Example: Call \f(CW\*(C`stdin_readable_cb\*(C'\fR when \s-1STDIN_FILENO\s0 has become, well
1198readable, but only once. Since it is likely line\-buffered, you could
1199attempt to read a whole line in the callback.
1200.PP
1201.Vb 6
1202\& static void
1203\& stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1204\& {
1205\& ev_io_stop (loop, w);
1206\& .. read from stdin here (or from w->fd) and haqndle any I/O errors
1207\& }
1208.Ve
1209.PP
1210.Vb 6
1211\& ...
1212\& struct ev_loop *loop = ev_default_init (0);
1213\& struct ev_io stdin_readable;
1214\& ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
1215\& ev_io_start (loop, &stdin_readable);
1216\& ev_loop (loop, 0);
1217.Ve
646.ie n .Sh """ev_timer"" \- relative and optionally recurring timeouts" 1218.ie n .Sh """ev_timer"" \- relative and optionally repeating timeouts"
647.el .Sh "\f(CWev_timer\fP \- relative and optionally recurring timeouts" 1219.el .Sh "\f(CWev_timer\fP \- relative and optionally repeating timeouts"
648.IX Subsection "ev_timer - relative and optionally recurring timeouts" 1220.IX Subsection "ev_timer - relative and optionally repeating timeouts"
649Timer watchers are simple relative timers that generate an event after a 1221Timer watchers are simple relative timers that generate an event after a
650given time, and optionally repeating in regular intervals after that. 1222given time, and optionally repeating in regular intervals after that.
651.PP 1223.PP
652The timers are based on real time, that is, if you register an event that 1224The timers are based on real time, that is, if you register an event that
653times out after an hour and you reset your system clock to last years 1225times out after an hour and you reset your system clock to last years
… …
666.Ve 1238.Ve
667.PP 1239.PP
668The callback is guarenteed to be invoked only when its timeout has passed, 1240The callback is guarenteed to be invoked only when its timeout has passed,
669but if multiple timers become ready during the same loop iteration then 1241but if multiple timers become ready during the same loop iteration then
670order of execution is undefined. 1242order of execution is undefined.
1243.PP
1244\fIWatcher-Specific Functions and Data Members\fR
1245.IX Subsection "Watcher-Specific Functions and Data Members"
671.IP "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)" 4 1246.IP "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)" 4
672.IX Item "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)" 1247.IX Item "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)"
673.PD 0 1248.PD 0
674.IP "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)" 4 1249.IP "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)" 4
675.IX Item "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)" 1250.IX Item "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)"
… …
687.IP "ev_timer_again (loop)" 4 1262.IP "ev_timer_again (loop)" 4
688.IX Item "ev_timer_again (loop)" 1263.IX Item "ev_timer_again (loop)"
689This will act as if the timer timed out and restart it again if it is 1264This will act as if the timer timed out and restart it again if it is
690repeating. The exact semantics are: 1265repeating. The exact semantics are:
691.Sp 1266.Sp
1267If the timer is pending, its pending status is cleared.
1268.Sp
692If the timer is started but nonrepeating, stop it. 1269If the timer is started but nonrepeating, stop it (as if it timed out).
693.Sp 1270.Sp
694If the timer is repeating, either start it if necessary (with the repeat 1271If the timer is repeating, either start it if necessary (with the
695value), or reset the running timer to the repeat value. 1272\&\f(CW\*(C`repeat\*(C'\fR value), or reset the running timer to the \f(CW\*(C`repeat\*(C'\fR value.
696.Sp 1273.Sp
697This sounds a bit complicated, but here is a useful and typical 1274This sounds a bit complicated, but here is a useful and typical
698example: Imagine you have a tcp connection and you want a so-called idle 1275example: Imagine you have a tcp connection and you want a so-called idle
699timeout, that is, you want to be called when there have been, say, 60 1276timeout, that is, you want to be called when there have been, say, 60
700seconds of inactivity on the socket. The easiest way to do this is to 1277seconds of inactivity on the socket. The easiest way to do this is to
701configure an \f(CW\*(C`ev_timer\*(C'\fR with after=repeat=60 and calling ev_timer_again each 1278configure an \f(CW\*(C`ev_timer\*(C'\fR with a \f(CW\*(C`repeat\*(C'\fR value of \f(CW60\fR and then call
702time you successfully read or write some data. If you go into an idle 1279\&\f(CW\*(C`ev_timer_again\*(C'\fR each time you successfully read or write some data. If
703state where you do not expect data to travel on the socket, you can stop 1280you go into an idle state where you do not expect data to travel on the
704the timer, and again will automatically restart it if need be. 1281socket, you can \f(CW\*(C`ev_timer_stop\*(C'\fR the timer, and \f(CW\*(C`ev_timer_again\*(C'\fR will
1282automatically restart it if need be.
1283.Sp
1284That means you can ignore the \f(CW\*(C`after\*(C'\fR value and \f(CW\*(C`ev_timer_start\*(C'\fR
1285altogether and only ever use the \f(CW\*(C`repeat\*(C'\fR value and \f(CW\*(C`ev_timer_again\*(C'\fR:
1286.Sp
1287.Vb 8
1288\& ev_timer_init (timer, callback, 0., 5.);
1289\& ev_timer_again (loop, timer);
1290\& ...
1291\& timer->again = 17.;
1292\& ev_timer_again (loop, timer);
1293\& ...
1294\& timer->again = 10.;
1295\& ev_timer_again (loop, timer);
1296.Ve
1297.Sp
1298This is more slightly efficient then stopping/starting the timer each time
1299you want to modify its timeout value.
1300.IP "ev_tstamp repeat [read\-write]" 4
1301.IX Item "ev_tstamp repeat [read-write]"
1302The current \f(CW\*(C`repeat\*(C'\fR value. Will be used each time the watcher times out
1303or \f(CW\*(C`ev_timer_again\*(C'\fR is called and determines the next timeout (if any),
1304which is also when any modifications are taken into account.
1305.PP
1306Example: Create a timer that fires after 60 seconds.
1307.PP
1308.Vb 5
1309\& static void
1310\& one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1311\& {
1312\& .. one minute over, w is actually stopped right here
1313\& }
1314.Ve
1315.PP
1316.Vb 3
1317\& struct ev_timer mytimer;
1318\& ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1319\& ev_timer_start (loop, &mytimer);
1320.Ve
1321.PP
1322Example: Create a timeout timer that times out after 10 seconds of
1323inactivity.
1324.PP
1325.Vb 5
1326\& static void
1327\& timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1328\& {
1329\& .. ten seconds without any activity
1330\& }
1331.Ve
1332.PP
1333.Vb 4
1334\& struct ev_timer mytimer;
1335\& ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1336\& ev_timer_again (&mytimer); /* start timer */
1337\& ev_loop (loop, 0);
1338.Ve
1339.PP
1340.Vb 3
1341\& // and in some piece of code that gets executed on any "activity":
1342\& // reset the timeout to start ticking again at 10 seconds
1343\& ev_timer_again (&mytimer);
1344.Ve
705.ie n .Sh """ev_periodic"" \- to cron or not to cron" 1345.ie n .Sh """ev_periodic"" \- to cron or not to cron?"
706.el .Sh "\f(CWev_periodic\fP \- to cron or not to cron" 1346.el .Sh "\f(CWev_periodic\fP \- to cron or not to cron?"
707.IX Subsection "ev_periodic - to cron or not to cron" 1347.IX Subsection "ev_periodic - to cron or not to cron?"
708Periodic watchers are also timers of a kind, but they are very versatile 1348Periodic watchers are also timers of a kind, but they are very versatile
709(and unfortunately a bit complex). 1349(and unfortunately a bit complex).
710.PP 1350.PP
711Unlike \f(CW\*(C`ev_timer\*(C'\fR's, they are not based on real time (or relative time) 1351Unlike \f(CW\*(C`ev_timer\*(C'\fR's, they are not based on real time (or relative time)
712but on wallclock time (absolute time). You can tell a periodic watcher 1352but on wallclock time (absolute time). You can tell a periodic watcher
713to trigger \*(L"at\*(R" some specific point in time. For example, if you tell a 1353to trigger \*(L"at\*(R" some specific point in time. For example, if you tell a
714periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now () 1354periodic watcher to trigger in 10 seconds (by specifiying e.g. \f(CW\*(C`ev_now ()
715+ 10.>) and then reset your system clock to the last year, then it will 1355+ 10.\*(C'\fR) and then reset your system clock to the last year, then it will
716take a year to trigger the event (unlike an \f(CW\*(C`ev_timer\*(C'\fR, which would trigger 1356take a year to trigger the event (unlike an \f(CW\*(C`ev_timer\*(C'\fR, which would trigger
717roughly 10 seconds later and of course not if you reset your system time 1357roughly 10 seconds later).
718again).
719.PP 1358.PP
720They can also be used to implement vastly more complex timers, such as 1359They can also be used to implement vastly more complex timers, such as
721triggering an event on eahc midnight, local time. 1360triggering an event on each midnight, local time or other, complicated,
1361rules.
722.PP 1362.PP
723As with timers, the callback is guarenteed to be invoked only when the 1363As with timers, the callback is guarenteed to be invoked only when the
724time (\f(CW\*(C`at\*(C'\fR) has been passed, but if multiple periodic timers become ready 1364time (\f(CW\*(C`at\*(C'\fR) has been passed, but if multiple periodic timers become ready
725during the same loop iteration then order of execution is undefined. 1365during the same loop iteration then order of execution is undefined.
1366.PP
1367\fIWatcher-Specific Functions and Data Members\fR
1368.IX Subsection "Watcher-Specific Functions and Data Members"
726.IP "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 4 1369.IP "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 4
727.IX Item "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 1370.IX Item "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)"
728.PD 0 1371.PD 0
729.IP "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)" 4 1372.IP "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)" 4
730.IX Item "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)" 1373.IX Item "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)"
731.PD 1374.PD
732Lots of arguments, lets sort it out... There are basically three modes of 1375Lots of arguments, lets sort it out... There are basically three modes of
733operation, and we will explain them from simplest to complex: 1376operation, and we will explain them from simplest to complex:
734.RS 4 1377.RS 4
735.IP "* absolute timer (interval = reschedule_cb = 0)" 4 1378.IP "* absolute timer (at = time, interval = reschedule_cb = 0)" 4
736.IX Item "absolute timer (interval = reschedule_cb = 0)" 1379.IX Item "absolute timer (at = time, interval = reschedule_cb = 0)"
737In this configuration the watcher triggers an event at the wallclock time 1380In this configuration the watcher triggers an event at the wallclock time
738\&\f(CW\*(C`at\*(C'\fR and doesn't repeat. It will not adjust when a time jump occurs, 1381\&\f(CW\*(C`at\*(C'\fR and doesn't repeat. It will not adjust when a time jump occurs,
739that is, if it is to be run at January 1st 2011 then it will run when the 1382that is, if it is to be run at January 1st 2011 then it will run when the
740system time reaches or surpasses this time. 1383system time reaches or surpasses this time.
741.IP "* non-repeating interval timer (interval > 0, reschedule_cb = 0)" 4 1384.IP "* non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)" 4
742.IX Item "non-repeating interval timer (interval > 0, reschedule_cb = 0)" 1385.IX Item "non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)"
743In this mode the watcher will always be scheduled to time out at the next 1386In this mode the watcher will always be scheduled to time out at the next
744\&\f(CW\*(C`at + N * interval\*(C'\fR time (for some integer N) and then repeat, regardless 1387\&\f(CW\*(C`at + N * interval\*(C'\fR time (for some integer N, which can also be negative)
745of any time jumps. 1388and then repeat, regardless of any time jumps.
746.Sp 1389.Sp
747This can be used to create timers that do not drift with respect to system 1390This can be used to create timers that do not drift with respect to system
748time: 1391time:
749.Sp 1392.Sp
750.Vb 1 1393.Vb 1
… …
757by 3600. 1400by 3600.
758.Sp 1401.Sp
759Another way to think about it (for the mathematically inclined) is that 1402Another way to think about it (for the mathematically inclined) is that
760\&\f(CW\*(C`ev_periodic\*(C'\fR will try to run the callback in this mode at the next possible 1403\&\f(CW\*(C`ev_periodic\*(C'\fR will try to run the callback in this mode at the next possible
761time where \f(CW\*(C`time = at (mod interval)\*(C'\fR, regardless of any time jumps. 1404time where \f(CW\*(C`time = at (mod interval)\*(C'\fR, regardless of any time jumps.
1405.Sp
1406For numerical stability it is preferable that the \f(CW\*(C`at\*(C'\fR value is near
1407\&\f(CW\*(C`ev_now ()\*(C'\fR (the current time), but there is no range requirement for
1408this value.
762.IP "* manual reschedule mode (reschedule_cb = callback)" 4 1409.IP "* manual reschedule mode (at and interval ignored, reschedule_cb = callback)" 4
763.IX Item "manual reschedule mode (reschedule_cb = callback)" 1410.IX Item "manual reschedule mode (at and interval ignored, reschedule_cb = callback)"
764In this mode the values for \f(CW\*(C`interval\*(C'\fR and \f(CW\*(C`at\*(C'\fR are both being 1411In this mode the values for \f(CW\*(C`interval\*(C'\fR and \f(CW\*(C`at\*(C'\fR are both being
765ignored. Instead, each time the periodic watcher gets scheduled, the 1412ignored. Instead, each time the periodic watcher gets scheduled, the
766reschedule callback will be called with the watcher as first, and the 1413reschedule callback will be called with the watcher as first, and the
767current time as second argument. 1414current time as second argument.
768.Sp 1415.Sp
769\&\s-1NOTE:\s0 \fIThis callback \s-1MUST\s0 \s-1NOT\s0 stop or destroy any periodic watcher, 1416\&\s-1NOTE:\s0 \fIThis callback \s-1MUST\s0 \s-1NOT\s0 stop or destroy any periodic watcher,
770ever, or make any event loop modifications\fR. If you need to stop it, 1417ever, or make any event loop modifications\fR. If you need to stop it,
771return \f(CW\*(C`now + 1e30\*(C'\fR (or so, fudge fudge) and stop it afterwards (e.g. by 1418return \f(CW\*(C`now + 1e30\*(C'\fR (or so, fudge fudge) and stop it afterwards (e.g. by
772starting a prepare watcher). 1419starting an \f(CW\*(C`ev_prepare\*(C'\fR watcher, which is legal).
773.Sp 1420.Sp
774Its prototype is \f(CW\*(C`ev_tstamp (*reschedule_cb)(struct ev_periodic *w, 1421Its prototype is \f(CW\*(C`ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
775ev_tstamp now)\*(C'\fR, e.g.: 1422ev_tstamp now)\*(C'\fR, e.g.:
776.Sp 1423.Sp
777.Vb 4 1424.Vb 4
… …
801.IX Item "ev_periodic_again (loop, ev_periodic *)" 1448.IX Item "ev_periodic_again (loop, ev_periodic *)"
802Simply stops and restarts the periodic watcher again. This is only useful 1449Simply stops and restarts the periodic watcher again. This is only useful
803when you changed some parameters or the reschedule callback would return 1450when you changed some parameters or the reschedule callback would return
804a different time than the last time it was called (e.g. in a crond like 1451a different time than the last time it was called (e.g. in a crond like
805program when the crontabs have changed). 1452program when the crontabs have changed).
1453.IP "ev_tstamp offset [read\-write]" 4
1454.IX Item "ev_tstamp offset [read-write]"
1455When repeating, this contains the offset value, otherwise this is the
1456absolute point in time (the \f(CW\*(C`at\*(C'\fR value passed to \f(CW\*(C`ev_periodic_set\*(C'\fR).
1457.Sp
1458Can be modified any time, but changes only take effect when the periodic
1459timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being called.
1460.IP "ev_tstamp interval [read\-write]" 4
1461.IX Item "ev_tstamp interval [read-write]"
1462The current interval value. Can be modified any time, but changes only
1463take effect when the periodic timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being
1464called.
1465.IP "ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read\-write]" 4
1466.IX Item "ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]"
1467The current reschedule callback, or \f(CW0\fR, if this functionality is
1468switched off. Can be changed any time, but changes only take effect when
1469the periodic timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being called.
1470.IP "ev_tstamp at [read\-only]" 4
1471.IX Item "ev_tstamp at [read-only]"
1472When active, contains the absolute time that the watcher is supposed to
1473trigger next.
1474.PP
1475Example: Call a callback every hour, or, more precisely, whenever the
1476system clock is divisible by 3600. The callback invocation times have
1477potentially a lot of jittering, but good long-term stability.
1478.PP
1479.Vb 5
1480\& static void
1481\& clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1482\& {
1483\& ... its now a full hour (UTC, or TAI or whatever your clock follows)
1484\& }
1485.Ve
1486.PP
1487.Vb 3
1488\& struct ev_periodic hourly_tick;
1489\& ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1490\& ev_periodic_start (loop, &hourly_tick);
1491.Ve
1492.PP
1493Example: The same as above, but use a reschedule callback to do it:
1494.PP
1495.Vb 1
1496\& #include <math.h>
1497.Ve
1498.PP
1499.Vb 5
1500\& static ev_tstamp
1501\& my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1502\& {
1503\& return fmod (now, 3600.) + 3600.;
1504\& }
1505.Ve
1506.PP
1507.Vb 1
1508\& ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1509.Ve
1510.PP
1511Example: Call a callback every hour, starting now:
1512.PP
1513.Vb 4
1514\& struct ev_periodic hourly_tick;
1515\& ev_periodic_init (&hourly_tick, clock_cb,
1516\& fmod (ev_now (loop), 3600.), 3600., 0);
1517\& ev_periodic_start (loop, &hourly_tick);
1518.Ve
806.ie n .Sh """ev_signal"" \- signal me when a signal gets signalled" 1519.ie n .Sh """ev_signal"" \- signal me when a signal gets signalled!"
807.el .Sh "\f(CWev_signal\fP \- signal me when a signal gets signalled" 1520.el .Sh "\f(CWev_signal\fP \- signal me when a signal gets signalled!"
808.IX Subsection "ev_signal - signal me when a signal gets signalled" 1521.IX Subsection "ev_signal - signal me when a signal gets signalled!"
809Signal watchers will trigger an event when the process receives a specific 1522Signal watchers will trigger an event when the process receives a specific
810signal one or more times. Even though signals are very asynchronous, libev 1523signal one or more times. Even though signals are very asynchronous, libev
811will try it's best to deliver signals synchronously, i.e. as part of the 1524will try it's best to deliver signals synchronously, i.e. as part of the
812normal event processing, like any other event. 1525normal event processing, like any other event.
813.PP 1526.PP
… …
815first watcher gets started will libev actually register a signal watcher 1528first watcher gets started will libev actually register a signal watcher
816with the kernel (thus it coexists with your own signal handlers as long 1529with the kernel (thus it coexists with your own signal handlers as long
817as you don't register any with libev). Similarly, when the last signal 1530as you don't register any with libev). Similarly, when the last signal
818watcher for a signal is stopped libev will reset the signal handler to 1531watcher for a signal is stopped libev will reset the signal handler to
819\&\s-1SIG_DFL\s0 (regardless of what it was set to before). 1532\&\s-1SIG_DFL\s0 (regardless of what it was set to before).
1533.PP
1534\fIWatcher-Specific Functions and Data Members\fR
1535.IX Subsection "Watcher-Specific Functions and Data Members"
820.IP "ev_signal_init (ev_signal *, callback, int signum)" 4 1536.IP "ev_signal_init (ev_signal *, callback, int signum)" 4
821.IX Item "ev_signal_init (ev_signal *, callback, int signum)" 1537.IX Item "ev_signal_init (ev_signal *, callback, int signum)"
822.PD 0 1538.PD 0
823.IP "ev_signal_set (ev_signal *, int signum)" 4 1539.IP "ev_signal_set (ev_signal *, int signum)" 4
824.IX Item "ev_signal_set (ev_signal *, int signum)" 1540.IX Item "ev_signal_set (ev_signal *, int signum)"
825.PD 1541.PD
826Configures the watcher to trigger on the given signal number (usually one 1542Configures the watcher to trigger on the given signal number (usually one
827of the \f(CW\*(C`SIGxxx\*(C'\fR constants). 1543of the \f(CW\*(C`SIGxxx\*(C'\fR constants).
1544.IP "int signum [read\-only]" 4
1545.IX Item "int signum [read-only]"
1546The signal the watcher watches out for.
828.ie n .Sh """ev_child"" \- wait for pid status changes" 1547.ie n .Sh """ev_child"" \- watch out for process status changes"
829.el .Sh "\f(CWev_child\fP \- wait for pid status changes" 1548.el .Sh "\f(CWev_child\fP \- watch out for process status changes"
830.IX Subsection "ev_child - wait for pid status changes" 1549.IX Subsection "ev_child - watch out for process status changes"
831Child watchers trigger when your process receives a \s-1SIGCHLD\s0 in response to 1550Child watchers trigger when your process receives a \s-1SIGCHLD\s0 in response to
832some child status changes (most typically when a child of yours dies). 1551some child status changes (most typically when a child of yours dies).
1552.PP
1553\fIWatcher-Specific Functions and Data Members\fR
1554.IX Subsection "Watcher-Specific Functions and Data Members"
833.IP "ev_child_init (ev_child *, callback, int pid)" 4 1555.IP "ev_child_init (ev_child *, callback, int pid)" 4
834.IX Item "ev_child_init (ev_child *, callback, int pid)" 1556.IX Item "ev_child_init (ev_child *, callback, int pid)"
835.PD 0 1557.PD 0
836.IP "ev_child_set (ev_child *, int pid)" 4 1558.IP "ev_child_set (ev_child *, int pid)" 4
837.IX Item "ev_child_set (ev_child *, int pid)" 1559.IX Item "ev_child_set (ev_child *, int pid)"
… …
840\&\fIany\fR process if \f(CW\*(C`pid\*(C'\fR is specified as \f(CW0\fR). The callback can look 1562\&\fIany\fR process if \f(CW\*(C`pid\*(C'\fR is specified as \f(CW0\fR). The callback can look
841at the \f(CW\*(C`rstatus\*(C'\fR member of the \f(CW\*(C`ev_child\*(C'\fR watcher structure to see 1563at the \f(CW\*(C`rstatus\*(C'\fR member of the \f(CW\*(C`ev_child\*(C'\fR watcher structure to see
842the status word (use the macros from \f(CW\*(C`sys/wait.h\*(C'\fR and see your systems 1564the status word (use the macros from \f(CW\*(C`sys/wait.h\*(C'\fR and see your systems
843\&\f(CW\*(C`waitpid\*(C'\fR documentation). The \f(CW\*(C`rpid\*(C'\fR member contains the pid of the 1565\&\f(CW\*(C`waitpid\*(C'\fR documentation). The \f(CW\*(C`rpid\*(C'\fR member contains the pid of the
844process causing the status change. 1566process causing the status change.
1567.IP "int pid [read\-only]" 4
1568.IX Item "int pid [read-only]"
1569The process id this watcher watches out for, or \f(CW0\fR, meaning any process id.
1570.IP "int rpid [read\-write]" 4
1571.IX Item "int rpid [read-write]"
1572The process id that detected a status change.
1573.IP "int rstatus [read\-write]" 4
1574.IX Item "int rstatus [read-write]"
1575The process exit/trace status caused by \f(CW\*(C`rpid\*(C'\fR (see your systems
1576\&\f(CW\*(C`waitpid\*(C'\fR and \f(CW\*(C`sys/wait.h\*(C'\fR documentation for details).
1577.PP
1578Example: Try to exit cleanly on \s-1SIGINT\s0 and \s-1SIGTERM\s0.
1579.PP
1580.Vb 5
1581\& static void
1582\& sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1583\& {
1584\& ev_unloop (loop, EVUNLOOP_ALL);
1585\& }
1586.Ve
1587.PP
1588.Vb 3
1589\& struct ev_signal signal_watcher;
1590\& ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1591\& ev_signal_start (loop, &sigint_cb);
1592.Ve
1593.ie n .Sh """ev_stat"" \- did the file attributes just change?"
1594.el .Sh "\f(CWev_stat\fP \- did the file attributes just change?"
1595.IX Subsection "ev_stat - did the file attributes just change?"
1596This watches a filesystem path for attribute changes. That is, it calls
1597\&\f(CW\*(C`stat\*(C'\fR regularly (or when the \s-1OS\s0 says it changed) and sees if it changed
1598compared to the last time, invoking the callback if it did.
1599.PP
1600The path does not need to exist: changing from \*(L"path exists\*(R" to \*(L"path does
1601not exist\*(R" is a status change like any other. The condition \*(L"path does
1602not exist\*(R" is signified by the \f(CW\*(C`st_nlink\*(C'\fR field being zero (which is
1603otherwise always forced to be at least one) and all the other fields of
1604the stat buffer having unspecified contents.
1605.PP
1606The path \fIshould\fR be absolute and \fImust not\fR end in a slash. If it is
1607relative and your working directory changes, the behaviour is undefined.
1608.PP
1609Since there is no standard to do this, the portable implementation simply
1610calls \f(CW\*(C`stat (2)\*(C'\fR regularly on the path to see if it changed somehow. You
1611can specify a recommended polling interval for this case. If you specify
1612a polling interval of \f(CW0\fR (highly recommended!) then a \fIsuitable,
1613unspecified default\fR value will be used (which you can expect to be around
1614five seconds, although this might change dynamically). Libev will also
1615impose a minimum interval which is currently around \f(CW0.1\fR, but thats
1616usually overkill.
1617.PP
1618This watcher type is not meant for massive numbers of stat watchers,
1619as even with OS-supported change notifications, this can be
1620resource\-intensive.
1621.PP
1622At the time of this writing, only the Linux inotify interface is
1623implemented (implementing kqueue support is left as an exercise for the
1624reader). Inotify will be used to give hints only and should not change the
1625semantics of \f(CW\*(C`ev_stat\*(C'\fR watchers, which means that libev sometimes needs
1626to fall back to regular polling again even with inotify, but changes are
1627usually detected immediately, and if the file exists there will be no
1628polling.
1629.PP
1630\fIWatcher-Specific Functions and Data Members\fR
1631.IX Subsection "Watcher-Specific Functions and Data Members"
1632.IP "ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)" 4
1633.IX Item "ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)"
1634.PD 0
1635.IP "ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)" 4
1636.IX Item "ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)"
1637.PD
1638Configures the watcher to wait for status changes of the given
1639\&\f(CW\*(C`path\*(C'\fR. The \f(CW\*(C`interval\*(C'\fR is a hint on how quickly a change is expected to
1640be detected and should normally be specified as \f(CW0\fR to let libev choose
1641a suitable value. The memory pointed to by \f(CW\*(C`path\*(C'\fR must point to the same
1642path for as long as the watcher is active.
1643.Sp
1644The callback will be receive \f(CW\*(C`EV_STAT\*(C'\fR when a change was detected,
1645relative to the attributes at the time the watcher was started (or the
1646last change was detected).
1647.IP "ev_stat_stat (ev_stat *)" 4
1648.IX Item "ev_stat_stat (ev_stat *)"
1649Updates the stat buffer immediately with new values. If you change the
1650watched path in your callback, you could call this fucntion to avoid
1651detecting this change (while introducing a race condition). Can also be
1652useful simply to find out the new values.
1653.IP "ev_statdata attr [read\-only]" 4
1654.IX Item "ev_statdata attr [read-only]"
1655The most-recently detected attributes of the file. Although the type is of
1656\&\f(CW\*(C`ev_statdata\*(C'\fR, this is usually the (or one of the) \f(CW\*(C`struct stat\*(C'\fR types
1657suitable for your system. If the \f(CW\*(C`st_nlink\*(C'\fR member is \f(CW0\fR, then there
1658was some error while \f(CW\*(C`stat\*(C'\fRing the file.
1659.IP "ev_statdata prev [read\-only]" 4
1660.IX Item "ev_statdata prev [read-only]"
1661The previous attributes of the file. The callback gets invoked whenever
1662\&\f(CW\*(C`prev\*(C'\fR != \f(CW\*(C`attr\*(C'\fR.
1663.IP "ev_tstamp interval [read\-only]" 4
1664.IX Item "ev_tstamp interval [read-only]"
1665The specified interval.
1666.IP "const char *path [read\-only]" 4
1667.IX Item "const char *path [read-only]"
1668The filesystem path that is being watched.
1669.PP
1670Example: Watch \f(CW\*(C`/etc/passwd\*(C'\fR for attribute changes.
1671.PP
1672.Vb 15
1673\& static void
1674\& passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1675\& {
1676\& /* /etc/passwd changed in some way */
1677\& if (w->attr.st_nlink)
1678\& {
1679\& printf ("passwd current size %ld\en", (long)w->attr.st_size);
1680\& printf ("passwd current atime %ld\en", (long)w->attr.st_mtime);
1681\& printf ("passwd current mtime %ld\en", (long)w->attr.st_mtime);
1682\& }
1683\& else
1684\& /* you shalt not abuse printf for puts */
1685\& puts ("wow, /etc/passwd is not there, expect problems. "
1686\& "if this is windows, they already arrived\en");
1687\& }
1688.Ve
1689.PP
1690.Vb 2
1691\& ...
1692\& ev_stat passwd;
1693.Ve
1694.PP
1695.Vb 2
1696\& ev_stat_init (&passwd, passwd_cb, "/etc/passwd");
1697\& ev_stat_start (loop, &passwd);
1698.Ve
845.ie n .Sh """ev_idle"" \- when you've got nothing better to do" 1699.ie n .Sh """ev_idle"" \- when you've got nothing better to do..."
846.el .Sh "\f(CWev_idle\fP \- when you've got nothing better to do" 1700.el .Sh "\f(CWev_idle\fP \- when you've got nothing better to do..."
847.IX Subsection "ev_idle - when you've got nothing better to do" 1701.IX Subsection "ev_idle - when you've got nothing better to do..."
848Idle watchers trigger events when there are no other events are pending 1702Idle watchers trigger events when no other events of the same or higher
849(prepare, check and other idle watchers do not count). That is, as long 1703priority are pending (prepare, check and other idle watchers do not
850as your process is busy handling sockets or timeouts (or even signals, 1704count).
851imagine) it will not be triggered. But when your process is idle all idle 1705.PP
852watchers are being called again and again, once per event loop iteration \- 1706That is, as long as your process is busy handling sockets or timeouts
1707(or even signals, imagine) of the same or higher priority it will not be
1708triggered. But when your process is idle (or only lower-priority watchers
1709are pending), the idle watchers are being called once per event loop
853until stopped, that is, or your process receives more events and becomes 1710iteration \- until stopped, that is, or your process receives more events
854busy. 1711and becomes busy again with higher priority stuff.
855.PP 1712.PP
856The most noteworthy effect is that as long as any idle watchers are 1713The most noteworthy effect is that as long as any idle watchers are
857active, the process will not block when waiting for new events. 1714active, the process will not block when waiting for new events.
858.PP 1715.PP
859Apart from keeping your process non-blocking (which is a useful 1716Apart from keeping your process non-blocking (which is a useful
860effect on its own sometimes), idle watchers are a good place to do 1717effect on its own sometimes), idle watchers are a good place to do
861\&\*(L"pseudo\-background processing\*(R", or delay processing stuff to after the 1718\&\*(L"pseudo\-background processing\*(R", or delay processing stuff to after the
862event loop has handled all outstanding events. 1719event loop has handled all outstanding events.
1720.PP
1721\fIWatcher-Specific Functions and Data Members\fR
1722.IX Subsection "Watcher-Specific Functions and Data Members"
863.IP "ev_idle_init (ev_signal *, callback)" 4 1723.IP "ev_idle_init (ev_signal *, callback)" 4
864.IX Item "ev_idle_init (ev_signal *, callback)" 1724.IX Item "ev_idle_init (ev_signal *, callback)"
865Initialises and configures the idle watcher \- it has no parameters of any 1725Initialises and configures the idle watcher \- it has no parameters of any
866kind. There is a \f(CW\*(C`ev_idle_set\*(C'\fR macro, but using it is utterly pointless, 1726kind. There is a \f(CW\*(C`ev_idle_set\*(C'\fR macro, but using it is utterly pointless,
867believe me. 1727believe me.
1728.PP
1729Example: Dynamically allocate an \f(CW\*(C`ev_idle\*(C'\fR watcher, start it, and in the
1730callback, free it. Also, use no error checking, as usual.
1731.PP
1732.Vb 7
1733\& static void
1734\& idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1735\& {
1736\& free (w);
1737\& // now do something you wanted to do when the program has
1738\& // no longer asnything immediate to do.
1739\& }
1740.Ve
1741.PP
1742.Vb 3
1743\& struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1744\& ev_idle_init (idle_watcher, idle_cb);
1745\& ev_idle_start (loop, idle_cb);
1746.Ve
868.ie n .Sh """ev_prepare""\fP and \f(CW""ev_check"" \- customise your event loop" 1747.ie n .Sh """ev_prepare""\fP and \f(CW""ev_check"" \- customise your event loop!"
869.el .Sh "\f(CWev_prepare\fP and \f(CWev_check\fP \- customise your event loop" 1748.el .Sh "\f(CWev_prepare\fP and \f(CWev_check\fP \- customise your event loop!"
870.IX Subsection "ev_prepare and ev_check - customise your event loop" 1749.IX Subsection "ev_prepare and ev_check - customise your event loop!"
871Prepare and check watchers are usually (but not always) used in tandem: 1750Prepare and check watchers are usually (but not always) used in tandem:
872prepare watchers get invoked before the process blocks and check watchers 1751prepare watchers get invoked before the process blocks and check watchers
873afterwards. 1752afterwards.
874.PP 1753.PP
1754You \fImust not\fR call \f(CW\*(C`ev_loop\*(C'\fR or similar functions that enter
1755the current event loop from either \f(CW\*(C`ev_prepare\*(C'\fR or \f(CW\*(C`ev_check\*(C'\fR
1756watchers. Other loops than the current one are fine, however. The
1757rationale behind this is that you do not need to check for recursion in
1758those watchers, i.e. the sequence will always be \f(CW\*(C`ev_prepare\*(C'\fR, blocking,
1759\&\f(CW\*(C`ev_check\*(C'\fR so if you have one watcher of each kind they will always be
1760called in pairs bracketing the blocking call.
1761.PP
875Their main purpose is to integrate other event mechanisms into libev. This 1762Their main purpose is to integrate other event mechanisms into libev and
876could be used, for example, to track variable changes, implement your own 1763their use is somewhat advanced. This could be used, for example, to track
877watchers, integrate net-snmp or a coroutine library and lots more. 1764variable changes, implement your own watchers, integrate net-snmp or a
1765coroutine library and lots more. They are also occasionally useful if
1766you cache some data and want to flush it before blocking (for example,
1767in X programs you might want to do an \f(CW\*(C`XFlush ()\*(C'\fR in an \f(CW\*(C`ev_prepare\*(C'\fR
1768watcher).
878.PP 1769.PP
879This is done by examining in each prepare call which file descriptors need 1770This is done by examining in each prepare call which file descriptors need
880to be watched by the other library, registering \f(CW\*(C`ev_io\*(C'\fR watchers for 1771to be watched by the other library, registering \f(CW\*(C`ev_io\*(C'\fR watchers for
881them and starting an \f(CW\*(C`ev_timer\*(C'\fR watcher for any timeouts (many libraries 1772them and starting an \f(CW\*(C`ev_timer\*(C'\fR watcher for any timeouts (many libraries
882provide just this functionality). Then, in the check watcher you check for 1773provide just this functionality). Then, in the check watcher you check for
… …
891are ready to run (it's actually more complicated: it only runs coroutines 1782are ready to run (it's actually more complicated: it only runs coroutines
892with priority higher than or equal to the event loop and one coroutine 1783with priority higher than or equal to the event loop and one coroutine
893of lower priority, but only once, using idle watchers to keep the event 1784of lower priority, but only once, using idle watchers to keep the event
894loop from blocking if lower-priority coroutines are active, thus mapping 1785loop from blocking if lower-priority coroutines are active, thus mapping
895low-priority coroutines to idle/background tasks). 1786low-priority coroutines to idle/background tasks).
1787.PP
1788It is recommended to give \f(CW\*(C`ev_check\*(C'\fR watchers highest (\f(CW\*(C`EV_MAXPRI\*(C'\fR)
1789priority, to ensure that they are being run before any other watchers
1790after the poll. Also, \f(CW\*(C`ev_check\*(C'\fR watchers (and \f(CW\*(C`ev_prepare\*(C'\fR watchers,
1791too) should not activate (\*(L"feed\*(R") events into libev. While libev fully
1792supports this, they will be called before other \f(CW\*(C`ev_check\*(C'\fR watchers
1793did their job. As \f(CW\*(C`ev_check\*(C'\fR watchers are often used to embed other
1794(non\-libev) event loops those other event loops might be in an unusable
1795state until their \f(CW\*(C`ev_check\*(C'\fR watcher ran (always remind yourself to
1796coexist peacefully with others).
1797.PP
1798\fIWatcher-Specific Functions and Data Members\fR
1799.IX Subsection "Watcher-Specific Functions and Data Members"
896.IP "ev_prepare_init (ev_prepare *, callback)" 4 1800.IP "ev_prepare_init (ev_prepare *, callback)" 4
897.IX Item "ev_prepare_init (ev_prepare *, callback)" 1801.IX Item "ev_prepare_init (ev_prepare *, callback)"
898.PD 0 1802.PD 0
899.IP "ev_check_init (ev_check *, callback)" 4 1803.IP "ev_check_init (ev_check *, callback)" 4
900.IX Item "ev_check_init (ev_check *, callback)" 1804.IX Item "ev_check_init (ev_check *, callback)"
901.PD 1805.PD
902Initialises and configures the prepare or check watcher \- they have no 1806Initialises and configures the prepare or check watcher \- they have no
903parameters of any kind. There are \f(CW\*(C`ev_prepare_set\*(C'\fR and \f(CW\*(C`ev_check_set\*(C'\fR 1807parameters of any kind. There are \f(CW\*(C`ev_prepare_set\*(C'\fR and \f(CW\*(C`ev_check_set\*(C'\fR
904macros, but using them is utterly, utterly and completely pointless. 1808macros, but using them is utterly, utterly and completely pointless.
1809.PP
1810There are a number of principal ways to embed other event loops or modules
1811into libev. Here are some ideas on how to include libadns into libev
1812(there is a Perl module named \f(CW\*(C`EV::ADNS\*(C'\fR that does this, which you could
1813use for an actually working example. Another Perl module named \f(CW\*(C`EV::Glib\*(C'\fR
1814embeds a Glib main context into libev, and finally, \f(CW\*(C`Glib::EV\*(C'\fR embeds \s-1EV\s0
1815into the Glib event loop).
1816.PP
1817Method 1: Add \s-1IO\s0 watchers and a timeout watcher in a prepare handler,
1818and in a check watcher, destroy them and call into libadns. What follows
1819is pseudo-code only of course. This requires you to either use a low
1820priority for the check watcher or use \f(CW\*(C`ev_clear_pending\*(C'\fR explicitly, as
1821the callbacks for the IO/timeout watchers might not have been called yet.
1822.PP
1823.Vb 2
1824\& static ev_io iow [nfd];
1825\& static ev_timer tw;
1826.Ve
1827.PP
1828.Vb 4
1829\& static void
1830\& io_cb (ev_loop *loop, ev_io *w, int revents)
1831\& {
1832\& }
1833.Ve
1834.PP
1835.Vb 8
1836\& // create io watchers for each fd and a timer before blocking
1837\& static void
1838\& adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1839\& {
1840\& int timeout = 3600000;
1841\& struct pollfd fds [nfd];
1842\& // actual code will need to loop here and realloc etc.
1843\& adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1844.Ve
1845.PP
1846.Vb 3
1847\& /* the callback is illegal, but won't be called as we stop during check */
1848\& ev_timer_init (&tw, 0, timeout * 1e-3);
1849\& ev_timer_start (loop, &tw);
1850.Ve
1851.PP
1852.Vb 6
1853\& // create one ev_io per pollfd
1854\& for (int i = 0; i < nfd; ++i)
1855\& {
1856\& ev_io_init (iow + i, io_cb, fds [i].fd,
1857\& ((fds [i].events & POLLIN ? EV_READ : 0)
1858\& | (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1859.Ve
1860.PP
1861.Vb 4
1862\& fds [i].revents = 0;
1863\& ev_io_start (loop, iow + i);
1864\& }
1865\& }
1866.Ve
1867.PP
1868.Vb 5
1869\& // stop all watchers after blocking
1870\& static void
1871\& adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1872\& {
1873\& ev_timer_stop (loop, &tw);
1874.Ve
1875.PP
1876.Vb 8
1877\& for (int i = 0; i < nfd; ++i)
1878\& {
1879\& // set the relevant poll flags
1880\& // could also call adns_processreadable etc. here
1881\& struct pollfd *fd = fds + i;
1882\& int revents = ev_clear_pending (iow + i);
1883\& if (revents & EV_READ ) fd->revents |= fd->events & POLLIN;
1884\& if (revents & EV_WRITE) fd->revents |= fd->events & POLLOUT;
1885.Ve
1886.PP
1887.Vb 3
1888\& // now stop the watcher
1889\& ev_io_stop (loop, iow + i);
1890\& }
1891.Ve
1892.PP
1893.Vb 2
1894\& adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1895\& }
1896.Ve
1897.PP
1898Method 2: This would be just like method 1, but you run \f(CW\*(C`adns_afterpoll\*(C'\fR
1899in the prepare watcher and would dispose of the check watcher.
1900.PP
1901Method 3: If the module to be embedded supports explicit event
1902notification (adns does), you can also make use of the actual watcher
1903callbacks, and only destroy/create the watchers in the prepare watcher.
1904.PP
1905.Vb 5
1906\& static void
1907\& timer_cb (EV_P_ ev_timer *w, int revents)
1908\& {
1909\& adns_state ads = (adns_state)w->data;
1910\& update_now (EV_A);
1911.Ve
1912.PP
1913.Vb 2
1914\& adns_processtimeouts (ads, &tv_now);
1915\& }
1916.Ve
1917.PP
1918.Vb 5
1919\& static void
1920\& io_cb (EV_P_ ev_io *w, int revents)
1921\& {
1922\& adns_state ads = (adns_state)w->data;
1923\& update_now (EV_A);
1924.Ve
1925.PP
1926.Vb 3
1927\& if (revents & EV_READ ) adns_processreadable (ads, w->fd, &tv_now);
1928\& if (revents & EV_WRITE) adns_processwriteable (ads, w->fd, &tv_now);
1929\& }
1930.Ve
1931.PP
1932.Vb 1
1933\& // do not ever call adns_afterpoll
1934.Ve
1935.PP
1936Method 4: Do not use a prepare or check watcher because the module you
1937want to embed is too inflexible to support it. Instead, youc na override
1938their poll function. The drawback with this solution is that the main
1939loop is now no longer controllable by \s-1EV\s0. The \f(CW\*(C`Glib::EV\*(C'\fR module does
1940this.
1941.PP
1942.Vb 4
1943\& static gint
1944\& event_poll_func (GPollFD *fds, guint nfds, gint timeout)
1945\& {
1946\& int got_events = 0;
1947.Ve
1948.PP
1949.Vb 2
1950\& for (n = 0; n < nfds; ++n)
1951\& // create/start io watcher that sets the relevant bits in fds[n] and increment got_events
1952.Ve
1953.PP
1954.Vb 2
1955\& if (timeout >= 0)
1956\& // create/start timer
1957.Ve
1958.PP
1959.Vb 2
1960\& // poll
1961\& ev_loop (EV_A_ 0);
1962.Ve
1963.PP
1964.Vb 3
1965\& // stop timer again
1966\& if (timeout >= 0)
1967\& ev_timer_stop (EV_A_ &to);
1968.Ve
1969.PP
1970.Vb 3
1971\& // stop io watchers again - their callbacks should have set
1972\& for (n = 0; n < nfds; ++n)
1973\& ev_io_stop (EV_A_ iow [n]);
1974.Ve
1975.PP
1976.Vb 2
1977\& return got_events;
1978\& }
1979.Ve
1980.ie n .Sh """ev_embed"" \- when one backend isn't enough..."
1981.el .Sh "\f(CWev_embed\fP \- when one backend isn't enough..."
1982.IX Subsection "ev_embed - when one backend isn't enough..."
1983This is a rather advanced watcher type that lets you embed one event loop
1984into another (currently only \f(CW\*(C`ev_io\*(C'\fR events are supported in the embedded
1985loop, other types of watchers might be handled in a delayed or incorrect
1986fashion and must not be used).
1987.PP
1988There are primarily two reasons you would want that: work around bugs and
1989prioritise I/O.
1990.PP
1991As an example for a bug workaround, the kqueue backend might only support
1992sockets on some platform, so it is unusable as generic backend, but you
1993still want to make use of it because you have many sockets and it scales
1994so nicely. In this case, you would create a kqueue-based loop and embed it
1995into your default loop (which might use e.g. poll). Overall operation will
1996be a bit slower because first libev has to poll and then call kevent, but
1997at least you can use both at what they are best.
1998.PP
1999As for prioritising I/O: rarely you have the case where some fds have
2000to be watched and handled very quickly (with low latency), and even
2001priorities and idle watchers might have too much overhead. In this case
2002you would put all the high priority stuff in one loop and all the rest in
2003a second one, and embed the second one in the first.
2004.PP
2005As long as the watcher is active, the callback will be invoked every time
2006there might be events pending in the embedded loop. The callback must then
2007call \f(CW\*(C`ev_embed_sweep (mainloop, watcher)\*(C'\fR to make a single sweep and invoke
2008their callbacks (you could also start an idle watcher to give the embedded
2009loop strictly lower priority for example). You can also set the callback
2010to \f(CW0\fR, in which case the embed watcher will automatically execute the
2011embedded loop sweep.
2012.PP
2013As long as the watcher is started it will automatically handle events. The
2014callback will be invoked whenever some events have been handled. You can
2015set the callback to \f(CW0\fR to avoid having to specify one if you are not
2016interested in that.
2017.PP
2018Also, there have not currently been made special provisions for forking:
2019when you fork, you not only have to call \f(CW\*(C`ev_loop_fork\*(C'\fR on both loops,
2020but you will also have to stop and restart any \f(CW\*(C`ev_embed\*(C'\fR watchers
2021yourself.
2022.PP
2023Unfortunately, not all backends are embeddable, only the ones returned by
2024\&\f(CW\*(C`ev_embeddable_backends\*(C'\fR are, which, unfortunately, does not include any
2025portable one.
2026.PP
2027So when you want to use this feature you will always have to be prepared
2028that you cannot get an embeddable loop. The recommended way to get around
2029this is to have a separate variables for your embeddable loop, try to
2030create it, and if that fails, use the normal loop for everything:
2031.PP
2032.Vb 3
2033\& struct ev_loop *loop_hi = ev_default_init (0);
2034\& struct ev_loop *loop_lo = 0;
2035\& struct ev_embed embed;
2036.Ve
2037.PP
2038.Vb 5
2039\& // see if there is a chance of getting one that works
2040\& // (remember that a flags value of 0 means autodetection)
2041\& loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
2042\& ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
2043\& : 0;
2044.Ve
2045.PP
2046.Vb 8
2047\& // if we got one, then embed it, otherwise default to loop_hi
2048\& if (loop_lo)
2049\& {
2050\& ev_embed_init (&embed, 0, loop_lo);
2051\& ev_embed_start (loop_hi, &embed);
2052\& }
2053\& else
2054\& loop_lo = loop_hi;
2055.Ve
2056.PP
2057\fIWatcher-Specific Functions and Data Members\fR
2058.IX Subsection "Watcher-Specific Functions and Data Members"
2059.IP "ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)" 4
2060.IX Item "ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)"
2061.PD 0
2062.IP "ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)" 4
2063.IX Item "ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)"
2064.PD
2065Configures the watcher to embed the given loop, which must be
2066embeddable. If the callback is \f(CW0\fR, then \f(CW\*(C`ev_embed_sweep\*(C'\fR will be
2067invoked automatically, otherwise it is the responsibility of the callback
2068to invoke it (it will continue to be called until the sweep has been done,
2069if you do not want thta, you need to temporarily stop the embed watcher).
2070.IP "ev_embed_sweep (loop, ev_embed *)" 4
2071.IX Item "ev_embed_sweep (loop, ev_embed *)"
2072Make a single, non-blocking sweep over the embedded loop. This works
2073similarly to \f(CW\*(C`ev_loop (embedded_loop, EVLOOP_NONBLOCK)\*(C'\fR, but in the most
2074apropriate way for embedded loops.
2075.IP "struct ev_loop *other [read\-only]" 4
2076.IX Item "struct ev_loop *other [read-only]"
2077The embedded event loop.
2078.ie n .Sh """ev_fork"" \- the audacity to resume the event loop after a fork"
2079.el .Sh "\f(CWev_fork\fP \- the audacity to resume the event loop after a fork"
2080.IX Subsection "ev_fork - the audacity to resume the event loop after a fork"
2081Fork watchers are called when a \f(CW\*(C`fork ()\*(C'\fR was detected (usually because
2082whoever is a good citizen cared to tell libev about it by calling
2083\&\f(CW\*(C`ev_default_fork\*(C'\fR or \f(CW\*(C`ev_loop_fork\*(C'\fR). The invocation is done before the
2084event loop blocks next and before \f(CW\*(C`ev_check\*(C'\fR watchers are being called,
2085and only in the child after the fork. If whoever good citizen calling
2086\&\f(CW\*(C`ev_default_fork\*(C'\fR cheats and calls it in the wrong process, the fork
2087handlers will be invoked, too, of course.
2088.PP
2089\fIWatcher-Specific Functions and Data Members\fR
2090.IX Subsection "Watcher-Specific Functions and Data Members"
2091.IP "ev_fork_init (ev_signal *, callback)" 4
2092.IX Item "ev_fork_init (ev_signal *, callback)"
2093Initialises and configures the fork watcher \- it has no parameters of any
2094kind. There is a \f(CW\*(C`ev_fork_set\*(C'\fR macro, but using it is utterly pointless,
2095believe me.
905.SH "OTHER FUNCTIONS" 2096.SH "OTHER FUNCTIONS"
906.IX Header "OTHER FUNCTIONS" 2097.IX Header "OTHER FUNCTIONS"
907There are some other functions of possible interest. Described. Here. Now. 2098There are some other functions of possible interest. Described. Here. Now.
908.IP "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 4 2099.IP "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 4
909.IX Item "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 2100.IX Item "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)"
… …
938.Ve 2129.Ve
939.Sp 2130.Sp
940.Vb 1 2131.Vb 1
941\& ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0); 2132\& ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
942.Ve 2133.Ve
943.IP "ev_feed_event (loop, watcher, int events)" 4 2134.IP "ev_feed_event (ev_loop *, watcher *, int revents)" 4
944.IX Item "ev_feed_event (loop, watcher, int events)" 2135.IX Item "ev_feed_event (ev_loop *, watcher *, int revents)"
945Feeds the given event set into the event loop, as if the specified event 2136Feeds the given event set into the event loop, as if the specified event
946had happened for the specified watcher (which must be a pointer to an 2137had happened for the specified watcher (which must be a pointer to an
947initialised but not necessarily started event watcher). 2138initialised but not necessarily started event watcher).
948.IP "ev_feed_fd_event (loop, int fd, int revents)" 4 2139.IP "ev_feed_fd_event (ev_loop *, int fd, int revents)" 4
949.IX Item "ev_feed_fd_event (loop, int fd, int revents)" 2140.IX Item "ev_feed_fd_event (ev_loop *, int fd, int revents)"
950Feed an event on the given fd, as if a file descriptor backend detected 2141Feed an event on the given fd, as if a file descriptor backend detected
951the given events it. 2142the given events it.
952.IP "ev_feed_signal_event (loop, int signum)" 4 2143.IP "ev_feed_signal_event (ev_loop *loop, int signum)" 4
953.IX Item "ev_feed_signal_event (loop, int signum)" 2144.IX Item "ev_feed_signal_event (ev_loop *loop, int signum)"
954Feed an event as if the given signal occured (loop must be the default loop!). 2145Feed an event as if the given signal occured (\f(CW\*(C`loop\*(C'\fR must be the default
2146loop!).
955.SH "LIBEVENT EMULATION" 2147.SH "LIBEVENT EMULATION"
956.IX Header "LIBEVENT EMULATION" 2148.IX Header "LIBEVENT EMULATION"
957Libev offers a compatibility emulation layer for libevent. It cannot 2149Libev offers a compatibility emulation layer for libevent. It cannot
958emulate the internals of libevent, so here are some usage hints: 2150emulate the internals of libevent, so here are some usage hints:
959.IP "* Use it by including <event.h>, as usual." 4 2151.IP "* Use it by including <event.h>, as usual." 4
… …
970.IP "* The libev emulation is \fInot\fR \s-1ABI\s0 compatible to libevent, you need to use the libev header file and library." 4 2162.IP "* The libev emulation is \fInot\fR \s-1ABI\s0 compatible to libevent, you need to use the libev header file and library." 4
971.IX Item "The libev emulation is not ABI compatible to libevent, you need to use the libev header file and library." 2163.IX Item "The libev emulation is not ABI compatible to libevent, you need to use the libev header file and library."
972.PD 2164.PD
973.SH "\*(C+ SUPPORT" 2165.SH "\*(C+ SUPPORT"
974.IX Header " SUPPORT" 2166.IX Header " SUPPORT"
975\&\s-1TBD\s0. 2167Libev comes with some simplistic wrapper classes for \*(C+ that mainly allow
2168you to use some convinience methods to start/stop watchers and also change
2169the callback model to a model using method callbacks on objects.
2170.PP
2171To use it,
2172.PP
2173.Vb 1
2174\& #include <ev++.h>
2175.Ve
2176.PP
2177This automatically includes \fIev.h\fR and puts all of its definitions (many
2178of them macros) into the global namespace. All \*(C+ specific things are
2179put into the \f(CW\*(C`ev\*(C'\fR namespace. It should support all the same embedding
2180options as \fIev.h\fR, most notably \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR.
2181.PP
2182Care has been taken to keep the overhead low. The only data member the \*(C+
2183classes add (compared to plain C\-style watchers) is the event loop pointer
2184that the watcher is associated with (or no additional members at all if
2185you disable \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR when embedding libev).
2186.PP
2187Currently, functions, and static and non-static member functions can be
2188used as callbacks. Other types should be easy to add as long as they only
2189need one additional pointer for context. If you need support for other
2190types of functors please contact the author (preferably after implementing
2191it).
2192.PP
2193Here is a list of things available in the \f(CW\*(C`ev\*(C'\fR namespace:
2194.ie n .IP """ev::READ""\fR, \f(CW""ev::WRITE"" etc." 4
2195.el .IP "\f(CWev::READ\fR, \f(CWev::WRITE\fR etc." 4
2196.IX Item "ev::READ, ev::WRITE etc."
2197These are just enum values with the same values as the \f(CW\*(C`EV_READ\*(C'\fR etc.
2198macros from \fIev.h\fR.
2199.ie n .IP """ev::tstamp""\fR, \f(CW""ev::now""" 4
2200.el .IP "\f(CWev::tstamp\fR, \f(CWev::now\fR" 4
2201.IX Item "ev::tstamp, ev::now"
2202Aliases to the same types/functions as with the \f(CW\*(C`ev_\*(C'\fR prefix.
2203.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
2204.el .IP "\f(CWev::io\fR, \f(CWev::timer\fR, \f(CWev::periodic\fR, \f(CWev::idle\fR, \f(CWev::sig\fR etc." 4
2205.IX Item "ev::io, ev::timer, ev::periodic, ev::idle, ev::sig etc."
2206For each \f(CW\*(C`ev_TYPE\*(C'\fR watcher in \fIev.h\fR there is a corresponding class of
2207the same name in the \f(CW\*(C`ev\*(C'\fR namespace, with the exception of \f(CW\*(C`ev_signal\*(C'\fR
2208which is called \f(CW\*(C`ev::sig\*(C'\fR to avoid clashes with the \f(CW\*(C`signal\*(C'\fR macro
2209defines by many implementations.
2210.Sp
2211All of those classes have these methods:
2212.RS 4
2213.IP "ev::TYPE::TYPE ()" 4
2214.IX Item "ev::TYPE::TYPE ()"
2215.PD 0
2216.IP "ev::TYPE::TYPE (struct ev_loop *)" 4
2217.IX Item "ev::TYPE::TYPE (struct ev_loop *)"
2218.IP "ev::TYPE::~TYPE" 4
2219.IX Item "ev::TYPE::~TYPE"
2220.PD
2221The constructor (optionally) takes an event loop to associate the watcher
2222with. If it is omitted, it will use \f(CW\*(C`EV_DEFAULT\*(C'\fR.
2223.Sp
2224The constructor calls \f(CW\*(C`ev_init\*(C'\fR for you, which means you have to call the
2225\&\f(CW\*(C`set\*(C'\fR method before starting it.
2226.Sp
2227It will not set a callback, however: You have to call the templated \f(CW\*(C`set\*(C'\fR
2228method to set a callback before you can start the watcher.
2229.Sp
2230(The reason why you have to use a method is a limitation in \*(C+ which does
2231not allow explicit template arguments for constructors).
2232.Sp
2233The destructor automatically stops the watcher if it is active.
2234.IP "w\->set<class, &class::method> (object *)" 4
2235.IX Item "w->set<class, &class::method> (object *)"
2236This method sets the callback method to call. The method has to have a
2237signature of \f(CW\*(C`void (*)(ev_TYPE &, int)\*(C'\fR, it receives the watcher as
2238first argument and the \f(CW\*(C`revents\*(C'\fR as second. The object must be given as
2239parameter and is stored in the \f(CW\*(C`data\*(C'\fR member of the watcher.
2240.Sp
2241This method synthesizes efficient thunking code to call your method from
2242the C callback that libev requires. If your compiler can inline your
2243callback (i.e. it is visible to it at the place of the \f(CW\*(C`set\*(C'\fR call and
2244your compiler is good :), then the method will be fully inlined into the
2245thunking function, making it as fast as a direct C callback.
2246.Sp
2247Example: simple class declaration and watcher initialisation
2248.Sp
2249.Vb 4
2250\& struct myclass
2251\& {
2252\& void io_cb (ev::io &w, int revents) { }
2253\& }
2254.Ve
2255.Sp
2256.Vb 3
2257\& myclass obj;
2258\& ev::io iow;
2259\& iow.set <myclass, &myclass::io_cb> (&obj);
2260.Ve
2261.IP "w\->set<function> (void *data = 0)" 4
2262.IX Item "w->set<function> (void *data = 0)"
2263Also sets a callback, but uses a static method or plain function as
2264callback. The optional \f(CW\*(C`data\*(C'\fR argument will be stored in the watcher's
2265\&\f(CW\*(C`data\*(C'\fR member and is free for you to use.
2266.Sp
2267The prototype of the \f(CW\*(C`function\*(C'\fR must be \f(CW\*(C`void (*)(ev::TYPE &w, int)\*(C'\fR.
2268.Sp
2269See the method\-\f(CW\*(C`set\*(C'\fR above for more details.
2270.Sp
2271Example:
2272.Sp
2273.Vb 2
2274\& static void io_cb (ev::io &w, int revents) { }
2275\& iow.set <io_cb> ();
2276.Ve
2277.IP "w\->set (struct ev_loop *)" 4
2278.IX Item "w->set (struct ev_loop *)"
2279Associates a different \f(CW\*(C`struct ev_loop\*(C'\fR with this watcher. You can only
2280do this when the watcher is inactive (and not pending either).
2281.IP "w\->set ([args])" 4
2282.IX Item "w->set ([args])"
2283Basically the same as \f(CW\*(C`ev_TYPE_set\*(C'\fR, with the same args. Must be
2284called at least once. Unlike the C counterpart, an active watcher gets
2285automatically stopped and restarted when reconfiguring it with this
2286method.
2287.IP "w\->start ()" 4
2288.IX Item "w->start ()"
2289Starts the watcher. Note that there is no \f(CW\*(C`loop\*(C'\fR argument, as the
2290constructor already stores the event loop.
2291.IP "w\->stop ()" 4
2292.IX Item "w->stop ()"
2293Stops the watcher if it is active. Again, no \f(CW\*(C`loop\*(C'\fR argument.
2294.ie n .IP "w\->again () (""ev::timer""\fR, \f(CW""ev::periodic"" only)" 4
2295.el .IP "w\->again () (\f(CWev::timer\fR, \f(CWev::periodic\fR only)" 4
2296.IX Item "w->again () (ev::timer, ev::periodic only)"
2297For \f(CW\*(C`ev::timer\*(C'\fR and \f(CW\*(C`ev::periodic\*(C'\fR, this invokes the corresponding
2298\&\f(CW\*(C`ev_TYPE_again\*(C'\fR function.
2299.ie n .IP "w\->sweep () (""ev::embed"" only)" 4
2300.el .IP "w\->sweep () (\f(CWev::embed\fR only)" 4
2301.IX Item "w->sweep () (ev::embed only)"
2302Invokes \f(CW\*(C`ev_embed_sweep\*(C'\fR.
2303.ie n .IP "w\->update () (""ev::stat"" only)" 4
2304.el .IP "w\->update () (\f(CWev::stat\fR only)" 4
2305.IX Item "w->update () (ev::stat only)"
2306Invokes \f(CW\*(C`ev_stat_stat\*(C'\fR.
2307.RE
2308.RS 4
2309.RE
2310.PP
2311Example: Define a class with an \s-1IO\s0 and idle watcher, start one of them in
2312the constructor.
2313.PP
2314.Vb 4
2315\& class myclass
2316\& {
2317\& ev_io io; void io_cb (ev::io &w, int revents);
2318\& ev_idle idle void idle_cb (ev::idle &w, int revents);
2319.Ve
2320.PP
2321.Vb 2
2322\& myclass ();
2323\& }
2324.Ve
2325.PP
2326.Vb 4
2327\& myclass::myclass (int fd)
2328\& {
2329\& io .set <myclass, &myclass::io_cb > (this);
2330\& idle.set <myclass, &myclass::idle_cb> (this);
2331.Ve
2332.PP
2333.Vb 2
2334\& io.start (fd, ev::READ);
2335\& }
2336.Ve
2337.SH "MACRO MAGIC"
2338.IX Header "MACRO MAGIC"
2339Libev can be compiled with a variety of options, the most fundamantal
2340of which is \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR. This option determines whether (most)
2341functions and callbacks have an initial \f(CW\*(C`struct ev_loop *\*(C'\fR argument.
2342.PP
2343To make it easier to write programs that cope with either variant, the
2344following macros are defined:
2345.ie n .IP """EV_A""\fR, \f(CW""EV_A_""" 4
2346.el .IP "\f(CWEV_A\fR, \f(CWEV_A_\fR" 4
2347.IX Item "EV_A, EV_A_"
2348This provides the loop \fIargument\fR for functions, if one is required (\*(L"ev
2349loop argument\*(R"). The \f(CW\*(C`EV_A\*(C'\fR form is used when this is the sole argument,
2350\&\f(CW\*(C`EV_A_\*(C'\fR is used when other arguments are following. Example:
2351.Sp
2352.Vb 3
2353\& ev_unref (EV_A);
2354\& ev_timer_add (EV_A_ watcher);
2355\& ev_loop (EV_A_ 0);
2356.Ve
2357.Sp
2358It assumes the variable \f(CW\*(C`loop\*(C'\fR of type \f(CW\*(C`struct ev_loop *\*(C'\fR is in scope,
2359which is often provided by the following macro.
2360.ie n .IP """EV_P""\fR, \f(CW""EV_P_""" 4
2361.el .IP "\f(CWEV_P\fR, \f(CWEV_P_\fR" 4
2362.IX Item "EV_P, EV_P_"
2363This provides the loop \fIparameter\fR for functions, if one is required (\*(L"ev
2364loop parameter\*(R"). The \f(CW\*(C`EV_P\*(C'\fR form is used when this is the sole parameter,
2365\&\f(CW\*(C`EV_P_\*(C'\fR is used when other parameters are following. Example:
2366.Sp
2367.Vb 2
2368\& // this is how ev_unref is being declared
2369\& static void ev_unref (EV_P);
2370.Ve
2371.Sp
2372.Vb 2
2373\& // this is how you can declare your typical callback
2374\& static void cb (EV_P_ ev_timer *w, int revents)
2375.Ve
2376.Sp
2377It declares a parameter \f(CW\*(C`loop\*(C'\fR of type \f(CW\*(C`struct ev_loop *\*(C'\fR, quite
2378suitable for use with \f(CW\*(C`EV_A\*(C'\fR.
2379.ie n .IP """EV_DEFAULT""\fR, \f(CW""EV_DEFAULT_""" 4
2380.el .IP "\f(CWEV_DEFAULT\fR, \f(CWEV_DEFAULT_\fR" 4
2381.IX Item "EV_DEFAULT, EV_DEFAULT_"
2382Similar to the other two macros, this gives you the value of the default
2383loop, if multiple loops are supported (\*(L"ev loop default\*(R").
2384.PP
2385Example: Declare and initialise a check watcher, utilising the above
2386macros so it will work regardless of whether multiple loops are supported
2387or not.
2388.PP
2389.Vb 5
2390\& static void
2391\& check_cb (EV_P_ ev_timer *w, int revents)
2392\& {
2393\& ev_check_stop (EV_A_ w);
2394\& }
2395.Ve
2396.PP
2397.Vb 4
2398\& ev_check check;
2399\& ev_check_init (&check, check_cb);
2400\& ev_check_start (EV_DEFAULT_ &check);
2401\& ev_loop (EV_DEFAULT_ 0);
2402.Ve
2403.SH "EMBEDDING"
2404.IX Header "EMBEDDING"
2405Libev can (and often is) directly embedded into host
2406applications. Examples of applications that embed it include the Deliantra
2407Game Server, the \s-1EV\s0 perl module, the \s-1GNU\s0 Virtual Private Ethernet (gvpe)
2408and rxvt\-unicode.
2409.PP
2410The goal is to enable you to just copy the necessary files into your
2411source directory without having to change even a single line in them, so
2412you can easily upgrade by simply copying (or having a checked-out copy of
2413libev somewhere in your source tree).
2414.Sh "\s-1FILESETS\s0"
2415.IX Subsection "FILESETS"
2416Depending on what features you need you need to include one or more sets of files
2417in your app.
2418.PP
2419\fI\s-1CORE\s0 \s-1EVENT\s0 \s-1LOOP\s0\fR
2420.IX Subsection "CORE EVENT LOOP"
2421.PP
2422To include only the libev core (all the \f(CW\*(C`ev_*\*(C'\fR functions), with manual
2423configuration (no autoconf):
2424.PP
2425.Vb 2
2426\& #define EV_STANDALONE 1
2427\& #include "ev.c"
2428.Ve
2429.PP
2430This will automatically include \fIev.h\fR, too, and should be done in a
2431single C source file only to provide the function implementations. To use
2432it, do the same for \fIev.h\fR in all files wishing to use this \s-1API\s0 (best
2433done by writing a wrapper around \fIev.h\fR that you can include instead and
2434where you can put other configuration options):
2435.PP
2436.Vb 2
2437\& #define EV_STANDALONE 1
2438\& #include "ev.h"
2439.Ve
2440.PP
2441Both header files and implementation files can be compiled with a \*(C+
2442compiler (at least, thats a stated goal, and breakage will be treated
2443as a bug).
2444.PP
2445You need the following files in your source tree, or in a directory
2446in your include path (e.g. in libev/ when using \-Ilibev):
2447.PP
2448.Vb 4
2449\& ev.h
2450\& ev.c
2451\& ev_vars.h
2452\& ev_wrap.h
2453.Ve
2454.PP
2455.Vb 1
2456\& ev_win32.c required on win32 platforms only
2457.Ve
2458.PP
2459.Vb 5
2460\& ev_select.c only when select backend is enabled (which is enabled by default)
2461\& ev_poll.c only when poll backend is enabled (disabled by default)
2462\& ev_epoll.c only when the epoll backend is enabled (disabled by default)
2463\& ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2464\& ev_port.c only when the solaris port backend is enabled (disabled by default)
2465.Ve
2466.PP
2467\&\fIev.c\fR includes the backend files directly when enabled, so you only need
2468to compile this single file.
2469.PP
2470\fI\s-1LIBEVENT\s0 \s-1COMPATIBILITY\s0 \s-1API\s0\fR
2471.IX Subsection "LIBEVENT COMPATIBILITY API"
2472.PP
2473To include the libevent compatibility \s-1API\s0, also include:
2474.PP
2475.Vb 1
2476\& #include "event.c"
2477.Ve
2478.PP
2479in the file including \fIev.c\fR, and:
2480.PP
2481.Vb 1
2482\& #include "event.h"
2483.Ve
2484.PP
2485in the files that want to use the libevent \s-1API\s0. This also includes \fIev.h\fR.
2486.PP
2487You need the following additional files for this:
2488.PP
2489.Vb 2
2490\& event.h
2491\& event.c
2492.Ve
2493.PP
2494\fI\s-1AUTOCONF\s0 \s-1SUPPORT\s0\fR
2495.IX Subsection "AUTOCONF SUPPORT"
2496.PP
2497Instead of using \f(CW\*(C`EV_STANDALONE=1\*(C'\fR and providing your config in
2498whatever way you want, you can also \f(CW\*(C`m4_include([libev.m4])\*(C'\fR in your
2499\&\fIconfigure.ac\fR and leave \f(CW\*(C`EV_STANDALONE\*(C'\fR undefined. \fIev.c\fR will then
2500include \fIconfig.h\fR and configure itself accordingly.
2501.PP
2502For this of course you need the m4 file:
2503.PP
2504.Vb 1
2505\& libev.m4
2506.Ve
2507.Sh "\s-1PREPROCESSOR\s0 \s-1SYMBOLS/MACROS\s0"
2508.IX Subsection "PREPROCESSOR SYMBOLS/MACROS"
2509Libev can be configured via a variety of preprocessor symbols you have to define
2510before including any of its files. The default is not to build for multiplicity
2511and only include the select backend.
2512.IP "\s-1EV_STANDALONE\s0" 4
2513.IX Item "EV_STANDALONE"
2514Must always be \f(CW1\fR if you do not use autoconf configuration, which
2515keeps libev from including \fIconfig.h\fR, and it also defines dummy
2516implementations for some libevent functions (such as logging, which is not
2517supported). It will also not define any of the structs usually found in
2518\&\fIevent.h\fR that are not directly supported by the libev core alone.
2519.IP "\s-1EV_USE_MONOTONIC\s0" 4
2520.IX Item "EV_USE_MONOTONIC"
2521If defined to be \f(CW1\fR, libev will try to detect the availability of the
2522monotonic clock option at both compiletime and runtime. Otherwise no use
2523of the monotonic clock option will be attempted. If you enable this, you
2524usually have to link against librt or something similar. Enabling it when
2525the functionality isn't available is safe, though, although you have
2526to make sure you link against any libraries where the \f(CW\*(C`clock_gettime\*(C'\fR
2527function is hiding in (often \fI\-lrt\fR).
2528.IP "\s-1EV_USE_REALTIME\s0" 4
2529.IX Item "EV_USE_REALTIME"
2530If defined to be \f(CW1\fR, libev will try to detect the availability of the
2531realtime clock option at compiletime (and assume its availability at
2532runtime if successful). Otherwise no use of the realtime clock option will
2533be attempted. This effectively replaces \f(CW\*(C`gettimeofday\*(C'\fR by \f(CW\*(C`clock_get
2534(CLOCK_REALTIME, ...)\*(C'\fR and will not normally affect correctness. See the
2535note about libraries in the description of \f(CW\*(C`EV_USE_MONOTONIC\*(C'\fR, though.
2536.IP "\s-1EV_USE_NANOSLEEP\s0" 4
2537.IX Item "EV_USE_NANOSLEEP"
2538If defined to be \f(CW1\fR, libev will assume that \f(CW\*(C`nanosleep ()\*(C'\fR is available
2539and will use it for delays. Otherwise it will use \f(CW\*(C`select ()\*(C'\fR.
2540.IP "\s-1EV_USE_SELECT\s0" 4
2541.IX Item "EV_USE_SELECT"
2542If undefined or defined to be \f(CW1\fR, libev will compile in support for the
2543\&\f(CW\*(C`select\*(C'\fR(2) backend. No attempt at autodetection will be done: if no
2544other method takes over, select will be it. Otherwise the select backend
2545will not be compiled in.
2546.IP "\s-1EV_SELECT_USE_FD_SET\s0" 4
2547.IX Item "EV_SELECT_USE_FD_SET"
2548If defined to \f(CW1\fR, then the select backend will use the system \f(CW\*(C`fd_set\*(C'\fR
2549structure. This is useful if libev doesn't compile due to a missing
2550\&\f(CW\*(C`NFDBITS\*(C'\fR or \f(CW\*(C`fd_mask\*(C'\fR definition or it misguesses the bitset layout on
2551exotic systems. This usually limits the range of file descriptors to some
2552low limit such as 1024 or might have other limitations (winsocket only
2553allows 64 sockets). The \f(CW\*(C`FD_SETSIZE\*(C'\fR macro, set before compilation, might
2554influence the size of the \f(CW\*(C`fd_set\*(C'\fR used.
2555.IP "\s-1EV_SELECT_IS_WINSOCKET\s0" 4
2556.IX Item "EV_SELECT_IS_WINSOCKET"
2557When defined to \f(CW1\fR, the select backend will assume that
2558select/socket/connect etc. don't understand file descriptors but
2559wants osf handles on win32 (this is the case when the select to
2560be used is the winsock select). This means that it will call
2561\&\f(CW\*(C`_get_osfhandle\*(C'\fR on the fd to convert it to an \s-1OS\s0 handle. Otherwise,
2562it is assumed that all these functions actually work on fds, even
2563on win32. Should not be defined on non\-win32 platforms.
2564.IP "\s-1EV_USE_POLL\s0" 4
2565.IX Item "EV_USE_POLL"
2566If defined to be \f(CW1\fR, libev will compile in support for the \f(CW\*(C`poll\*(C'\fR(2)
2567backend. Otherwise it will be enabled on non\-win32 platforms. It
2568takes precedence over select.
2569.IP "\s-1EV_USE_EPOLL\s0" 4
2570.IX Item "EV_USE_EPOLL"
2571If defined to be \f(CW1\fR, libev will compile in support for the Linux
2572\&\f(CW\*(C`epoll\*(C'\fR(7) backend. Its availability will be detected at runtime,
2573otherwise another method will be used as fallback. This is the
2574preferred backend for GNU/Linux systems.
2575.IP "\s-1EV_USE_KQUEUE\s0" 4
2576.IX Item "EV_USE_KQUEUE"
2577If defined to be \f(CW1\fR, libev will compile in support for the \s-1BSD\s0 style
2578\&\f(CW\*(C`kqueue\*(C'\fR(2) backend. Its actual availability will be detected at runtime,
2579otherwise another method will be used as fallback. This is the preferred
2580backend for \s-1BSD\s0 and BSD-like systems, although on most BSDs kqueue only
2581supports some types of fds correctly (the only platform we found that
2582supports ptys for example was NetBSD), so kqueue might be compiled in, but
2583not be used unless explicitly requested. The best way to use it is to find
2584out whether kqueue supports your type of fd properly and use an embedded
2585kqueue loop.
2586.IP "\s-1EV_USE_PORT\s0" 4
2587.IX Item "EV_USE_PORT"
2588If defined to be \f(CW1\fR, libev will compile in support for the Solaris
258910 port style backend. Its availability will be detected at runtime,
2590otherwise another method will be used as fallback. This is the preferred
2591backend for Solaris 10 systems.
2592.IP "\s-1EV_USE_DEVPOLL\s0" 4
2593.IX Item "EV_USE_DEVPOLL"
2594reserved for future expansion, works like the \s-1USE\s0 symbols above.
2595.IP "\s-1EV_USE_INOTIFY\s0" 4
2596.IX Item "EV_USE_INOTIFY"
2597If defined to be \f(CW1\fR, libev will compile in support for the Linux inotify
2598interface to speed up \f(CW\*(C`ev_stat\*(C'\fR watchers. Its actual availability will
2599be detected at runtime.
2600.IP "\s-1EV_H\s0" 4
2601.IX Item "EV_H"
2602The name of the \fIev.h\fR header file used to include it. The default if
2603undefined is \f(CW\*(C`<ev.h>\*(C'\fR in \fIevent.h\fR and \f(CW"ev.h"\fR in \fIev.c\fR. This
2604can be used to virtually rename the \fIev.h\fR header file in case of conflicts.
2605.IP "\s-1EV_CONFIG_H\s0" 4
2606.IX Item "EV_CONFIG_H"
2607If \f(CW\*(C`EV_STANDALONE\*(C'\fR isn't \f(CW1\fR, this variable can be used to override
2608\&\fIev.c\fR's idea of where to find the \fIconfig.h\fR file, similarly to
2609\&\f(CW\*(C`EV_H\*(C'\fR, above.
2610.IP "\s-1EV_EVENT_H\s0" 4
2611.IX Item "EV_EVENT_H"
2612Similarly to \f(CW\*(C`EV_H\*(C'\fR, this macro can be used to override \fIevent.c\fR's idea
2613of how the \fIevent.h\fR header can be found.
2614.IP "\s-1EV_PROTOTYPES\s0" 4
2615.IX Item "EV_PROTOTYPES"
2616If defined to be \f(CW0\fR, then \fIev.h\fR will not define any function
2617prototypes, but still define all the structs and other symbols. This is
2618occasionally useful if you want to provide your own wrapper functions
2619around libev functions.
2620.IP "\s-1EV_MULTIPLICITY\s0" 4
2621.IX Item "EV_MULTIPLICITY"
2622If undefined or defined to \f(CW1\fR, then all event-loop-specific functions
2623will have the \f(CW\*(C`struct ev_loop *\*(C'\fR as first argument, and you can create
2624additional independent event loops. Otherwise there will be no support
2625for multiple event loops and there is no first event loop pointer
2626argument. Instead, all functions act on the single default loop.
2627.IP "\s-1EV_MINPRI\s0" 4
2628.IX Item "EV_MINPRI"
2629.PD 0
2630.IP "\s-1EV_MAXPRI\s0" 4
2631.IX Item "EV_MAXPRI"
2632.PD
2633The range of allowed priorities. \f(CW\*(C`EV_MINPRI\*(C'\fR must be smaller or equal to
2634\&\f(CW\*(C`EV_MAXPRI\*(C'\fR, but otherwise there are no non-obvious limitations. You can
2635provide for more priorities by overriding those symbols (usually defined
2636to be \f(CW\*(C`\-2\*(C'\fR and \f(CW2\fR, respectively).
2637.Sp
2638When doing priority-based operations, libev usually has to linearly search
2639all the priorities, so having many of them (hundreds) uses a lot of space
2640and time, so using the defaults of five priorities (\-2 .. +2) is usually
2641fine.
2642.Sp
2643If your embedding app does not need any priorities, defining these both to
2644\&\f(CW0\fR will save some memory and cpu.
2645.IP "\s-1EV_PERIODIC_ENABLE\s0" 4
2646.IX Item "EV_PERIODIC_ENABLE"
2647If undefined or defined to be \f(CW1\fR, then periodic timers are supported. If
2648defined to be \f(CW0\fR, then they are not. Disabling them saves a few kB of
2649code.
2650.IP "\s-1EV_IDLE_ENABLE\s0" 4
2651.IX Item "EV_IDLE_ENABLE"
2652If undefined or defined to be \f(CW1\fR, then idle watchers are supported. If
2653defined to be \f(CW0\fR, then they are not. Disabling them saves a few kB of
2654code.
2655.IP "\s-1EV_EMBED_ENABLE\s0" 4
2656.IX Item "EV_EMBED_ENABLE"
2657If undefined or defined to be \f(CW1\fR, then embed watchers are supported. If
2658defined to be \f(CW0\fR, then they are not.
2659.IP "\s-1EV_STAT_ENABLE\s0" 4
2660.IX Item "EV_STAT_ENABLE"
2661If undefined or defined to be \f(CW1\fR, then stat watchers are supported. If
2662defined to be \f(CW0\fR, then they are not.
2663.IP "\s-1EV_FORK_ENABLE\s0" 4
2664.IX Item "EV_FORK_ENABLE"
2665If undefined or defined to be \f(CW1\fR, then fork watchers are supported. If
2666defined to be \f(CW0\fR, then they are not.
2667.IP "\s-1EV_MINIMAL\s0" 4
2668.IX Item "EV_MINIMAL"
2669If you need to shave off some kilobytes of code at the expense of some
2670speed, define this symbol to \f(CW1\fR. Currently only used for gcc to override
2671some inlining decisions, saves roughly 30% codesize of amd64.
2672.IP "\s-1EV_PID_HASHSIZE\s0" 4
2673.IX Item "EV_PID_HASHSIZE"
2674\&\f(CW\*(C`ev_child\*(C'\fR watchers use a small hash table to distribute workload by
2675pid. The default size is \f(CW16\fR (or \f(CW1\fR with \f(CW\*(C`EV_MINIMAL\*(C'\fR), usually more
2676than enough. If you need to manage thousands of children you might want to
2677increase this value (\fImust\fR be a power of two).
2678.IP "\s-1EV_INOTIFY_HASHSIZE\s0" 4
2679.IX Item "EV_INOTIFY_HASHSIZE"
2680\&\f(CW\*(C`ev_staz\*(C'\fR watchers use a small hash table to distribute workload by
2681inotify watch id. The default size is \f(CW16\fR (or \f(CW1\fR with \f(CW\*(C`EV_MINIMAL\*(C'\fR),
2682usually more than enough. If you need to manage thousands of \f(CW\*(C`ev_stat\*(C'\fR
2683watchers you might want to increase this value (\fImust\fR be a power of
2684two).
2685.IP "\s-1EV_COMMON\s0" 4
2686.IX Item "EV_COMMON"
2687By default, all watchers have a \f(CW\*(C`void *data\*(C'\fR member. By redefining
2688this macro to a something else you can include more and other types of
2689members. You have to define it each time you include one of the files,
2690though, and it must be identical each time.
2691.Sp
2692For example, the perl \s-1EV\s0 module uses something like this:
2693.Sp
2694.Vb 3
2695\& #define EV_COMMON \e
2696\& SV *self; /* contains this struct */ \e
2697\& SV *cb_sv, *fh /* note no trailing ";" */
2698.Ve
2699.IP "\s-1EV_CB_DECLARE\s0 (type)" 4
2700.IX Item "EV_CB_DECLARE (type)"
2701.PD 0
2702.IP "\s-1EV_CB_INVOKE\s0 (watcher, revents)" 4
2703.IX Item "EV_CB_INVOKE (watcher, revents)"
2704.IP "ev_set_cb (ev, cb)" 4
2705.IX Item "ev_set_cb (ev, cb)"
2706.PD
2707Can be used to change the callback member declaration in each watcher,
2708and the way callbacks are invoked and set. Must expand to a struct member
2709definition and a statement, respectively. See the \fIev.h\fR header file for
2710their default definitions. One possible use for overriding these is to
2711avoid the \f(CW\*(C`struct ev_loop *\*(C'\fR as first argument in all cases, or to use
2712method calls instead of plain function calls in \*(C+.
2713.Sh "\s-1EXPORTED\s0 \s-1API\s0 \s-1SYMBOLS\s0"
2714.IX Subsection "EXPORTED API SYMBOLS"
2715If you need to re-export the \s-1API\s0 (e.g. via a dll) and you need a list of
2716exported symbols, you can use the provided \fISymbol.*\fR files which list
2717all public symbols, one per line:
2718.Sp
2719.Vb 2
2720\& Symbols.ev for libev proper
2721\& Symbols.event for the libevent emulation
2722.Ve
2723.Sp
2724This can also be used to rename all public symbols to avoid clashes with
2725multiple versions of libev linked together (which is obviously bad in
2726itself, but sometimes it is inconvinient to avoid this).
2727.Sp
2728A sed command like this will create wrapper \f(CW\*(C`#define\*(C'\fR's that you need to
2729include before including \fIev.h\fR:
2730.Sp
2731.Vb 1
2732\& <Symbols.ev sed -e "s/.*/#define & myprefix_&/" >wrap.h
2733.Ve
2734.Sp
2735This would create a file \fIwrap.h\fR which essentially looks like this:
2736.Sp
2737.Vb 4
2738\& #define ev_backend myprefix_ev_backend
2739\& #define ev_check_start myprefix_ev_check_start
2740\& #define ev_check_stop myprefix_ev_check_stop
2741\& ...
2742.Ve
2743.Sh "\s-1EXAMPLES\s0"
2744.IX Subsection "EXAMPLES"
2745For a real-world example of a program the includes libev
2746verbatim, you can have a look at the \s-1EV\s0 perl module
2747(<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
2748the \fIlibev/\fR subdirectory and includes them in the \fI\s-1EV/EVAPI\s0.h\fR (public
2749interface) and \fI\s-1EV\s0.xs\fR (implementation) files. Only the \fI\s-1EV\s0.xs\fR file
2750will be compiled. It is pretty complex because it provides its own header
2751file.
2752.Sp
2753The usage in rxvt-unicode is simpler. It has a \fIev_cpp.h\fR header file
2754that everybody includes and which overrides some configure choices:
2755.Sp
2756.Vb 9
2757\& #define EV_MINIMAL 1
2758\& #define EV_USE_POLL 0
2759\& #define EV_MULTIPLICITY 0
2760\& #define EV_PERIODIC_ENABLE 0
2761\& #define EV_STAT_ENABLE 0
2762\& #define EV_FORK_ENABLE 0
2763\& #define EV_CONFIG_H <config.h>
2764\& #define EV_MINPRI 0
2765\& #define EV_MAXPRI 0
2766.Ve
2767.Sp
2768.Vb 1
2769\& #include "ev++.h"
2770.Ve
2771.Sp
2772And a \fIev_cpp.C\fR implementation file that contains libev proper and is compiled:
2773.Sp
2774.Vb 2
2775\& #include "ev_cpp.h"
2776\& #include "ev.c"
2777.Ve
2778.SH "COMPLEXITIES"
2779.IX Header "COMPLEXITIES"
2780In this section the complexities of (many of) the algorithms used inside
2781libev will be explained. For complexity discussions about backends see the
2782documentation for \f(CW\*(C`ev_default_init\*(C'\fR.
2783.Sp
2784All of the following are about amortised time: If an array needs to be
2785extended, libev needs to realloc and move the whole array, but this
2786happens asymptotically never with higher number of elements, so O(1) might
2787mean it might do a lengthy realloc operation in rare cases, but on average
2788it is much faster and asymptotically approaches constant time.
2789.RS 4
2790.IP "Starting and stopping timer/periodic watchers: O(log skipped_other_timers)" 4
2791.IX Item "Starting and stopping timer/periodic watchers: O(log skipped_other_timers)"
2792This means that, when you have a watcher that triggers in one hour and
2793there are 100 watchers that would trigger before that then inserting will
2794have to skip those 100 watchers.
2795.IP "Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)" 4
2796.IX Item "Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)"
2797That means that for changing a timer costs less than removing/adding them
2798as only the relative motion in the event queue has to be paid for.
2799.IP "Starting io/check/prepare/idle/signal/child watchers: O(1)" 4
2800.IX Item "Starting io/check/prepare/idle/signal/child watchers: O(1)"
2801These just add the watcher into an array or at the head of a list.
2802=item Stopping check/prepare/idle watchers: O(1)
2803.IP "Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % \s-1EV_PID_HASHSIZE\s0))" 4
2804.IX Item "Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))"
2805These watchers are stored in lists then need to be walked to find the
2806correct watcher to remove. The lists are usually short (you don't usually
2807have many watchers waiting for the same fd or signal).
2808.IP "Finding the next timer per loop iteration: O(1)" 4
2809.IX Item "Finding the next timer per loop iteration: O(1)"
2810.PD 0
2811.IP "Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)" 4
2812.IX Item "Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)"
2813.PD
2814A change means an I/O watcher gets started or stopped, which requires
2815libev to recalculate its status (and possibly tell the kernel).
2816.IP "Activating one watcher: O(1)" 4
2817.IX Item "Activating one watcher: O(1)"
2818.PD 0
2819.IP "Priority handling: O(number_of_priorities)" 4
2820.IX Item "Priority handling: O(number_of_priorities)"
2821.PD
2822Priorities are implemented by allocating some space for each
2823priority. When doing priority-based operations, libev usually has to
2824linearly search all the priorities.
2825.RE
2826.RS 4
976.SH "AUTHOR" 2827.SH "AUTHOR"
977.IX Header "AUTHOR" 2828.IX Header "AUTHOR"
978Marc Lehmann <libev@schmorp.de>. 2829Marc Lehmann <libev@schmorp.de>.

Diff Legend

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