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

Comparing libev/ev.3 (file contents):
Revision 1.1 by root, Tue Nov 13 03:11:57 2007 UTC vs.
Revision 1.60 by root, Mon Jan 28 12:23:02 2008 UTC

1.\" Automatically generated by Pod::Man v1.37, Pod::Parser v1.35 1.\" Automatically generated by Pod::Man 2.16 (Pod::Simple 3.05)
2.\" 2.\"
3.\" Standard preamble: 3.\" Standard preamble:
4.\" ======================================================================== 4.\" ========================================================================
5.de Sh \" Subsection heading 5.de Sh \" Subsection heading
6.br 6.br
… …
23.ft R 23.ft R
24.fi 24.fi
25.. 25..
26.\" Set up some character translations and predefined strings. \*(-- will 26.\" Set up some character translations and predefined strings. \*(-- will
27.\" give an unbreakable dash, \*(PI will give pi, \*(L" will give a left 27.\" give an unbreakable dash, \*(PI will give pi, \*(L" will give a left
28.\" double quote, and \*(R" will give a right double quote. | will give a 28.\" double quote, and \*(R" will give a right double quote. \*(C+ will
29.\" real vertical bar. \*(C+ will give a nicer C++. Capital omega is used to 29.\" give a nicer C++. Capital omega is used to do unbreakable dashes and
30.\" do unbreakable dashes and therefore won't be available. \*(C` and \*(C' 30.\" therefore won't be available. \*(C` and \*(C' expand to `' in nroff,
31.\" expand to `' in nroff, nothing in troff, for use with C<>. 31.\" nothing in troff, for use with C<>.
32.tr \(*W-|\(bv\*(Tr 32.tr \(*W-
33.ds C+ C\v'-.1v'\h'-1p'\s-2+\h'-1p'+\s0\v'.1v'\h'-1p' 33.ds C+ C\v'-.1v'\h'-1p'\s-2+\h'-1p'+\s0\v'.1v'\h'-1p'
34.ie n \{\ 34.ie n \{\
35. ds -- \(*W- 35. ds -- \(*W-
36. ds PI pi 36. ds PI pi
37. if (\n(.H=4u)&(1m=24u) .ds -- \(*W\h'-12u'\(*W\h'-12u'-\" diablo 10 pitch 37. if (\n(.H=4u)&(1m=24u) .ds -- \(*W\h'-12u'\(*W\h'-12u'-\" diablo 10 pitch
… …
46. ds PI \(*p 46. ds PI \(*p
47. ds L" `` 47. ds L" ``
48. ds R" '' 48. ds R" ''
49'br\} 49'br\}
50.\" 50.\"
51.\" Escape single quotes in literal strings from groff's Unicode transform.
52.ie \n(.g .ds Aq \(aq
53.el .ds Aq '
54.\"
51.\" If the F register is turned on, we'll generate index entries on stderr for 55.\" If the F register is turned on, we'll generate index entries on stderr for
52.\" titles (.TH), headers (.SH), subsections (.Sh), items (.Ip), and index 56.\" titles (.TH), headers (.SH), subsections (.Sh), items (.Ip), and index
53.\" entries marked with X<> in POD. Of course, you'll have to process the 57.\" entries marked with X<> in POD. Of course, you'll have to process the
54.\" output yourself in some meaningful fashion. 58.\" output yourself in some meaningful fashion.
55.if \nF \{\ 59.ie \nF \{\
56. de IX 60. de IX
57. tm Index:\\$1\t\\n%\t"\\$2" 61. tm Index:\\$1\t\\n%\t"\\$2"
58.. 62..
59. nr % 0 63. nr % 0
60. rr F 64. rr F
61.\} 65.\}
62.\" 66.el \{\
63.\" For nroff, turn off justification. Always turn off hyphenation; it makes 67. de IX
64.\" way too many mistakes in technical documents. 68..
65.hy 0 69.\}
66.if n .na
67.\" 70.\"
68.\" Accent mark definitions (@(#)ms.acc 1.5 88/02/08 SMI; from UCB 4.2). 71.\" Accent mark definitions (@(#)ms.acc 1.5 88/02/08 SMI; from UCB 4.2).
69.\" Fear. Run. Save yourself. No user-serviceable parts. 72.\" Fear. Run. Save yourself. No user-serviceable parts.
70. \" fudge factors for nroff and troff 73. \" fudge factors for nroff and troff
71.if n \{\ 74.if n \{\
… …
126. ds Ae AE 129. ds Ae AE
127.\} 130.\}
128.rm #[ #] #H #V #F C 131.rm #[ #] #H #V #F C
129.\" ======================================================================== 132.\" ========================================================================
130.\" 133.\"
131.IX Title ""<STANDARD INPUT>" 1" 134.IX Title "EV 1"
132.TH "<STANDARD INPUT>" 1 "2007-11-13" "perl v5.8.8" "User Contributed Perl Documentation" 135.TH EV 1 "2008-01-28" "perl v5.10.0" "User Contributed Perl Documentation"
136.\" For nroff, turn off justification. Always turn off hyphenation; it makes
137.\" way too many mistakes in technical documents.
138.if n .ad l
139.nh
133.SH "NAME" 140.SH "NAME"
134libev \- a high performance full\-featured event loop written in C 141libev \- a high performance full\-featured event loop written in C
135.SH "SYNOPSIS" 142.SH "SYNOPSIS"
136.IX Header "SYNOPSIS" 143.IX Header "SYNOPSIS"
137.Vb 1 144.Vb 1
138\& #include <ev.h> 145\& #include <ev.h>
139.Ve 146.Ve
147.Sh "\s-1EXAMPLE\s0 \s-1PROGRAM\s0"
148.IX Subsection "EXAMPLE PROGRAM"
149.Vb 1
150\& #include <ev.h>
151\&
152\& ev_io stdin_watcher;
153\& ev_timer timeout_watcher;
154\&
155\& /* called when data readable on stdin */
156\& static void
157\& stdin_cb (EV_P_ struct ev_io *w, int revents)
158\& {
159\& /* puts ("stdin ready"); */
160\& ev_io_stop (EV_A_ w); /* just a syntax example */
161\& ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
162\& }
163\&
164\& static void
165\& timeout_cb (EV_P_ struct ev_timer *w, int revents)
166\& {
167\& /* puts ("timeout"); */
168\& ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
169\& }
170\&
171\& int
172\& main (void)
173\& {
174\& struct ev_loop *loop = ev_default_loop (0);
175\&
176\& /* initialise an io watcher, then start it */
177\& ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
178\& ev_io_start (loop, &stdin_watcher);
179\&
180\& /* simple non\-repeating 5.5 second timeout */
181\& ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
182\& ev_timer_start (loop, &timeout_watcher);
183\&
184\& /* loop till timeout or data ready */
185\& ev_loop (loop, 0);
186\&
187\& return 0;
188\& }
189.Ve
140.SH "DESCRIPTION" 190.SH "DESCRIPTION"
141.IX Header "DESCRIPTION" 191.IX Header "DESCRIPTION"
192The newest version of this document is also available as a html-formatted
193web page you might find easier to navigate when reading it for the first
194time: <http://cvs.schmorp.de/libev/ev.html>.
195.PP
142Libev is an event loop: you register interest in certain events (such as a 196Libev 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 197file descriptor being readable or a timeout occurring), and it will manage
144these event sources and provide your program with events. 198these event sources and provide your program with events.
145.PP 199.PP
146To do this, it must take more or less complete control over your process 200To 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 201(or thread) by executing the \fIevent loop\fR handler, and will then
148communicate events via a callback mechanism. 202communicate events via a callback mechanism.
149.PP 203.PP
150You register interest in certain events by registering so-called \fIevent 204You register interest in certain events by registering so-called \fIevent
151watchers\fR, which are relatively small C structures you initialise with the 205watchers\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 206details of the event, and then hand it over to libev by \fIstarting\fR the
153watcher. 207watcher.
154.SH "FEATURES" 208.Sh "\s-1FEATURES\s0"
155.IX Header "FEATURES" 209.IX Subsection "FEATURES"
156Libev supports select, poll, the linux-specific epoll and the bsd-specific 210Libev 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 211BSD-specific \f(CW\*(C`kqueue\*(C'\fR and the Solaris-specific event port mechanisms
158timers with customised rescheduling, signal events, process status change 212for 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 213(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 214with customised rescheduling (\f(CW\*(C`ev_periodic\*(C'\fR), synchronous signals
161fast (see this benchmark comparing 215(\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). 216watchers dealing with the event loop mechanism itself (\f(CW\*(C`ev_idle\*(C'\fR,
217\&\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
218file watchers (\f(CW\*(C`ev_stat\*(C'\fR) and even limited support for fork events
219(\f(CW\*(C`ev_fork\*(C'\fR).
220.PP
221It also is quite fast (see this
222benchmark comparing it to libevent
223for example).
163.SH "CONVENTIONS" 224.Sh "\s-1CONVENTIONS\s0"
164.IX Header "CONVENTIONS" 225.IX Subsection "CONVENTIONS"
165Libev is very configurable. In this manual the default configuration 226Libev is very configurable. In this manual the default configuration will
166will be described, which supports multiple event loops. For more info 227be described, which supports multiple event loops. For more info about
167about various configuration options please have a look at the file 228various 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 229this manual. If libev was configured without support for multiple event
169support for multiple event loops, then all functions taking an initial 230loops, 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) 231(which is always of type \f(CW\*(C`struct ev_loop *\*(C'\fR) will not have this argument.
171will not have this argument. 232.Sh "\s-1TIME\s0 \s-1REPRESENTATION\s0"
172.SH "TIME REPRESENTATION"
173.IX Header "TIME REPRESENTATION" 233.IX Subsection "TIME REPRESENTATION"
174Libev represents time as a single floating point number, representing the 234Libev represents time as a single floating point number, representing the
175(fractional) number of seconds since the (\s-1POSIX\s0) epoch (somewhere near 235(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 236the 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 237called \f(CW\*(C`ev_tstamp\*(C'\fR, which is what you should use too. It usually aliases
178to the double type in C. 238to the \f(CW\*(C`double\*(C'\fR type in C, and when you need to do any calculations on
239it, you should treat it as some floatingpoint value. Unlike the name
240component \f(CW\*(C`stamp\*(C'\fR might indicate, it is also used for time differences
241throughout libev.
179.SH "GLOBAL FUNCTIONS" 242.SH "GLOBAL FUNCTIONS"
180.IX Header "GLOBAL FUNCTIONS" 243.IX Header "GLOBAL FUNCTIONS"
181These functions can be called anytime, even before initialising the 244These functions can be called anytime, even before initialising the
182library in any way. 245library in any way.
183.IP "ev_tstamp ev_time ()" 4 246.IP "ev_tstamp ev_time ()" 4
184.IX Item "ev_tstamp ev_time ()" 247.IX Item "ev_tstamp ev_time ()"
185Returns the current time as libev would use it. 248Returns the current time as libev would use it. Please note that the
249\&\f(CW\*(C`ev_now\*(C'\fR function is usually faster and also often returns the timestamp
250you actually want to know.
251.IP "ev_sleep (ev_tstamp interval)" 4
252.IX Item "ev_sleep (ev_tstamp interval)"
253Sleep for the given interval: The current thread will be blocked until
254either it is interrupted or the given time interval has passed. Basically
255this is a subsecond-resolution \f(CW\*(C`sleep ()\*(C'\fR.
186.IP "int ev_version_major ()" 4 256.IP "int ev_version_major ()" 4
187.IX Item "int ev_version_major ()" 257.IX Item "int ev_version_major ()"
188.PD 0 258.PD 0
189.IP "int ev_version_minor ()" 4 259.IP "int ev_version_minor ()" 4
190.IX Item "int ev_version_minor ()" 260.IX Item "int ev_version_minor ()"
191.PD 261.PD
192You can find out the major and minor version numbers of the library 262You can find out the major and minor \s-1ABI\s0 version numbers of the library
193you linked against by calling the functions \f(CW\*(C`ev_version_major\*(C'\fR and 263you linked against by calling the functions \f(CW\*(C`ev_version_major\*(C'\fR and
194\&\f(CW\*(C`ev_version_minor\*(C'\fR. If you want, you can compare against the global 264\&\f(CW\*(C`ev_version_minor\*(C'\fR. If you want, you can compare against the global
195symbols \f(CW\*(C`EV_VERSION_MAJOR\*(C'\fR and \f(CW\*(C`EV_VERSION_MINOR\*(C'\fR, which specify the 265symbols \f(CW\*(C`EV_VERSION_MAJOR\*(C'\fR and \f(CW\*(C`EV_VERSION_MINOR\*(C'\fR, which specify the
196version of the library your program was compiled against. 266version of the library your program was compiled against.
197.Sp 267.Sp
268These version numbers refer to the \s-1ABI\s0 version of the library, not the
269release version.
270.Sp
198Usually, it's a good idea to terminate if the major versions mismatch, 271Usually, it's a good idea to terminate if the major versions mismatch,
199as this indicates an incompatible change. Minor versions are usually 272as this indicates an incompatible change. Minor versions are usually
200compatible to older versions, so a larger minor version alone is usually 273compatible to older versions, so a larger minor version alone is usually
201not a problem. 274not a problem.
275.Sp
276Example: Make sure we haven't accidentally been linked against the wrong
277version.
278.Sp
279.Vb 3
280\& assert (("libev version mismatch",
281\& ev_version_major () == EV_VERSION_MAJOR
282\& && ev_version_minor () >= EV_VERSION_MINOR));
283.Ve
284.IP "unsigned int ev_supported_backends ()" 4
285.IX Item "unsigned int ev_supported_backends ()"
286Return the set of all backends (i.e. their corresponding \f(CW\*(C`EV_BACKEND_*\*(C'\fR
287value) compiled into this binary of libev (independent of their
288availability on the system you are running on). See \f(CW\*(C`ev_default_loop\*(C'\fR for
289a description of the set values.
290.Sp
291Example: make sure we have the epoll method, because yeah this is cool and
292a must have and can we have a torrent of it please!!!11
293.Sp
294.Vb 2
295\& assert (("sorry, no epoll, no sex",
296\& ev_supported_backends () & EVBACKEND_EPOLL));
297.Ve
298.IP "unsigned int ev_recommended_backends ()" 4
299.IX Item "unsigned int ev_recommended_backends ()"
300Return the set of all backends compiled into this binary of libev and also
301recommended for this platform. This set is often smaller than the one
302returned by \f(CW\*(C`ev_supported_backends\*(C'\fR, as for example kqueue is broken on
303most BSDs and will not be autodetected unless you explicitly request it
304(assuming you know what you are doing). This is the set of backends that
305libev will probe for if you specify no backends explicitly.
306.IP "unsigned int ev_embeddable_backends ()" 4
307.IX Item "unsigned int ev_embeddable_backends ()"
308Returns the set of backends that are embeddable in other event loops. This
309is the theoretical, all-platform, value. To find which backends
310might be supported on the current system, you would need to look at
311\&\f(CW\*(C`ev_embeddable_backends () & ev_supported_backends ()\*(C'\fR, likewise for
312recommended ones.
313.Sp
314See the description of \f(CW\*(C`ev_embed\*(C'\fR watchers for more info.
202.IP "ev_set_allocator (void *(*cb)(void *ptr, long size))" 4 315.IP "ev_set_allocator (void *(*cb)(void *ptr, long size))" 4
203.IX Item "ev_set_allocator (void *(*cb)(void *ptr, long size))" 316.IX Item "ev_set_allocator (void *(*cb)(void *ptr, long size))"
204Sets the allocation function to use (the prototype is similar to the 317Sets the allocation function to use (the prototype is similar \- the
205realloc C function, the semantics are identical). It is used to allocate 318semantics is identical \- to the realloc C function). It is used to
206and free memory (no surprises here). If it returns zero when memory 319allocate and free memory (no surprises here). If it returns zero when
207needs to be allocated, the library might abort or take some potentially 320memory needs to be allocated, the library might abort or take some
208destructive action. The default is your system realloc function. 321potentially destructive action. The default is your system realloc
322function.
209.Sp 323.Sp
210You could override this function in high-availability programs to, say, 324You could override this function in high-availability programs to, say,
211free some memory if it cannot allocate memory, to use a special allocator, 325free some memory if it cannot allocate memory, to use a special allocator,
212or even to sleep a while and retry until some memory is available. 326or even to sleep a while and retry until some memory is available.
327.Sp
328Example: Replace the libev allocator with one that waits a bit and then
329retries).
330.Sp
331.Vb 6
332\& static void *
333\& persistent_realloc (void *ptr, size_t size)
334\& {
335\& for (;;)
336\& {
337\& void *newptr = realloc (ptr, size);
338\&
339\& if (newptr)
340\& return newptr;
341\&
342\& sleep (60);
343\& }
344\& }
345\&
346\& ...
347\& ev_set_allocator (persistent_realloc);
348.Ve
213.IP "ev_set_syserr_cb (void (*cb)(const char *msg));" 4 349.IP "ev_set_syserr_cb (void (*cb)(const char *msg));" 4
214.IX Item "ev_set_syserr_cb (void (*cb)(const char *msg));" 350.IX Item "ev_set_syserr_cb (void (*cb)(const char *msg));"
215Set the callback function to call on a retryable syscall error (such 351Set the callback function to call on a retryable syscall error (such
216as failed select, poll, epoll_wait). The message is a printable string 352as failed select, poll, epoll_wait). The message is a printable string
217indicating the system call or subsystem causing the problem. If this 353indicating the system call or subsystem causing the problem. If this
218callback is set, then libev will expect it to remedy the sitution, no 354callback is set, then libev will expect it to remedy the sitution, no
219matter what, when it returns. That is, libev will generally retry the 355matter what, when it returns. That is, libev will generally retry the
220requested operation, or, if the condition doesn't go away, do bad stuff 356requested operation, or, if the condition doesn't go away, do bad stuff
221(such as abort). 357(such as abort).
358.Sp
359Example: This is basically the same thing that libev does internally, too.
360.Sp
361.Vb 6
362\& static void
363\& fatal_error (const char *msg)
364\& {
365\& perror (msg);
366\& abort ();
367\& }
368\&
369\& ...
370\& ev_set_syserr_cb (fatal_error);
371.Ve
222.SH "FUNCTIONS CONTROLLING THE EVENT LOOP" 372.SH "FUNCTIONS CONTROLLING THE EVENT LOOP"
223.IX Header "FUNCTIONS CONTROLLING THE EVENT LOOP" 373.IX Header "FUNCTIONS CONTROLLING THE EVENT LOOP"
224An event loop is described by a \f(CW\*(C`struct ev_loop *\*(C'\fR. The library knows two 374An event loop is described by a \f(CW\*(C`struct ev_loop *\*(C'\fR. The library knows two
225types of such loops, the \fIdefault\fR loop, which supports signals and child 375types of such loops, the \fIdefault\fR loop, which supports signals and child
226events, and dynamically created loops which do not. 376events, and dynamically created loops which do not.
… …
234.IP "struct ev_loop *ev_default_loop (unsigned int flags)" 4 384.IP "struct ev_loop *ev_default_loop (unsigned int flags)" 4
235.IX Item "struct ev_loop *ev_default_loop (unsigned int flags)" 385.IX Item "struct ev_loop *ev_default_loop (unsigned int flags)"
236This will initialise the default event loop if it hasn't been initialised 386This will initialise the default event loop if it hasn't been initialised
237yet and return it. If the default loop could not be initialised, returns 387yet and return it. If the default loop could not be initialised, returns
238false. If it already was initialised it simply returns it (and ignores the 388false. If it already was initialised it simply returns it (and ignores the
239flags). 389flags. If that is troubling you, check \f(CW\*(C`ev_backend ()\*(C'\fR afterwards).
240.Sp 390.Sp
241If you don't know what event loop to use, use the one returned from this 391If you don't know what event loop to use, use the one returned from this
242function. 392function.
243.Sp 393.Sp
394The default loop is the only loop that can handle \f(CW\*(C`ev_signal\*(C'\fR and
395\&\f(CW\*(C`ev_child\*(C'\fR watchers, and to do this, it always registers a handler
396for \f(CW\*(C`SIGCHLD\*(C'\fR. If this is a problem for your app you can either
397create a dynamic loop with \f(CW\*(C`ev_loop_new\*(C'\fR that doesn't do that, or you
398can simply overwrite the \f(CW\*(C`SIGCHLD\*(C'\fR signal handler \fIafter\fR calling
399\&\f(CW\*(C`ev_default_init\*(C'\fR.
400.Sp
244The flags argument can be used to specify special behaviour or specific 401The flags argument can be used to specify special behaviour or specific
245backends to use, and is usually specified as 0 (or \s-1EVFLAG_AUTO\s0). 402backends to use, and is usually specified as \f(CW0\fR (or \f(CW\*(C`EVFLAG_AUTO\*(C'\fR).
246.Sp 403.Sp
247It supports the following flags: 404The following flags are supported:
248.RS 4 405.RS 4
249.ie n .IP """EVFLAG_AUTO""" 4 406.ie n .IP """EVFLAG_AUTO""" 4
250.el .IP "\f(CWEVFLAG_AUTO\fR" 4 407.el .IP "\f(CWEVFLAG_AUTO\fR" 4
251.IX Item "EVFLAG_AUTO" 408.IX Item "EVFLAG_AUTO"
252The default flags value. Use this if you have no clue (it's the right 409The default flags value. Use this if you have no clue (it's the right
… …
258or setgid) then libev will \fInot\fR look at the environment variable 415or setgid) then libev will \fInot\fR look at the environment variable
259\&\f(CW\*(C`LIBEV_FLAGS\*(C'\fR. Otherwise (the default), this environment variable will 416\&\f(CW\*(C`LIBEV_FLAGS\*(C'\fR. Otherwise (the default), this environment variable will
260override the flags completely if it is found in the environment. This is 417override the flags completely if it is found in the environment. This is
261useful to try out specific backends to test their performance, or to work 418useful to try out specific backends to test their performance, or to work
262around bugs. 419around bugs.
420.ie n .IP """EVFLAG_FORKCHECK""" 4
421.el .IP "\f(CWEVFLAG_FORKCHECK\fR" 4
422.IX Item "EVFLAG_FORKCHECK"
423Instead of calling \f(CW\*(C`ev_default_fork\*(C'\fR or \f(CW\*(C`ev_loop_fork\*(C'\fR manually after
424a fork, you can also make libev check for a fork in each iteration by
425enabling this flag.
426.Sp
427This works by calling \f(CW\*(C`getpid ()\*(C'\fR on every iteration of the loop,
428and thus this might slow down your event loop if you do a lot of loop
429iterations and little real work, but is usually not noticeable (on my
430Linux system for example, \f(CW\*(C`getpid\*(C'\fR is actually a simple 5\-insn sequence
431without a syscall and thus \fIvery\fR fast, but my Linux system also has
432\&\f(CW\*(C`pthread_atfork\*(C'\fR which is even faster).
433.Sp
434The big advantage of this flag is that you can forget about fork (and
435forget about forgetting to tell libev about forking) when you use this
436flag.
437.Sp
438This flag setting cannot be overriden or specified in the \f(CW\*(C`LIBEV_FLAGS\*(C'\fR
439environment variable.
263.ie n .IP """EVMETHOD_SELECT"" (portable select backend)" 4 440.ie n .IP """EVBACKEND_SELECT"" (value 1, portable select backend)" 4
264.el .IP "\f(CWEVMETHOD_SELECT\fR (portable select backend)" 4 441.el .IP "\f(CWEVBACKEND_SELECT\fR (value 1, portable select backend)" 4
265.IX Item "EVMETHOD_SELECT (portable select backend)" 442.IX Item "EVBACKEND_SELECT (value 1, portable select backend)"
266.PD 0 443This is your standard \fIselect\fR\|(2) backend. Not \fIcompletely\fR standard, as
444libev tries to roll its own fd_set with no limits on the number of fds,
445but if that fails, expect a fairly low limit on the number of fds when
446using this backend. It doesn't scale too well (O(highest_fd)), but its
447usually the fastest backend for a low number of (low-numbered :) fds.
448.Sp
449To get good performance out of this backend you need a high amount of
450parallelity (most of the file descriptors should be busy). If you are
451writing a server, you should \f(CW\*(C`accept ()\*(C'\fR in a loop to accept as many
452connections as possible during one iteration. You might also want to have
453a look at \f(CW\*(C`ev_set_io_collect_interval ()\*(C'\fR to increase the amount of
454readyness notifications you get per iteration.
267.ie n .IP """EVMETHOD_POLL"" (poll backend, available everywhere except on windows)" 4 455.ie n .IP """EVBACKEND_POLL"" (value 2, poll backend, available everywhere except on windows)" 4
268.el .IP "\f(CWEVMETHOD_POLL\fR (poll backend, available everywhere except on windows)" 4 456.el .IP "\f(CWEVBACKEND_POLL\fR (value 2, poll backend, available everywhere except on windows)" 4
269.IX Item "EVMETHOD_POLL (poll backend, available everywhere except on windows)" 457.IX Item "EVBACKEND_POLL (value 2, poll backend, available everywhere except on windows)"
458And this is your standard \fIpoll\fR\|(2) backend. It's more complicated
459than select, but handles sparse fds better and has no artificial
460limit on the number of fds you can use (except it will slow down
461considerably with a lot of inactive fds). It scales similarly to select,
462i.e. O(total_fds). See the entry for \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR, above, for
463performance tips.
270.ie n .IP """EVMETHOD_EPOLL"" (linux only)" 4 464.ie n .IP """EVBACKEND_EPOLL"" (value 4, Linux)" 4
271.el .IP "\f(CWEVMETHOD_EPOLL\fR (linux only)" 4 465.el .IP "\f(CWEVBACKEND_EPOLL\fR (value 4, Linux)" 4
272.IX Item "EVMETHOD_EPOLL (linux only)" 466.IX Item "EVBACKEND_EPOLL (value 4, Linux)"
273.ie n .IP """EVMETHOD_KQUEUE"" (some bsds only)" 4 467For few fds, this backend is a bit little slower than poll and select,
274.el .IP "\f(CWEVMETHOD_KQUEUE\fR (some bsds only)" 4 468but it scales phenomenally better. While poll and select usually scale
275.IX Item "EVMETHOD_KQUEUE (some bsds only)" 469like O(total_fds) where n is the total number of fds (or the highest fd),
470epoll scales either O(1) or O(active_fds). The epoll design has a number
471of shortcomings, such as silently dropping events in some hard-to-detect
472cases and rewiring a syscall per fd change, no fork support and bad
473support for dup.
474.Sp
475While stopping, setting and starting an I/O watcher in the same iteration
476will result in some caching, there is still a syscall per such incident
477(because the fd could point to a different file description now), so its
478best to avoid that. Also, \f(CW\*(C`dup ()\*(C'\fR'ed file descriptors might not work
479very well if you register events for both fds.
480.Sp
481Please note that epoll sometimes generates spurious notifications, so you
482need to use non-blocking I/O or other means to avoid blocking when no data
483(or space) is available.
484.Sp
485Best performance from this backend is achieved by not unregistering all
486watchers for a file descriptor until it has been closed, if possible, i.e.
487keep at least one watcher active per fd at all times.
488.Sp
489While nominally embeddeble in other event loops, this feature is broken in
490all kernel versions tested so far.
491.ie n .IP """EVBACKEND_KQUEUE"" (value 8, most \s-1BSD\s0 clones)" 4
492.el .IP "\f(CWEVBACKEND_KQUEUE\fR (value 8, most \s-1BSD\s0 clones)" 4
493.IX Item "EVBACKEND_KQUEUE (value 8, most BSD clones)"
494Kqueue deserves special mention, as at the time of this writing, it
495was broken on all BSDs except NetBSD (usually it doesn't work reliably
496with anything but sockets and pipes, except on Darwin, where of course
497it's completely useless). For this reason it's not being \*(L"autodetected\*(R"
498unless you explicitly specify it explicitly in the flags (i.e. using
499\&\f(CW\*(C`EVBACKEND_KQUEUE\*(C'\fR) or libev was compiled on a known-to-be-good (\-enough)
500system like NetBSD.
501.Sp
502You still can embed kqueue into a normal poll or select backend and use it
503only for sockets (after having made sure that sockets work with kqueue on
504the target platform). See \f(CW\*(C`ev_embed\*(C'\fR watchers for more info.
505.Sp
506It scales in the same way as the epoll backend, but the interface to the
507kernel is more efficient (which says nothing about its actual speed, of
508course). While stopping, setting and starting an I/O watcher does never
509cause an extra syscall as with \f(CW\*(C`EVBACKEND_EPOLL\*(C'\fR, it still adds up to
510two event changes per incident, support for \f(CW\*(C`fork ()\*(C'\fR is very bad and it
511drops fds silently in similarly hard-to-detect cases.
512.Sp
513This backend usually performs well under most conditions.
514.Sp
515While nominally embeddable in other event loops, this doesn't work
516everywhere, so you might need to test for this. And since it is broken
517almost everywhere, you should only use it when you have a lot of sockets
518(for which it usually works), by embedding it into another event loop
519(e.g. \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or \f(CW\*(C`EVBACKEND_POLL\*(C'\fR) and using it only for
520sockets.
276.ie n .IP """EVMETHOD_DEVPOLL"" (solaris 8 only)" 4 521.ie n .IP """EVBACKEND_DEVPOLL"" (value 16, Solaris 8)" 4
277.el .IP "\f(CWEVMETHOD_DEVPOLL\fR (solaris 8 only)" 4 522.el .IP "\f(CWEVBACKEND_DEVPOLL\fR (value 16, Solaris 8)" 4
278.IX Item "EVMETHOD_DEVPOLL (solaris 8 only)" 523.IX Item "EVBACKEND_DEVPOLL (value 16, Solaris 8)"
524This is not implemented yet (and might never be, unless you send me an
525implementation). According to reports, \f(CW\*(C`/dev/poll\*(C'\fR only supports sockets
526and is not embeddable, which would limit the usefulness of this backend
527immensely.
279.ie n .IP """EVMETHOD_PORT"" (solaris 10 only)" 4 528.ie n .IP """EVBACKEND_PORT"" (value 32, Solaris 10)" 4
280.el .IP "\f(CWEVMETHOD_PORT\fR (solaris 10 only)" 4 529.el .IP "\f(CWEVBACKEND_PORT\fR (value 32, Solaris 10)" 4
281.IX Item "EVMETHOD_PORT (solaris 10 only)" 530.IX Item "EVBACKEND_PORT (value 32, Solaris 10)"
282.PD 531This uses the Solaris 10 event port mechanism. As with everything on Solaris,
283If one or more of these are ored into the flags value, then only these 532it's really slow, but it still scales very well (O(active_fds)).
284backends will be tried (in the reverse order as given here). If one are 533.Sp
285specified, any backend will do. 534Please note that solaris event ports can deliver a lot of spurious
535notifications, so you need to use non-blocking I/O or other means to avoid
536blocking when no data (or space) is available.
537.Sp
538While this backend scales well, it requires one system call per active
539file descriptor per loop iteration. For small and medium numbers of file
540descriptors a \*(L"slow\*(R" \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or \f(CW\*(C`EVBACKEND_POLL\*(C'\fR backend
541might perform better.
542.Sp
543On the positive side, ignoring the spurious readyness notifications, this
544backend actually performed to specification in all tests and is fully
545embeddable, which is a rare feat among the OS-specific backends.
546.ie n .IP """EVBACKEND_ALL""" 4
547.el .IP "\f(CWEVBACKEND_ALL\fR" 4
548.IX Item "EVBACKEND_ALL"
549Try all backends (even potentially broken ones that wouldn't be tried
550with \f(CW\*(C`EVFLAG_AUTO\*(C'\fR). Since this is a mask, you can do stuff such as
551\&\f(CW\*(C`EVBACKEND_ALL & ~EVBACKEND_KQUEUE\*(C'\fR.
552.Sp
553It is definitely not recommended to use this flag.
286.RE 554.RE
287.RS 4 555.RS 4
556.Sp
557If one or more of these are ored into the flags value, then only these
558backends will be tried (in the reverse order as listed here). If none are
559specified, all backends in \f(CW\*(C`ev_recommended_backends ()\*(C'\fR will be tried.
560.Sp
561The most typical usage is like this:
562.Sp
563.Vb 2
564\& if (!ev_default_loop (0))
565\& fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
566.Ve
567.Sp
568Restrict libev to the select and poll backends, and do not allow
569environment settings to be taken into account:
570.Sp
571.Vb 1
572\& ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
573.Ve
574.Sp
575Use whatever libev has to offer, but make sure that kqueue is used if
576available (warning, breaks stuff, best use only with your own private
577event loop and only if you know the \s-1OS\s0 supports your types of fds):
578.Sp
579.Vb 1
580\& ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
581.Ve
288.RE 582.RE
289.IP "struct ev_loop *ev_loop_new (unsigned int flags)" 4 583.IP "struct ev_loop *ev_loop_new (unsigned int flags)" 4
290.IX Item "struct ev_loop *ev_loop_new (unsigned int flags)" 584.IX Item "struct ev_loop *ev_loop_new (unsigned int flags)"
291Similar to \f(CW\*(C`ev_default_loop\*(C'\fR, but always creates a new event loop that is 585Similar to \f(CW\*(C`ev_default_loop\*(C'\fR, but always creates a new event loop that is
292always distinct from the default loop. Unlike the default loop, it cannot 586always distinct from the default loop. Unlike the default loop, it cannot
293handle signal and child watchers, and attempts to do so will be greeted by 587handle signal and child watchers, and attempts to do so will be greeted by
294undefined behaviour (or a failed assertion if assertions are enabled). 588undefined behaviour (or a failed assertion if assertions are enabled).
589.Sp
590Example: Try to create a event loop that uses epoll and nothing else.
591.Sp
592.Vb 3
593\& struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
594\& if (!epoller)
595\& fatal ("no epoll found here, maybe it hides under your chair");
596.Ve
295.IP "ev_default_destroy ()" 4 597.IP "ev_default_destroy ()" 4
296.IX Item "ev_default_destroy ()" 598.IX Item "ev_default_destroy ()"
297Destroys the default loop again (frees all memory and kernel state 599Destroys the default loop again (frees all memory and kernel state
298etc.). This stops all registered event watchers (by not touching them in 600etc.). None of the active event watchers will be stopped in the normal
299any way whatsoever, although you cannot rely on this :). 601sense, so e.g. \f(CW\*(C`ev_is_active\*(C'\fR might still return true. It is your
602responsibility to either stop all watchers cleanly yoursef \fIbefore\fR
603calling this function, or cope with the fact afterwards (which is usually
604the easiest thing, you can just ignore the watchers and/or \f(CW\*(C`free ()\*(C'\fR them
605for example).
606.Sp
607Note that certain global state, such as signal state, will not be freed by
608this function, and related watchers (such as signal and child watchers)
609would need to be stopped manually.
610.Sp
611In general it is not advisable to call this function except in the
612rare occasion where you really need to free e.g. the signal handling
613pipe fds. If you need dynamically allocated loops it is better to use
614\&\f(CW\*(C`ev_loop_new\*(C'\fR and \f(CW\*(C`ev_loop_destroy\*(C'\fR).
300.IP "ev_loop_destroy (loop)" 4 615.IP "ev_loop_destroy (loop)" 4
301.IX Item "ev_loop_destroy (loop)" 616.IX Item "ev_loop_destroy (loop)"
302Like \f(CW\*(C`ev_default_destroy\*(C'\fR, but destroys an event loop created by an 617Like \f(CW\*(C`ev_default_destroy\*(C'\fR, but destroys an event loop created by an
303earlier call to \f(CW\*(C`ev_loop_new\*(C'\fR. 618earlier call to \f(CW\*(C`ev_loop_new\*(C'\fR.
304.IP "ev_default_fork ()" 4 619.IP "ev_default_fork ()" 4
305.IX Item "ev_default_fork ()" 620.IX Item "ev_default_fork ()"
621This function sets a flag that causes subsequent \f(CW\*(C`ev_loop\*(C'\fR iterations
306This function reinitialises the kernel state for backends that have 622to reinitialise the kernel state for backends that have one. Despite the
307one. Despite the name, you can call it anytime, but it makes most sense 623name, you can call it anytime, but it makes most sense after forking, in
308after forking, in either the parent or child process (or both, but that 624the child process (or both child and parent, but that again makes little
309again makes little sense). 625sense). You \fImust\fR call it in the child before using any of the libev
626functions, and it will only take effect at the next \f(CW\*(C`ev_loop\*(C'\fR iteration.
310.Sp 627.Sp
311You \fImust\fR call this function after forking if and only if you want to 628On the other hand, you only need to call this function in the child
312use the event library in both processes. If you just fork+exec, you don't 629process if and only if you want to use the event library in the child. If
313have to call it. 630you just fork+exec, you don't have to call it at all.
314.Sp 631.Sp
315The function itself is quite fast and it's usually not a problem to call 632The function itself is quite fast and it's usually not a problem to call
316it just in case after a fork. To make this easy, the function will fit in 633it just in case after a fork. To make this easy, the function will fit in
317quite nicely into a call to \f(CW\*(C`pthread_atfork\*(C'\fR: 634quite nicely into a call to \f(CW\*(C`pthread_atfork\*(C'\fR:
318.Sp 635.Sp
… …
322.IP "ev_loop_fork (loop)" 4 639.IP "ev_loop_fork (loop)" 4
323.IX Item "ev_loop_fork (loop)" 640.IX Item "ev_loop_fork (loop)"
324Like \f(CW\*(C`ev_default_fork\*(C'\fR, but acts on an event loop created by 641Like \f(CW\*(C`ev_default_fork\*(C'\fR, but acts on an event loop created by
325\&\f(CW\*(C`ev_loop_new\*(C'\fR. Yes, you have to call this on every allocated event loop 642\&\f(CW\*(C`ev_loop_new\*(C'\fR. Yes, you have to call this on every allocated event loop
326after fork, and how you do this is entirely your own problem. 643after fork, and how you do this is entirely your own problem.
644.IP "unsigned int ev_loop_count (loop)" 4
645.IX Item "unsigned int ev_loop_count (loop)"
646Returns the count of loop iterations for the loop, which is identical to
647the number of times libev did poll for new events. It starts at \f(CW0\fR and
648happily wraps around with enough iterations.
649.Sp
650This value can sometimes be useful as a generation counter of sorts (it
651\&\*(L"ticks\*(R" the number of loop iterations), as it roughly corresponds with
652\&\f(CW\*(C`ev_prepare\*(C'\fR and \f(CW\*(C`ev_check\*(C'\fR calls.
327.IP "unsigned int ev_method (loop)" 4 653.IP "unsigned int ev_backend (loop)" 4
328.IX Item "unsigned int ev_method (loop)" 654.IX Item "unsigned int ev_backend (loop)"
329Returns one of the \f(CW\*(C`EVMETHOD_*\*(C'\fR flags indicating the event backend in 655Returns one of the \f(CW\*(C`EVBACKEND_*\*(C'\fR flags indicating the event backend in
330use. 656use.
331.IP "ev_tstamp ev_now (loop)" 4 657.IP "ev_tstamp ev_now (loop)" 4
332.IX Item "ev_tstamp ev_now (loop)" 658.IX Item "ev_tstamp ev_now (loop)"
333Returns the current \*(L"event loop time\*(R", which is the time the event loop 659Returns the current \*(L"event loop time\*(R", which is the time the event loop
334got events and started processing them. This timestamp does not change 660received events and started processing them. This timestamp does not
335as long as callbacks are being processed, and this is also the base time 661change as long as callbacks are being processed, and this is also the base
336used for relative timers. You can treat it as the timestamp of the event 662time used for relative timers. You can treat it as the timestamp of the
337occuring (or more correctly, the mainloop finding out about it). 663event occurring (or more correctly, libev finding out about it).
338.IP "ev_loop (loop, int flags)" 4 664.IP "ev_loop (loop, int flags)" 4
339.IX Item "ev_loop (loop, int flags)" 665.IX Item "ev_loop (loop, int flags)"
340Finally, this is it, the event handler. This function usually is called 666Finally, this is it, the event handler. This function usually is called
341after you initialised all your watchers and you want to start handling 667after you initialised all your watchers and you want to start handling
342events. 668events.
343.Sp 669.Sp
344If the flags argument is specified as 0, it will not return until either 670If the flags argument is specified as \f(CW0\fR, it will not return until
345no event watchers are active anymore or \f(CW\*(C`ev_unloop\*(C'\fR was called. 671either no event watchers are active anymore or \f(CW\*(C`ev_unloop\*(C'\fR was called.
672.Sp
673Please note that an explicit \f(CW\*(C`ev_unloop\*(C'\fR is usually better than
674relying on all watchers to be stopped when deciding when a program has
675finished (especially in interactive programs), but having a program that
676automatically loops as long as it has to and no longer by virtue of
677relying on its watchers stopping correctly is a thing of beauty.
346.Sp 678.Sp
347A flags value of \f(CW\*(C`EVLOOP_NONBLOCK\*(C'\fR will look for new events, will handle 679A flags value of \f(CW\*(C`EVLOOP_NONBLOCK\*(C'\fR will look for new events, will handle
348those events and any outstanding ones, but will not block your process in 680those events and any outstanding ones, but will not block your process in
349case there are no events and will return after one iteration of the loop. 681case there are no events and will return after one iteration of the loop.
350.Sp 682.Sp
351A flags value of \f(CW\*(C`EVLOOP_ONESHOT\*(C'\fR will look for new events (waiting if 683A flags value of \f(CW\*(C`EVLOOP_ONESHOT\*(C'\fR will look for new events (waiting if
352neccessary) and will handle those and any outstanding ones. It will block 684neccessary) and will handle those and any outstanding ones. It will block
353your process until at least one new event arrives, and will return after 685your process until at least one new event arrives, and will return after
354one iteration of the loop. 686one iteration of the loop. This is useful if you are waiting for some
687external event in conjunction with something not expressible using other
688libev watchers. However, a pair of \f(CW\*(C`ev_prepare\*(C'\fR/\f(CW\*(C`ev_check\*(C'\fR watchers is
689usually a better approach for this kind of thing.
355.Sp 690.Sp
356This flags value could be used to implement alternative looping 691Here are the gory details of what \f(CW\*(C`ev_loop\*(C'\fR does:
357constructs, but the \f(CW\*(C`prepare\*(C'\fR and \f(CW\*(C`check\*(C'\fR watchers provide a better and 692.Sp
358more generic mechanism. 693.Vb 10
694\& \- Before the first iteration, call any pending watchers.
695\& * If EVFLAG_FORKCHECK was used, check for a fork.
696\& \- If a fork was detected, queue and call all fork watchers.
697\& \- Queue and call all prepare watchers.
698\& \- If we have been forked, recreate the kernel state.
699\& \- Update the kernel state with all outstanding changes.
700\& \- Update the "event loop time".
701\& \- Calculate for how long to sleep or block, if at all
702\& (active idle watchers, EVLOOP_NONBLOCK or not having
703\& any active watchers at all will result in not sleeping).
704\& \- Sleep if the I/O and timer collect interval say so.
705\& \- Block the process, waiting for any events.
706\& \- Queue all outstanding I/O (fd) events.
707\& \- Update the "event loop time" and do time jump handling.
708\& \- Queue all outstanding timers.
709\& \- Queue all outstanding periodics.
710\& \- If no events are pending now, queue all idle watchers.
711\& \- Queue all check watchers.
712\& \- Call all queued watchers in reverse order (i.e. check watchers first).
713\& Signals and child watchers are implemented as I/O watchers, and will
714\& be handled here by queueing them when their watcher gets executed.
715\& \- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
716\& were used, or there are no active watchers, return, otherwise
717\& continue with step *.
718.Ve
719.Sp
720Example: Queue some jobs and then loop until no events are outstanding
721anymore.
722.Sp
723.Vb 4
724\& ... queue jobs here, make sure they register event watchers as long
725\& ... as they still have work to do (even an idle watcher will do..)
726\& ev_loop (my_loop, 0);
727\& ... jobs done. yeah!
728.Ve
359.IP "ev_unloop (loop, how)" 4 729.IP "ev_unloop (loop, how)" 4
360.IX Item "ev_unloop (loop, how)" 730.IX Item "ev_unloop (loop, how)"
361Can be used to make a call to \f(CW\*(C`ev_loop\*(C'\fR return early (but only after it 731Can be used to make a call to \f(CW\*(C`ev_loop\*(C'\fR return early (but only after it
362has processed all outstanding events). The \f(CW\*(C`how\*(C'\fR argument must be either 732has processed all outstanding events). The \f(CW\*(C`how\*(C'\fR argument must be either
363\&\f(CW\*(C`EVUNLOOP_ONE\*(C'\fR, which will make the innermost \f(CW\*(C`ev_loop\*(C'\fR call return, or 733\&\f(CW\*(C`EVUNLOOP_ONE\*(C'\fR, which will make the innermost \f(CW\*(C`ev_loop\*(C'\fR call return, or
364\&\f(CW\*(C`EVUNLOOP_ALL\*(C'\fR, which will make all nested \f(CW\*(C`ev_loop\*(C'\fR calls return. 734\&\f(CW\*(C`EVUNLOOP_ALL\*(C'\fR, which will make all nested \f(CW\*(C`ev_loop\*(C'\fR calls return.
735.Sp
736This \*(L"unloop state\*(R" will be cleared when entering \f(CW\*(C`ev_loop\*(C'\fR again.
365.IP "ev_ref (loop)" 4 737.IP "ev_ref (loop)" 4
366.IX Item "ev_ref (loop)" 738.IX Item "ev_ref (loop)"
367.PD 0 739.PD 0
368.IP "ev_unref (loop)" 4 740.IP "ev_unref (loop)" 4
369.IX Item "ev_unref (loop)" 741.IX Item "ev_unref (loop)"
… …
375returning, \fIev_unref()\fR after starting, and \fIev_ref()\fR before stopping it. For 747returning, \fIev_unref()\fR after starting, and \fIev_ref()\fR before stopping it. For
376example, libev itself uses this for its internal signal pipe: It is not 748example, libev itself uses this for its internal signal pipe: It is not
377visible to the libev user and should not keep \f(CW\*(C`ev_loop\*(C'\fR from exiting if 749visible to the libev user and should not keep \f(CW\*(C`ev_loop\*(C'\fR from exiting if
378no event watchers registered by it are active. It is also an excellent 750no event watchers registered by it are active. It is also an excellent
379way to do this for generic recurring timers or from within third-party 751way to do this for generic recurring timers or from within third-party
380libraries. Just remember to \fIunref after start\fR and \fIref before stop\fR. 752libraries. Just remember to \fIunref after start\fR and \fIref before stop\fR
753(but only if the watcher wasn't active before, or was active before,
754respectively).
755.Sp
756Example: Create a signal watcher, but keep it from keeping \f(CW\*(C`ev_loop\*(C'\fR
757running when nothing else is active.
758.Sp
759.Vb 4
760\& struct ev_signal exitsig;
761\& ev_signal_init (&exitsig, sig_cb, SIGINT);
762\& ev_signal_start (loop, &exitsig);
763\& evf_unref (loop);
764.Ve
765.Sp
766Example: For some weird reason, unregister the above signal handler again.
767.Sp
768.Vb 2
769\& ev_ref (loop);
770\& ev_signal_stop (loop, &exitsig);
771.Ve
772.IP "ev_set_io_collect_interval (loop, ev_tstamp interval)" 4
773.IX Item "ev_set_io_collect_interval (loop, ev_tstamp interval)"
774.PD 0
775.IP "ev_set_timeout_collect_interval (loop, ev_tstamp interval)" 4
776.IX Item "ev_set_timeout_collect_interval (loop, ev_tstamp interval)"
777.PD
778These advanced functions influence the time that libev will spend waiting
779for events. Both are by default \f(CW0\fR, meaning that libev will try to
780invoke timer/periodic callbacks and I/O callbacks with minimum latency.
781.Sp
782Setting these to a higher value (the \f(CW\*(C`interval\*(C'\fR \fImust\fR be >= \f(CW0\fR)
783allows libev to delay invocation of I/O and timer/periodic callbacks to
784increase efficiency of loop iterations.
785.Sp
786The background is that sometimes your program runs just fast enough to
787handle one (or very few) event(s) per loop iteration. While this makes
788the program responsive, it also wastes a lot of \s-1CPU\s0 time to poll for new
789events, especially with backends like \f(CW\*(C`select ()\*(C'\fR which have a high
790overhead for the actual polling but can deliver many events at once.
791.Sp
792By setting a higher \fIio collect interval\fR you allow libev to spend more
793time collecting I/O events, so you can handle more events per iteration,
794at the cost of increasing latency. Timeouts (both \f(CW\*(C`ev_periodic\*(C'\fR and
795\&\f(CW\*(C`ev_timer\*(C'\fR) will be not affected. Setting this to a non-null value will
796introduce an additional \f(CW\*(C`ev_sleep ()\*(C'\fR call into most loop iterations.
797.Sp
798Likewise, by setting a higher \fItimeout collect interval\fR you allow libev
799to spend more time collecting timeouts, at the expense of increased
800latency (the watcher callback will be called later). \f(CW\*(C`ev_io\*(C'\fR watchers
801will not be affected. Setting this to a non-null value will not introduce
802any overhead in libev.
803.Sp
804Many (busy) programs can usually benefit by setting the io collect
805interval to a value near \f(CW0.1\fR or so, which is often enough for
806interactive servers (of course not for games), likewise for timeouts. It
807usually doesn't make much sense to set it to a lower value than \f(CW0.01\fR,
808as this approsaches the timing granularity of most systems.
381.SH "ANATOMY OF A WATCHER" 809.SH "ANATOMY OF A WATCHER"
382.IX Header "ANATOMY OF A WATCHER" 810.IX Header "ANATOMY OF A WATCHER"
383A watcher is a structure that you create and register to record your 811A watcher is a structure that you create and register to record your
384interest in some event. For instance, if you want to wait for \s-1STDIN\s0 to 812interest in some event. For instance, if you want to wait for \s-1STDIN\s0 to
385become readable, you would create an \f(CW\*(C`ev_io\*(C'\fR watcher for that: 813become readable, you would create an \f(CW\*(C`ev_io\*(C'\fR watcher for that:
… …
388\& static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents) 816\& static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
389\& { 817\& {
390\& ev_io_stop (w); 818\& ev_io_stop (w);
391\& ev_unloop (loop, EVUNLOOP_ALL); 819\& ev_unloop (loop, EVUNLOOP_ALL);
392\& } 820\& }
393.Ve 821\&
394.PP
395.Vb 6
396\& struct ev_loop *loop = ev_default_loop (0); 822\& struct ev_loop *loop = ev_default_loop (0);
397\& struct ev_io stdin_watcher; 823\& struct ev_io stdin_watcher;
398\& ev_init (&stdin_watcher, my_cb); 824\& ev_init (&stdin_watcher, my_cb);
399\& ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ); 825\& ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
400\& ev_io_start (loop, &stdin_watcher); 826\& ev_io_start (loop, &stdin_watcher);
… …
421*)\*(C'\fR), and you can stop watching for events at any time by calling the 847*)\*(C'\fR), and you can stop watching for events at any time by calling the
422corresponding stop function (\f(CW\*(C`ev_<type>_stop (loop, watcher *)\*(C'\fR. 848corresponding stop function (\f(CW\*(C`ev_<type>_stop (loop, watcher *)\*(C'\fR.
423.PP 849.PP
424As long as your watcher is active (has been started but not stopped) you 850As long as your watcher is active (has been started but not stopped) you
425must not touch the values stored in it. Most specifically you must never 851must not touch the values stored in it. Most specifically you must never
426reinitialise it or call its set method. 852reinitialise it or call its \f(CW\*(C`set\*(C'\fR macro.
427.PP
428You can check whether an event is active by calling the \f(CW\*(C`ev_is_active
429(watcher *)\*(C'\fR macro. To see whether an event is outstanding (but the
430callback for it has not been called yet) you can use the \f(CW\*(C`ev_is_pending
431(watcher *)\*(C'\fR macro.
432.PP 853.PP
433Each and every callback receives the event loop pointer as first, the 854Each and every callback receives the event loop pointer as first, the
434registered watcher structure as second, and a bitset of received events as 855registered watcher structure as second, and a bitset of received events as
435third argument. 856third argument.
436.PP 857.PP
… …
461The signal specified in the \f(CW\*(C`ev_signal\*(C'\fR watcher has been received by a thread. 882The signal specified in the \f(CW\*(C`ev_signal\*(C'\fR watcher has been received by a thread.
462.ie n .IP """EV_CHILD""" 4 883.ie n .IP """EV_CHILD""" 4
463.el .IP "\f(CWEV_CHILD\fR" 4 884.el .IP "\f(CWEV_CHILD\fR" 4
464.IX Item "EV_CHILD" 885.IX Item "EV_CHILD"
465The pid specified in the \f(CW\*(C`ev_child\*(C'\fR watcher has received a status change. 886The pid specified in the \f(CW\*(C`ev_child\*(C'\fR watcher has received a status change.
887.ie n .IP """EV_STAT""" 4
888.el .IP "\f(CWEV_STAT\fR" 4
889.IX Item "EV_STAT"
890The path specified in the \f(CW\*(C`ev_stat\*(C'\fR watcher changed its attributes somehow.
466.ie n .IP """EV_IDLE""" 4 891.ie n .IP """EV_IDLE""" 4
467.el .IP "\f(CWEV_IDLE\fR" 4 892.el .IP "\f(CWEV_IDLE\fR" 4
468.IX Item "EV_IDLE" 893.IX Item "EV_IDLE"
469The \f(CW\*(C`ev_idle\*(C'\fR watcher has determined that you have nothing better to do. 894The \f(CW\*(C`ev_idle\*(C'\fR watcher has determined that you have nothing better to do.
470.ie n .IP """EV_PREPARE""" 4 895.ie n .IP """EV_PREPARE""" 4
… …
480\&\f(CW\*(C`ev_loop\*(C'\fR has gathered them, but before it invokes any callbacks for any 905\&\f(CW\*(C`ev_loop\*(C'\fR has gathered them, but before it invokes any callbacks for any
481received events. Callbacks of both watcher types can start and stop as 906received events. Callbacks of both watcher types can start and stop as
482many watchers as they want, and all of them will be taken into account 907many watchers as they want, and all of them will be taken into account
483(for example, a \f(CW\*(C`ev_prepare\*(C'\fR watcher might start an idle watcher to keep 908(for example, a \f(CW\*(C`ev_prepare\*(C'\fR watcher might start an idle watcher to keep
484\&\f(CW\*(C`ev_loop\*(C'\fR from blocking). 909\&\f(CW\*(C`ev_loop\*(C'\fR from blocking).
910.ie n .IP """EV_EMBED""" 4
911.el .IP "\f(CWEV_EMBED\fR" 4
912.IX Item "EV_EMBED"
913The embedded event loop specified in the \f(CW\*(C`ev_embed\*(C'\fR watcher needs attention.
914.ie n .IP """EV_FORK""" 4
915.el .IP "\f(CWEV_FORK\fR" 4
916.IX Item "EV_FORK"
917The event loop has been resumed in the child process after fork (see
918\&\f(CW\*(C`ev_fork\*(C'\fR).
485.ie n .IP """EV_ERROR""" 4 919.ie n .IP """EV_ERROR""" 4
486.el .IP "\f(CWEV_ERROR\fR" 4 920.el .IP "\f(CWEV_ERROR\fR" 4
487.IX Item "EV_ERROR" 921.IX Item "EV_ERROR"
488An unspecified error has occured, the watcher has been stopped. This might 922An unspecified error has occured, the watcher has been stopped. This might
489happen because the watcher could not be properly started because libev 923happen because the watcher could not be properly started because libev
… …
494Libev will usually signal a few \*(L"dummy\*(R" events together with an error, 928Libev will usually signal a few \*(L"dummy\*(R" events together with an error,
495for example it might indicate that a fd is readable or writable, and if 929for example it might indicate that a fd is readable or writable, and if
496your callbacks is well-written it can just attempt the operation and cope 930your callbacks is well-written it can just attempt the operation and cope
497with the error from \fIread()\fR or \fIwrite()\fR. This will not work in multithreaded 931with the error from \fIread()\fR or \fIwrite()\fR. This will not work in multithreaded
498programs, though, so beware. 932programs, though, so beware.
933.Sh "\s-1GENERIC\s0 \s-1WATCHER\s0 \s-1FUNCTIONS\s0"
934.IX Subsection "GENERIC WATCHER FUNCTIONS"
935In the following description, \f(CW\*(C`TYPE\*(C'\fR stands for the watcher type,
936e.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.
937.ie n .IP """ev_init"" (ev_TYPE *watcher, callback)" 4
938.el .IP "\f(CWev_init\fR (ev_TYPE *watcher, callback)" 4
939.IX Item "ev_init (ev_TYPE *watcher, callback)"
940This macro initialises the generic portion of a watcher. The contents
941of the watcher object can be arbitrary (so \f(CW\*(C`malloc\*(C'\fR will do). Only
942the generic parts of the watcher are initialised, you \fIneed\fR to call
943the type-specific \f(CW\*(C`ev_TYPE_set\*(C'\fR macro afterwards to initialise the
944type-specific parts. For each type there is also a \f(CW\*(C`ev_TYPE_init\*(C'\fR macro
945which rolls both calls into one.
946.Sp
947You can reinitialise a watcher at any time as long as it has been stopped
948(or never started) and there are no pending events outstanding.
949.Sp
950The callback is always of type \f(CW\*(C`void (*)(ev_loop *loop, ev_TYPE *watcher,
951int revents)\*(C'\fR.
952.ie n .IP """ev_TYPE_set"" (ev_TYPE *, [args])" 4
953.el .IP "\f(CWev_TYPE_set\fR (ev_TYPE *, [args])" 4
954.IX Item "ev_TYPE_set (ev_TYPE *, [args])"
955This macro initialises the type-specific parts of a watcher. You need to
956call \f(CW\*(C`ev_init\*(C'\fR at least once before you call this macro, but you can
957call \f(CW\*(C`ev_TYPE_set\*(C'\fR any number of times. You must not, however, call this
958macro on a watcher that is active (it can be pending, however, which is a
959difference to the \f(CW\*(C`ev_init\*(C'\fR macro).
960.Sp
961Although some watcher types do not have type-specific arguments
962(e.g. \f(CW\*(C`ev_prepare\*(C'\fR) you still need to call its \f(CW\*(C`set\*(C'\fR macro.
963.ie n .IP """ev_TYPE_init"" (ev_TYPE *watcher, callback, [args])" 4
964.el .IP "\f(CWev_TYPE_init\fR (ev_TYPE *watcher, callback, [args])" 4
965.IX Item "ev_TYPE_init (ev_TYPE *watcher, callback, [args])"
966This convinience macro rolls both \f(CW\*(C`ev_init\*(C'\fR and \f(CW\*(C`ev_TYPE_set\*(C'\fR macro
967calls into a single call. This is the most convinient method to initialise
968a watcher. The same limitations apply, of course.
969.ie n .IP """ev_TYPE_start"" (loop *, ev_TYPE *watcher)" 4
970.el .IP "\f(CWev_TYPE_start\fR (loop *, ev_TYPE *watcher)" 4
971.IX Item "ev_TYPE_start (loop *, ev_TYPE *watcher)"
972Starts (activates) the given watcher. Only active watchers will receive
973events. If the watcher is already active nothing will happen.
974.ie n .IP """ev_TYPE_stop"" (loop *, ev_TYPE *watcher)" 4
975.el .IP "\f(CWev_TYPE_stop\fR (loop *, ev_TYPE *watcher)" 4
976.IX Item "ev_TYPE_stop (loop *, ev_TYPE *watcher)"
977Stops the given watcher again (if active) and clears the pending
978status. It is possible that stopped watchers are pending (for example,
979non-repeating timers are being stopped when they become pending), but
980\&\f(CW\*(C`ev_TYPE_stop\*(C'\fR ensures that the watcher is neither active nor pending. If
981you want to free or reuse the memory used by the watcher it is therefore a
982good idea to always call its \f(CW\*(C`ev_TYPE_stop\*(C'\fR function.
983.IP "bool ev_is_active (ev_TYPE *watcher)" 4
984.IX Item "bool ev_is_active (ev_TYPE *watcher)"
985Returns a true value iff the watcher is active (i.e. it has been started
986and not yet been stopped). As long as a watcher is active you must not modify
987it.
988.IP "bool ev_is_pending (ev_TYPE *watcher)" 4
989.IX Item "bool ev_is_pending (ev_TYPE *watcher)"
990Returns a true value iff the watcher is pending, (i.e. it has outstanding
991events but its callback has not yet been invoked). As long as a watcher
992is pending (but not active) you must not call an init function on it (but
993\&\f(CW\*(C`ev_TYPE_set\*(C'\fR is safe), you must not change its priority, and you must
994make sure the watcher is available to libev (e.g. you cannot \f(CW\*(C`free ()\*(C'\fR
995it).
996.IP "callback ev_cb (ev_TYPE *watcher)" 4
997.IX Item "callback ev_cb (ev_TYPE *watcher)"
998Returns the callback currently set on the watcher.
999.IP "ev_cb_set (ev_TYPE *watcher, callback)" 4
1000.IX Item "ev_cb_set (ev_TYPE *watcher, callback)"
1001Change the callback. You can change the callback at virtually any time
1002(modulo threads).
1003.IP "ev_set_priority (ev_TYPE *watcher, priority)" 4
1004.IX Item "ev_set_priority (ev_TYPE *watcher, priority)"
1005.PD 0
1006.IP "int ev_priority (ev_TYPE *watcher)" 4
1007.IX Item "int ev_priority (ev_TYPE *watcher)"
1008.PD
1009Set and query the priority of the watcher. The priority is a small
1010integer between \f(CW\*(C`EV_MAXPRI\*(C'\fR (default: \f(CW2\fR) and \f(CW\*(C`EV_MINPRI\*(C'\fR
1011(default: \f(CW\*(C`\-2\*(C'\fR). Pending watchers with higher priority will be invoked
1012before watchers with lower priority, but priority will not keep watchers
1013from being executed (except for \f(CW\*(C`ev_idle\*(C'\fR watchers).
1014.Sp
1015This means that priorities are \fIonly\fR used for ordering callback
1016invocation after new events have been received. This is useful, for
1017example, to reduce latency after idling, or more often, to bind two
1018watchers on the same event and make sure one is called first.
1019.Sp
1020If you need to suppress invocation when higher priority events are pending
1021you need to look at \f(CW\*(C`ev_idle\*(C'\fR watchers, which provide this functionality.
1022.Sp
1023You \fImust not\fR change the priority of a watcher as long as it is active or
1024pending.
1025.Sp
1026The default priority used by watchers when no priority has been set is
1027always \f(CW0\fR, which is supposed to not be too high and not be too low :).
1028.Sp
1029Setting a priority outside the range of \f(CW\*(C`EV_MINPRI\*(C'\fR to \f(CW\*(C`EV_MAXPRI\*(C'\fR is
1030fine, as long as you do not mind that the priority value you query might
1031or might not have been adjusted to be within valid range.
1032.IP "ev_invoke (loop, ev_TYPE *watcher, int revents)" 4
1033.IX Item "ev_invoke (loop, ev_TYPE *watcher, int revents)"
1034Invoke the \f(CW\*(C`watcher\*(C'\fR with the given \f(CW\*(C`loop\*(C'\fR and \f(CW\*(C`revents\*(C'\fR. Neither
1035\&\f(CW\*(C`loop\*(C'\fR nor \f(CW\*(C`revents\*(C'\fR need to be valid as long as the watcher callback
1036can deal with that fact.
1037.IP "int ev_clear_pending (loop, ev_TYPE *watcher)" 4
1038.IX Item "int ev_clear_pending (loop, ev_TYPE *watcher)"
1039If the watcher is pending, this function returns clears its pending status
1040and returns its \f(CW\*(C`revents\*(C'\fR bitset (as if its callback was invoked). If the
1041watcher isn't pending it does nothing and returns \f(CW0\fR.
499.Sh "\s-1ASSOCIATING\s0 \s-1CUSTOM\s0 \s-1DATA\s0 \s-1WITH\s0 A \s-1WATCHER\s0" 1042.Sh "\s-1ASSOCIATING\s0 \s-1CUSTOM\s0 \s-1DATA\s0 \s-1WITH\s0 A \s-1WATCHER\s0"
500.IX Subsection "ASSOCIATING CUSTOM DATA WITH A WATCHER" 1043.IX Subsection "ASSOCIATING CUSTOM DATA WITH A WATCHER"
501Each watcher has, by default, a member \f(CW\*(C`void *data\*(C'\fR that you can change 1044Each watcher has, by default, a member \f(CW\*(C`void *data\*(C'\fR that you can change
502and read at any time, libev will completely ignore it. This can be used 1045and read at any time, libev will completely ignore it. This can be used
503to associate arbitrary data with your watcher. If you need more data and 1046to associate arbitrary data with your watcher. If you need more data and
… …
524\& struct my_io *w = (struct my_io *)w_; 1067\& struct my_io *w = (struct my_io *)w_;
525\& ... 1068\& ...
526\& } 1069\& }
527.Ve 1070.Ve
528.PP 1071.PP
529More interesting and less C\-conformant ways of catsing your callback type 1072More interesting and less C\-conformant ways of casting your callback type
530have been omitted.... 1073instead have been omitted.
1074.PP
1075Another common scenario is having some data structure with multiple
1076watchers:
1077.PP
1078.Vb 6
1079\& struct my_biggy
1080\& {
1081\& int some_data;
1082\& ev_timer t1;
1083\& ev_timer t2;
1084\& }
1085.Ve
1086.PP
1087In this case getting the pointer to \f(CW\*(C`my_biggy\*(C'\fR is a bit more complicated,
1088you need to use \f(CW\*(C`offsetof\*(C'\fR:
1089.PP
1090.Vb 1
1091\& #include <stddef.h>
1092\&
1093\& static void
1094\& t1_cb (EV_P_ struct ev_timer *w, int revents)
1095\& {
1096\& struct my_biggy big = (struct my_biggy *
1097\& (((char *)w) \- offsetof (struct my_biggy, t1));
1098\& }
1099\&
1100\& static void
1101\& t2_cb (EV_P_ struct ev_timer *w, int revents)
1102\& {
1103\& struct my_biggy big = (struct my_biggy *
1104\& (((char *)w) \- offsetof (struct my_biggy, t2));
1105\& }
1106.Ve
531.SH "WATCHER TYPES" 1107.SH "WATCHER TYPES"
532.IX Header "WATCHER TYPES" 1108.IX Header "WATCHER TYPES"
533This section describes each watcher in detail, but will not repeat 1109This section describes each watcher in detail, but will not repeat
534information given in the last section. 1110information given in the last section. Any initialisation/set macros,
1111functions and members specific to the watcher type are explained.
1112.PP
1113Members are additionally marked with either \fI[read\-only]\fR, meaning that,
1114while the watcher is active, you can look at the member and expect some
1115sensible content, but you must not modify it (you can modify it while the
1116watcher is stopped to your hearts content), or \fI[read\-write]\fR, which
1117means you can expect it to have some sensible content while the watcher
1118is active, but you can also modify it. Modifying it may not do something
1119sensible or take immediate effect (or do anything at all), but libev will
1120not crash or malfunction in any way.
535.ie n .Sh """ev_io"" \- is this file descriptor readable or writable" 1121.ie n .Sh """ev_io"" \- is this file descriptor readable or writable?"
536.el .Sh "\f(CWev_io\fP \- is this file descriptor readable or writable" 1122.el .Sh "\f(CWev_io\fP \- is this file descriptor readable or writable?"
537.IX Subsection "ev_io - is this file descriptor readable or writable" 1123.IX Subsection "ev_io - is this file descriptor readable or writable?"
538I/O watchers check whether a file descriptor is readable or writable 1124I/O watchers check whether a file descriptor is readable or writable
539in each iteration of the event loop (This behaviour is called 1125in each iteration of the event loop, or, more precisely, when reading
540level-triggering because you keep receiving events as long as the 1126would not block the process and writing would at least be able to write
541condition persists. Remember you can stop the watcher if you don't want to 1127some data. This behaviour is called level-triggering because you keep
542act on the event and neither want to receive future events). 1128receiving events as long as the condition persists. Remember you can stop
1129the watcher if you don't want to act on the event and neither want to
1130receive future events.
543.PP 1131.PP
544In general you can register as many read and/or write event watchers per 1132In general you can register as many read and/or write event watchers per
545fd as you want (as long as you don't confuse yourself). Setting all file 1133fd as you want (as long as you don't confuse yourself). Setting all file
546descriptors to non-blocking mode is also usually a good idea (but not 1134descriptors to non-blocking mode is also usually a good idea (but not
547required if you know what you are doing). 1135required if you know what you are doing).
548.PP 1136.PP
549You have to be careful with dup'ed file descriptors, though. Some backends
550(the linux epoll backend is a notable example) cannot handle dup'ed file
551descriptors correctly if you register interest in two or more fds pointing
552to the same underlying file/socket etc. description (that is, they share
553the same underlying \*(L"file open\*(R").
554.PP
555If you must do this, then force the use of a known-to-be-good backend 1137If you must do this, then force the use of a known-to-be-good backend
556(at the time of this writing, this includes only \s-1EVMETHOD_SELECT\s0 and 1138(at the time of this writing, this includes only \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR and
557\&\s-1EVMETHOD_POLL\s0). 1139\&\f(CW\*(C`EVBACKEND_POLL\*(C'\fR).
1140.PP
1141Another thing you have to watch out for is that it is quite easy to
1142receive \*(L"spurious\*(R" readyness notifications, that is your callback might
1143be called with \f(CW\*(C`EV_READ\*(C'\fR but a subsequent \f(CW\*(C`read\*(C'\fR(2) will actually block
1144because there is no data. Not only are some backends known to create a
1145lot of those (for example solaris ports), it is very easy to get into
1146this situation even with a relatively standard program structure. Thus
1147it is best to always use non-blocking I/O: An extra \f(CW\*(C`read\*(C'\fR(2) returning
1148\&\f(CW\*(C`EAGAIN\*(C'\fR is far preferable to a program hanging until some data arrives.
1149.PP
1150If you cannot run the fd in non-blocking mode (for example you should not
1151play around with an Xlib connection), then you have to seperately re-test
1152whether a file descriptor is really ready with a known-to-be good interface
1153such as poll (fortunately in our Xlib example, Xlib already does this on
1154its own, so its quite safe to use).
1155.PP
1156\fIThe special problem of disappearing file descriptors\fR
1157.IX Subsection "The special problem of disappearing file descriptors"
1158.PP
1159Some backends (e.g. kqueue, epoll) need to be told about closing a file
1160descriptor (either by calling \f(CW\*(C`close\*(C'\fR explicitly or by any other means,
1161such as \f(CW\*(C`dup\*(C'\fR). The reason is that you register interest in some file
1162descriptor, but when it goes away, the operating system will silently drop
1163this interest. If another file descriptor with the same number then is
1164registered with libev, there is no efficient way to see that this is, in
1165fact, a different file descriptor.
1166.PP
1167To avoid having to explicitly tell libev about such cases, libev follows
1168the following policy: Each time \f(CW\*(C`ev_io_set\*(C'\fR is being called, libev
1169will assume that this is potentially a new file descriptor, otherwise
1170it is assumed that the file descriptor stays the same. That means that
1171you \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
1172descriptor even if the file descriptor number itself did not change.
1173.PP
1174This is how one would do it normally anyway, the important point is that
1175the libev application should not optimise around libev but should leave
1176optimisations to libev.
1177.PP
1178\fIThe special problem of dup'ed file descriptors\fR
1179.IX Subsection "The special problem of dup'ed file descriptors"
1180.PP
1181Some backends (e.g. epoll), cannot register events for file descriptors,
1182but only events for the underlying file descriptions. That means when you
1183have \f(CW\*(C`dup ()\*(C'\fR'ed file descriptors or weirder constellations, and register
1184events for them, only one file descriptor might actually receive events.
1185.PP
1186There is no workaround possible except not registering events
1187for potentially \f(CW\*(C`dup ()\*(C'\fR'ed file descriptors, or to resort to
1188\&\f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or \f(CW\*(C`EVBACKEND_POLL\*(C'\fR.
1189.PP
1190\fIThe special problem of fork\fR
1191.IX Subsection "The special problem of fork"
1192.PP
1193Some backends (epoll, kqueue) do not support \f(CW\*(C`fork ()\*(C'\fR at all or exhibit
1194useless behaviour. Libev fully supports fork, but needs to be told about
1195it in the child.
1196.PP
1197To support fork in your programs, you either have to call
1198\&\f(CW\*(C`ev_default_fork ()\*(C'\fR or \f(CW\*(C`ev_loop_fork ()\*(C'\fR after a fork in the child,
1199enable \f(CW\*(C`EVFLAG_FORKCHECK\*(C'\fR, or resort to \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or
1200\&\f(CW\*(C`EVBACKEND_POLL\*(C'\fR.
1201.PP
1202\fIWatcher-Specific Functions\fR
1203.IX Subsection "Watcher-Specific Functions"
558.IP "ev_io_init (ev_io *, callback, int fd, int events)" 4 1204.IP "ev_io_init (ev_io *, callback, int fd, int events)" 4
559.IX Item "ev_io_init (ev_io *, callback, int fd, int events)" 1205.IX Item "ev_io_init (ev_io *, callback, int fd, int events)"
560.PD 0 1206.PD 0
561.IP "ev_io_set (ev_io *, int fd, int events)" 4 1207.IP "ev_io_set (ev_io *, int fd, int events)" 4
562.IX Item "ev_io_set (ev_io *, int fd, int events)" 1208.IX Item "ev_io_set (ev_io *, int fd, int events)"
563.PD 1209.PD
564Configures an \f(CW\*(C`ev_io\*(C'\fR watcher. The fd is the file descriptor to rceeive 1210Configures an \f(CW\*(C`ev_io\*(C'\fR watcher. The \f(CW\*(C`fd\*(C'\fR is the file descriptor to
565events 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 | 1211rceeive events for and events is either \f(CW\*(C`EV_READ\*(C'\fR, \f(CW\*(C`EV_WRITE\*(C'\fR or
566EV_WRITE\*(C'\fR to receive the given events. 1212\&\f(CW\*(C`EV_READ | EV_WRITE\*(C'\fR to receive the given events.
1213.IP "int fd [read\-only]" 4
1214.IX Item "int fd [read-only]"
1215The file descriptor being watched.
1216.IP "int events [read\-only]" 4
1217.IX Item "int events [read-only]"
1218The events being watched.
1219.PP
1220\fIExamples\fR
1221.IX Subsection "Examples"
1222.PP
1223Example: Call \f(CW\*(C`stdin_readable_cb\*(C'\fR when \s-1STDIN_FILENO\s0 has become, well
1224readable, but only once. Since it is likely line-buffered, you could
1225attempt to read a whole line in the callback.
1226.PP
1227.Vb 6
1228\& static void
1229\& stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1230\& {
1231\& ev_io_stop (loop, w);
1232\& .. read from stdin here (or from w\->fd) and haqndle any I/O errors
1233\& }
1234\&
1235\& ...
1236\& struct ev_loop *loop = ev_default_init (0);
1237\& struct ev_io stdin_readable;
1238\& ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
1239\& ev_io_start (loop, &stdin_readable);
1240\& ev_loop (loop, 0);
1241.Ve
567.ie n .Sh """ev_timer"" \- relative and optionally recurring timeouts" 1242.ie n .Sh """ev_timer"" \- relative and optionally repeating timeouts"
568.el .Sh "\f(CWev_timer\fP \- relative and optionally recurring timeouts" 1243.el .Sh "\f(CWev_timer\fP \- relative and optionally repeating timeouts"
569.IX Subsection "ev_timer - relative and optionally recurring timeouts" 1244.IX Subsection "ev_timer - relative and optionally repeating timeouts"
570Timer watchers are simple relative timers that generate an event after a 1245Timer watchers are simple relative timers that generate an event after a
571given time, and optionally repeating in regular intervals after that. 1246given time, and optionally repeating in regular intervals after that.
572.PP 1247.PP
573The timers are based on real time, that is, if you register an event that 1248The timers are based on real time, that is, if you register an event that
574times out after an hour and you reset your system clock to last years 1249times out after an hour and you reset your system clock to last years
575time, it will still time out after (roughly) and hour. \*(L"Roughly\*(R" because 1250time, it will still time out after (roughly) and hour. \*(L"Roughly\*(R" because
576detecting time jumps is hard, and soem inaccuracies are unavoidable (the 1251detecting time jumps is hard, and some inaccuracies are unavoidable (the
577monotonic clock option helps a lot here). 1252monotonic clock option helps a lot here).
578.PP 1253.PP
579The relative timeouts are calculated relative to the \f(CW\*(C`ev_now ()\*(C'\fR 1254The relative timeouts are calculated relative to the \f(CW\*(C`ev_now ()\*(C'\fR
580time. This is usually the right thing as this timestamp refers to the time 1255time. This is usually the right thing as this timestamp refers to the time
581of the event triggering whatever timeout you are modifying/starting. If 1256of the event triggering whatever timeout you are modifying/starting. If
582you suspect event processing to be delayed and you *need* to base the timeout 1257you suspect event processing to be delayed and you \fIneed\fR to base the timeout
583on the current time, use something like this to adjust for this: 1258on the current time, use something like this to adjust for this:
584.PP 1259.PP
585.Vb 1 1260.Vb 1
586\& ev_timer_set (&timer, after + ev_now () - ev_time (), 0.); 1261\& ev_timer_set (&timer, after + ev_now () \- ev_time (), 0.);
587.Ve 1262.Ve
1263.PP
1264The callback is guarenteed to be invoked only when its timeout has passed,
1265but if multiple timers become ready during the same loop iteration then
1266order of execution is undefined.
1267.PP
1268\fIWatcher-Specific Functions and Data Members\fR
1269.IX Subsection "Watcher-Specific Functions and Data Members"
588.IP "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)" 4 1270.IP "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)" 4
589.IX Item "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)" 1271.IX Item "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)"
590.PD 0 1272.PD 0
591.IP "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)" 4 1273.IP "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)" 4
592.IX Item "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)" 1274.IX Item "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)"
… …
604.IP "ev_timer_again (loop)" 4 1286.IP "ev_timer_again (loop)" 4
605.IX Item "ev_timer_again (loop)" 1287.IX Item "ev_timer_again (loop)"
606This will act as if the timer timed out and restart it again if it is 1288This will act as if the timer timed out and restart it again if it is
607repeating. The exact semantics are: 1289repeating. The exact semantics are:
608.Sp 1290.Sp
1291If the timer is pending, its pending status is cleared.
1292.Sp
609If the timer is started but nonrepeating, stop it. 1293If the timer is started but nonrepeating, stop it (as if it timed out).
610.Sp 1294.Sp
611If the timer is repeating, either start it if necessary (with the repeat 1295If the timer is repeating, either start it if necessary (with the
612value), or reset the running timer to the repeat value. 1296\&\f(CW\*(C`repeat\*(C'\fR value), or reset the running timer to the \f(CW\*(C`repeat\*(C'\fR value.
613.Sp 1297.Sp
614This sounds a bit complicated, but here is a useful and typical 1298This sounds a bit complicated, but here is a useful and typical
615example: Imagine you have a tcp connection and you want a so-called idle 1299example: Imagine you have a tcp connection and you want a so-called idle
616timeout, that is, you want to be called when there have been, say, 60 1300timeout, that is, you want to be called when there have been, say, 60
617seconds of inactivity on the socket. The easiest way to do this is to 1301seconds of inactivity on the socket. The easiest way to do this is to
618configure an \f(CW\*(C`ev_timer\*(C'\fR with after=repeat=60 and calling ev_timer_again each 1302configure an \f(CW\*(C`ev_timer\*(C'\fR with a \f(CW\*(C`repeat\*(C'\fR value of \f(CW60\fR and then call
619time you successfully read or write some data. If you go into an idle 1303\&\f(CW\*(C`ev_timer_again\*(C'\fR each time you successfully read or write some data. If
620state where you do not expect data to travel on the socket, you can stop 1304you go into an idle state where you do not expect data to travel on the
621the timer, and again will automatically restart it if need be. 1305socket, you can \f(CW\*(C`ev_timer_stop\*(C'\fR the timer, and \f(CW\*(C`ev_timer_again\*(C'\fR will
1306automatically restart it if need be.
1307.Sp
1308That means you can ignore the \f(CW\*(C`after\*(C'\fR value and \f(CW\*(C`ev_timer_start\*(C'\fR
1309altogether and only ever use the \f(CW\*(C`repeat\*(C'\fR value and \f(CW\*(C`ev_timer_again\*(C'\fR:
1310.Sp
1311.Vb 8
1312\& ev_timer_init (timer, callback, 0., 5.);
1313\& ev_timer_again (loop, timer);
1314\& ...
1315\& timer\->again = 17.;
1316\& ev_timer_again (loop, timer);
1317\& ...
1318\& timer\->again = 10.;
1319\& ev_timer_again (loop, timer);
1320.Ve
1321.Sp
1322This is more slightly efficient then stopping/starting the timer each time
1323you want to modify its timeout value.
1324.IP "ev_tstamp repeat [read\-write]" 4
1325.IX Item "ev_tstamp repeat [read-write]"
1326The current \f(CW\*(C`repeat\*(C'\fR value. Will be used each time the watcher times out
1327or \f(CW\*(C`ev_timer_again\*(C'\fR is called and determines the next timeout (if any),
1328which is also when any modifications are taken into account.
1329.PP
1330\fIExamples\fR
1331.IX Subsection "Examples"
1332.PP
1333Example: Create a timer that fires after 60 seconds.
1334.PP
1335.Vb 5
1336\& static void
1337\& one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1338\& {
1339\& .. one minute over, w is actually stopped right here
1340\& }
1341\&
1342\& struct ev_timer mytimer;
1343\& ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
1344\& ev_timer_start (loop, &mytimer);
1345.Ve
1346.PP
1347Example: Create a timeout timer that times out after 10 seconds of
1348inactivity.
1349.PP
1350.Vb 5
1351\& static void
1352\& timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1353\& {
1354\& .. ten seconds without any activity
1355\& }
1356\&
1357\& struct ev_timer mytimer;
1358\& ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1359\& ev_timer_again (&mytimer); /* start timer */
1360\& ev_loop (loop, 0);
1361\&
1362\& // and in some piece of code that gets executed on any "activity":
1363\& // reset the timeout to start ticking again at 10 seconds
1364\& ev_timer_again (&mytimer);
1365.Ve
622.ie n .Sh """ev_periodic"" \- to cron or not to cron" 1366.ie n .Sh """ev_periodic"" \- to cron or not to cron?"
623.el .Sh "\f(CWev_periodic\fP \- to cron or not to cron" 1367.el .Sh "\f(CWev_periodic\fP \- to cron or not to cron?"
624.IX Subsection "ev_periodic - to cron or not to cron" 1368.IX Subsection "ev_periodic - to cron or not to cron?"
625Periodic watchers are also timers of a kind, but they are very versatile 1369Periodic watchers are also timers of a kind, but they are very versatile
626(and unfortunately a bit complex). 1370(and unfortunately a bit complex).
627.PP 1371.PP
628Unlike \f(CW\*(C`ev_timer\*(C'\fR's, they are not based on real time (or relative time) 1372Unlike \f(CW\*(C`ev_timer\*(C'\fR's, they are not based on real time (or relative time)
629but on wallclock time (absolute time). You can tell a periodic watcher 1373but on wallclock time (absolute time). You can tell a periodic watcher
630to trigger \*(L"at\*(R" some specific point in time. For example, if you tell a 1374to trigger \*(L"at\*(R" some specific point in time. For example, if you tell a
631periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now () 1375periodic watcher to trigger in 10 seconds (by specifiying e.g. \f(CW\*(C`ev_now ()
632+ 10.>) and then reset your system clock to the last year, then it will 1376+ 10.\*(C'\fR) and then reset your system clock to the last year, then it will
633take a year to trigger the event (unlike an \f(CW\*(C`ev_timer\*(C'\fR, which would trigger 1377take a year to trigger the event (unlike an \f(CW\*(C`ev_timer\*(C'\fR, which would trigger
634roughly 10 seconds later and of course not if you reset your system time 1378roughly 10 seconds later).
635again).
636.PP 1379.PP
637They can also be used to implement vastly more complex timers, such as 1380They can also be used to implement vastly more complex timers, such as
638triggering an event on eahc midnight, local time. 1381triggering an event on each midnight, local time or other, complicated,
1382rules.
1383.PP
1384As with timers, the callback is guarenteed to be invoked only when the
1385time (\f(CW\*(C`at\*(C'\fR) has been passed, but if multiple periodic timers become ready
1386during the same loop iteration then order of execution is undefined.
1387.PP
1388\fIWatcher-Specific Functions and Data Members\fR
1389.IX Subsection "Watcher-Specific Functions and Data Members"
639.IP "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 4 1390.IP "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 4
640.IX Item "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 1391.IX Item "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)"
641.PD 0 1392.PD 0
642.IP "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)" 4 1393.IP "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)" 4
643.IX Item "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)" 1394.IX Item "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)"
644.PD 1395.PD
645Lots of arguments, lets sort it out... There are basically three modes of 1396Lots of arguments, lets sort it out... There are basically three modes of
646operation, and we will explain them from simplest to complex: 1397operation, and we will explain them from simplest to complex:
647.RS 4 1398.RS 4
1399.IP "\(bu" 4
648.IP "* absolute timer (interval = reschedule_cb = 0)" 4 1400absolute timer (at = time, interval = reschedule_cb = 0)
649.IX Item "absolute timer (interval = reschedule_cb = 0)" 1401.Sp
650In this configuration the watcher triggers an event at the wallclock time 1402In this configuration the watcher triggers an event at the wallclock time
651\&\f(CW\*(C`at\*(C'\fR and doesn't repeat. It will not adjust when a time jump occurs, 1403\&\f(CW\*(C`at\*(C'\fR and doesn't repeat. It will not adjust when a time jump occurs,
652that is, if it is to be run at January 1st 2011 then it will run when the 1404that is, if it is to be run at January 1st 2011 then it will run when the
653system time reaches or surpasses this time. 1405system time reaches or surpasses this time.
1406.IP "\(bu" 4
654.IP "* non-repeating interval timer (interval > 0, reschedule_cb = 0)" 4 1407non-repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
655.IX Item "non-repeating interval timer (interval > 0, reschedule_cb = 0)" 1408.Sp
656In this mode the watcher will always be scheduled to time out at the next 1409In this mode the watcher will always be scheduled to time out at the next
657\&\f(CW\*(C`at + N * interval\*(C'\fR time (for some integer N) and then repeat, regardless 1410\&\f(CW\*(C`at + N * interval\*(C'\fR time (for some integer N, which can also be negative)
658of any time jumps. 1411and then repeat, regardless of any time jumps.
659.Sp 1412.Sp
660This can be used to create timers that do not drift with respect to system 1413This can be used to create timers that do not drift with respect to system
661time: 1414time:
662.Sp 1415.Sp
663.Vb 1 1416.Vb 1
… …
670by 3600. 1423by 3600.
671.Sp 1424.Sp
672Another way to think about it (for the mathematically inclined) is that 1425Another way to think about it (for the mathematically inclined) is that
673\&\f(CW\*(C`ev_periodic\*(C'\fR will try to run the callback in this mode at the next possible 1426\&\f(CW\*(C`ev_periodic\*(C'\fR will try to run the callback in this mode at the next possible
674time where \f(CW\*(C`time = at (mod interval)\*(C'\fR, regardless of any time jumps. 1427time where \f(CW\*(C`time = at (mod interval)\*(C'\fR, regardless of any time jumps.
675.IP "* manual reschedule mode (reschedule_cb = callback)" 4 1428.Sp
676.IX Item "manual reschedule mode (reschedule_cb = callback)" 1429For numerical stability it is preferable that the \f(CW\*(C`at\*(C'\fR value is near
1430\&\f(CW\*(C`ev_now ()\*(C'\fR (the current time), but there is no range requirement for
1431this value.
1432.IP "\(bu" 4
1433manual reschedule mode (at and interval ignored, reschedule_cb = callback)
1434.Sp
677In this mode the values for \f(CW\*(C`interval\*(C'\fR and \f(CW\*(C`at\*(C'\fR are both being 1435In this mode the values for \f(CW\*(C`interval\*(C'\fR and \f(CW\*(C`at\*(C'\fR are both being
678ignored. Instead, each time the periodic watcher gets scheduled, the 1436ignored. Instead, each time the periodic watcher gets scheduled, the
679reschedule callback will be called with the watcher as first, and the 1437reschedule callback will be called with the watcher as first, and the
680current time as second argument. 1438current time as second argument.
681.Sp 1439.Sp
682\&\s-1NOTE:\s0 \fIThis callback \s-1MUST\s0 \s-1NOT\s0 stop or destroy any periodic watcher, 1440\&\s-1NOTE:\s0 \fIThis callback \s-1MUST\s0 \s-1NOT\s0 stop or destroy any periodic watcher,
683ever, or make any event loop modifications\fR. If you need to stop it, 1441ever, or make any event loop modifications\fR. If you need to stop it,
684return \f(CW\*(C`now + 1e30\*(C'\fR (or so, fudge fudge) and stop it afterwards (e.g. by 1442return \f(CW\*(C`now + 1e30\*(C'\fR (or so, fudge fudge) and stop it afterwards (e.g. by
685starting a prepare watcher). 1443starting an \f(CW\*(C`ev_prepare\*(C'\fR watcher, which is legal).
686.Sp 1444.Sp
687Its prototype is \f(CW\*(C`ev_tstamp (*reschedule_cb)(struct ev_periodic *w, 1445Its prototype is \f(CW\*(C`ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
688ev_tstamp now)\*(C'\fR, e.g.: 1446ev_tstamp now)\*(C'\fR, e.g.:
689.Sp 1447.Sp
690.Vb 4 1448.Vb 4
… …
714.IX Item "ev_periodic_again (loop, ev_periodic *)" 1472.IX Item "ev_periodic_again (loop, ev_periodic *)"
715Simply stops and restarts the periodic watcher again. This is only useful 1473Simply stops and restarts the periodic watcher again. This is only useful
716when you changed some parameters or the reschedule callback would return 1474when you changed some parameters or the reschedule callback would return
717a different time than the last time it was called (e.g. in a crond like 1475a different time than the last time it was called (e.g. in a crond like
718program when the crontabs have changed). 1476program when the crontabs have changed).
1477.IP "ev_tstamp offset [read\-write]" 4
1478.IX Item "ev_tstamp offset [read-write]"
1479When repeating, this contains the offset value, otherwise this is the
1480absolute point in time (the \f(CW\*(C`at\*(C'\fR value passed to \f(CW\*(C`ev_periodic_set\*(C'\fR).
1481.Sp
1482Can be modified any time, but changes only take effect when the periodic
1483timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being called.
1484.IP "ev_tstamp interval [read\-write]" 4
1485.IX Item "ev_tstamp interval [read-write]"
1486The current interval value. Can be modified any time, but changes only
1487take effect when the periodic timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being
1488called.
1489.IP "ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read\-write]" 4
1490.IX Item "ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]"
1491The current reschedule callback, or \f(CW0\fR, if this functionality is
1492switched off. Can be changed any time, but changes only take effect when
1493the periodic timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being called.
1494.IP "ev_tstamp at [read\-only]" 4
1495.IX Item "ev_tstamp at [read-only]"
1496When active, contains the absolute time that the watcher is supposed to
1497trigger next.
1498.PP
1499\fIExamples\fR
1500.IX Subsection "Examples"
1501.PP
1502Example: Call a callback every hour, or, more precisely, whenever the
1503system clock is divisible by 3600. The callback invocation times have
1504potentially a lot of jittering, but good long-term stability.
1505.PP
1506.Vb 5
1507\& static void
1508\& clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1509\& {
1510\& ... its now a full hour (UTC, or TAI or whatever your clock follows)
1511\& }
1512\&
1513\& struct ev_periodic hourly_tick;
1514\& ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1515\& ev_periodic_start (loop, &hourly_tick);
1516.Ve
1517.PP
1518Example: The same as above, but use a reschedule callback to do it:
1519.PP
1520.Vb 1
1521\& #include <math.h>
1522\&
1523\& static ev_tstamp
1524\& my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1525\& {
1526\& return fmod (now, 3600.) + 3600.;
1527\& }
1528\&
1529\& ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1530.Ve
1531.PP
1532Example: Call a callback every hour, starting now:
1533.PP
1534.Vb 4
1535\& struct ev_periodic hourly_tick;
1536\& ev_periodic_init (&hourly_tick, clock_cb,
1537\& fmod (ev_now (loop), 3600.), 3600., 0);
1538\& ev_periodic_start (loop, &hourly_tick);
1539.Ve
719.ie n .Sh """ev_signal"" \- signal me when a signal gets signalled" 1540.ie n .Sh """ev_signal"" \- signal me when a signal gets signalled!"
720.el .Sh "\f(CWev_signal\fP \- signal me when a signal gets signalled" 1541.el .Sh "\f(CWev_signal\fP \- signal me when a signal gets signalled!"
721.IX Subsection "ev_signal - signal me when a signal gets signalled" 1542.IX Subsection "ev_signal - signal me when a signal gets signalled!"
722Signal watchers will trigger an event when the process receives a specific 1543Signal watchers will trigger an event when the process receives a specific
723signal one or more times. Even though signals are very asynchronous, libev 1544signal one or more times. Even though signals are very asynchronous, libev
724will try it's best to deliver signals synchronously, i.e. as part of the 1545will try it's best to deliver signals synchronously, i.e. as part of the
725normal event processing, like any other event. 1546normal event processing, like any other event.
726.PP 1547.PP
… …
728first watcher gets started will libev actually register a signal watcher 1549first watcher gets started will libev actually register a signal watcher
729with the kernel (thus it coexists with your own signal handlers as long 1550with the kernel (thus it coexists with your own signal handlers as long
730as you don't register any with libev). Similarly, when the last signal 1551as you don't register any with libev). Similarly, when the last signal
731watcher for a signal is stopped libev will reset the signal handler to 1552watcher for a signal is stopped libev will reset the signal handler to
732\&\s-1SIG_DFL\s0 (regardless of what it was set to before). 1553\&\s-1SIG_DFL\s0 (regardless of what it was set to before).
1554.PP
1555\fIWatcher-Specific Functions and Data Members\fR
1556.IX Subsection "Watcher-Specific Functions and Data Members"
733.IP "ev_signal_init (ev_signal *, callback, int signum)" 4 1557.IP "ev_signal_init (ev_signal *, callback, int signum)" 4
734.IX Item "ev_signal_init (ev_signal *, callback, int signum)" 1558.IX Item "ev_signal_init (ev_signal *, callback, int signum)"
735.PD 0 1559.PD 0
736.IP "ev_signal_set (ev_signal *, int signum)" 4 1560.IP "ev_signal_set (ev_signal *, int signum)" 4
737.IX Item "ev_signal_set (ev_signal *, int signum)" 1561.IX Item "ev_signal_set (ev_signal *, int signum)"
738.PD 1562.PD
739Configures the watcher to trigger on the given signal number (usually one 1563Configures the watcher to trigger on the given signal number (usually one
740of the \f(CW\*(C`SIGxxx\*(C'\fR constants). 1564of the \f(CW\*(C`SIGxxx\*(C'\fR constants).
1565.IP "int signum [read\-only]" 4
1566.IX Item "int signum [read-only]"
1567The signal the watcher watches out for.
741.ie n .Sh """ev_child"" \- wait for pid status changes" 1568.ie n .Sh """ev_child"" \- watch out for process status changes"
742.el .Sh "\f(CWev_child\fP \- wait for pid status changes" 1569.el .Sh "\f(CWev_child\fP \- watch out for process status changes"
743.IX Subsection "ev_child - wait for pid status changes" 1570.IX Subsection "ev_child - watch out for process status changes"
744Child watchers trigger when your process receives a \s-1SIGCHLD\s0 in response to 1571Child watchers trigger when your process receives a \s-1SIGCHLD\s0 in response to
745some child status changes (most typically when a child of yours dies). 1572some child status changes (most typically when a child of yours dies).
1573.PP
1574\fIWatcher-Specific Functions and Data Members\fR
1575.IX Subsection "Watcher-Specific Functions and Data Members"
746.IP "ev_child_init (ev_child *, callback, int pid)" 4 1576.IP "ev_child_init (ev_child *, callback, int pid, int trace)" 4
747.IX Item "ev_child_init (ev_child *, callback, int pid)" 1577.IX Item "ev_child_init (ev_child *, callback, int pid, int trace)"
748.PD 0 1578.PD 0
749.IP "ev_child_set (ev_child *, int pid)" 4 1579.IP "ev_child_set (ev_child *, int pid, int trace)" 4
750.IX Item "ev_child_set (ev_child *, int pid)" 1580.IX Item "ev_child_set (ev_child *, int pid, int trace)"
751.PD 1581.PD
752Configures the watcher to wait for status changes of process \f(CW\*(C`pid\*(C'\fR (or 1582Configures the watcher to wait for status changes of process \f(CW\*(C`pid\*(C'\fR (or
753\&\fIany\fR process if \f(CW\*(C`pid\*(C'\fR is specified as \f(CW0\fR). The callback can look 1583\&\fIany\fR process if \f(CW\*(C`pid\*(C'\fR is specified as \f(CW0\fR). The callback can look
754at the \f(CW\*(C`rstatus\*(C'\fR member of the \f(CW\*(C`ev_child\*(C'\fR watcher structure to see 1584at the \f(CW\*(C`rstatus\*(C'\fR member of the \f(CW\*(C`ev_child\*(C'\fR watcher structure to see
755the status word (use the macros from \f(CW\*(C`sys/wait.h\*(C'\fR and see your systems 1585the status word (use the macros from \f(CW\*(C`sys/wait.h\*(C'\fR and see your systems
756\&\f(CW\*(C`waitpid\*(C'\fR documentation). The \f(CW\*(C`rpid\*(C'\fR member contains the pid of the 1586\&\f(CW\*(C`waitpid\*(C'\fR documentation). The \f(CW\*(C`rpid\*(C'\fR member contains the pid of the
757process causing the status change. 1587process causing the status change. \f(CW\*(C`trace\*(C'\fR must be either \f(CW0\fR (only
1588activate the watcher when the process terminates) or \f(CW1\fR (additionally
1589activate the watcher when the process is stopped or continued).
1590.IP "int pid [read\-only]" 4
1591.IX Item "int pid [read-only]"
1592The process id this watcher watches out for, or \f(CW0\fR, meaning any process id.
1593.IP "int rpid [read\-write]" 4
1594.IX Item "int rpid [read-write]"
1595The process id that detected a status change.
1596.IP "int rstatus [read\-write]" 4
1597.IX Item "int rstatus [read-write]"
1598The process exit/trace status caused by \f(CW\*(C`rpid\*(C'\fR (see your systems
1599\&\f(CW\*(C`waitpid\*(C'\fR and \f(CW\*(C`sys/wait.h\*(C'\fR documentation for details).
1600.PP
1601\fIExamples\fR
1602.IX Subsection "Examples"
1603.PP
1604Example: Try to exit cleanly on \s-1SIGINT\s0 and \s-1SIGTERM\s0.
1605.PP
1606.Vb 5
1607\& static void
1608\& sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1609\& {
1610\& ev_unloop (loop, EVUNLOOP_ALL);
1611\& }
1612\&
1613\& struct ev_signal signal_watcher;
1614\& ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1615\& ev_signal_start (loop, &sigint_cb);
1616.Ve
1617.ie n .Sh """ev_stat"" \- did the file attributes just change?"
1618.el .Sh "\f(CWev_stat\fP \- did the file attributes just change?"
1619.IX Subsection "ev_stat - did the file attributes just change?"
1620This watches a filesystem path for attribute changes. That is, it calls
1621\&\f(CW\*(C`stat\*(C'\fR regularly (or when the \s-1OS\s0 says it changed) and sees if it changed
1622compared to the last time, invoking the callback if it did.
1623.PP
1624The path does not need to exist: changing from \*(L"path exists\*(R" to \*(L"path does
1625not exist\*(R" is a status change like any other. The condition \*(L"path does
1626not exist\*(R" is signified by the \f(CW\*(C`st_nlink\*(C'\fR field being zero (which is
1627otherwise always forced to be at least one) and all the other fields of
1628the stat buffer having unspecified contents.
1629.PP
1630The path \fIshould\fR be absolute and \fImust not\fR end in a slash. If it is
1631relative and your working directory changes, the behaviour is undefined.
1632.PP
1633Since there is no standard to do this, the portable implementation simply
1634calls \f(CW\*(C`stat (2)\*(C'\fR regularly on the path to see if it changed somehow. You
1635can specify a recommended polling interval for this case. If you specify
1636a polling interval of \f(CW0\fR (highly recommended!) then a \fIsuitable,
1637unspecified default\fR value will be used (which you can expect to be around
1638five seconds, although this might change dynamically). Libev will also
1639impose a minimum interval which is currently around \f(CW0.1\fR, but thats
1640usually overkill.
1641.PP
1642This watcher type is not meant for massive numbers of stat watchers,
1643as even with OS-supported change notifications, this can be
1644resource-intensive.
1645.PP
1646At the time of this writing, only the Linux inotify interface is
1647implemented (implementing kqueue support is left as an exercise for the
1648reader). Inotify will be used to give hints only and should not change the
1649semantics of \f(CW\*(C`ev_stat\*(C'\fR watchers, which means that libev sometimes needs
1650to fall back to regular polling again even with inotify, but changes are
1651usually detected immediately, and if the file exists there will be no
1652polling.
1653.PP
1654\fIInotify\fR
1655.IX Subsection "Inotify"
1656.PP
1657When \f(CW\*(C`inotify (7)\*(C'\fR support has been compiled into libev (generally only
1658available on Linux) and present at runtime, it will be used to speed up
1659change detection where possible. The inotify descriptor will be created lazily
1660when the first \f(CW\*(C`ev_stat\*(C'\fR watcher is being started.
1661.PP
1662Inotify presense does not change the semantics of \f(CW\*(C`ev_stat\*(C'\fR watchers
1663except that changes might be detected earlier, and in some cases, to avoid
1664making regular \f(CW\*(C`stat\*(C'\fR calls. Even in the presense of inotify support
1665there are many cases where libev has to resort to regular \f(CW\*(C`stat\*(C'\fR polling.
1666.PP
1667(There is no support for kqueue, as apparently it cannot be used to
1668implement this functionality, due to the requirement of having a file
1669descriptor open on the object at all times).
1670.PP
1671\fIThe special problem of stat time resolution\fR
1672.IX Subsection "The special problem of stat time resolution"
1673.PP
1674The \f(CW\*(C`stat ()\*(C'\fR syscall only supports full-second resolution portably, and
1675even on systems where the resolution is higher, many filesystems still
1676only support whole seconds.
1677.PP
1678That means that, if the time is the only thing that changes, you might
1679miss updates: on the first update, \f(CW\*(C`ev_stat\*(C'\fR detects a change and calls
1680your callback, which does something. When there is another update within
1681the same second, \f(CW\*(C`ev_stat\*(C'\fR will be unable to detect it.
1682.PP
1683The solution to this is to delay acting on a change for a second (or till
1684the next second boundary), using a roughly one-second delay \f(CW\*(C`ev_timer\*(C'\fR
1685(\f(CW\*(C`ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)\*(C'\fR). The \f(CW.01\fR
1686is added to work around small timing inconsistencies of some operating
1687systems.
1688.PP
1689\fIWatcher-Specific Functions and Data Members\fR
1690.IX Subsection "Watcher-Specific Functions and Data Members"
1691.IP "ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)" 4
1692.IX Item "ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)"
1693.PD 0
1694.IP "ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)" 4
1695.IX Item "ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)"
1696.PD
1697Configures the watcher to wait for status changes of the given
1698\&\f(CW\*(C`path\*(C'\fR. The \f(CW\*(C`interval\*(C'\fR is a hint on how quickly a change is expected to
1699be detected and should normally be specified as \f(CW0\fR to let libev choose
1700a suitable value. The memory pointed to by \f(CW\*(C`path\*(C'\fR must point to the same
1701path for as long as the watcher is active.
1702.Sp
1703The callback will be receive \f(CW\*(C`EV_STAT\*(C'\fR when a change was detected,
1704relative to the attributes at the time the watcher was started (or the
1705last change was detected).
1706.IP "ev_stat_stat (ev_stat *)" 4
1707.IX Item "ev_stat_stat (ev_stat *)"
1708Updates the stat buffer immediately with new values. If you change the
1709watched path in your callback, you could call this fucntion to avoid
1710detecting this change (while introducing a race condition). Can also be
1711useful simply to find out the new values.
1712.IP "ev_statdata attr [read\-only]" 4
1713.IX Item "ev_statdata attr [read-only]"
1714The most-recently detected attributes of the file. Although the type is of
1715\&\f(CW\*(C`ev_statdata\*(C'\fR, this is usually the (or one of the) \f(CW\*(C`struct stat\*(C'\fR types
1716suitable for your system. If the \f(CW\*(C`st_nlink\*(C'\fR member is \f(CW0\fR, then there
1717was some error while \f(CW\*(C`stat\*(C'\fRing the file.
1718.IP "ev_statdata prev [read\-only]" 4
1719.IX Item "ev_statdata prev [read-only]"
1720The previous attributes of the file. The callback gets invoked whenever
1721\&\f(CW\*(C`prev\*(C'\fR != \f(CW\*(C`attr\*(C'\fR.
1722.IP "ev_tstamp interval [read\-only]" 4
1723.IX Item "ev_tstamp interval [read-only]"
1724The specified interval.
1725.IP "const char *path [read\-only]" 4
1726.IX Item "const char *path [read-only]"
1727The filesystem path that is being watched.
1728.PP
1729\fIExamples\fR
1730.IX Subsection "Examples"
1731.PP
1732Example: Watch \f(CW\*(C`/etc/passwd\*(C'\fR for attribute changes.
1733.PP
1734.Vb 10
1735\& static void
1736\& passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1737\& {
1738\& /* /etc/passwd changed in some way */
1739\& if (w\->attr.st_nlink)
1740\& {
1741\& printf ("passwd current size %ld\en", (long)w\->attr.st_size);
1742\& printf ("passwd current atime %ld\en", (long)w\->attr.st_mtime);
1743\& printf ("passwd current mtime %ld\en", (long)w\->attr.st_mtime);
1744\& }
1745\& else
1746\& /* you shalt not abuse printf for puts */
1747\& puts ("wow, /etc/passwd is not there, expect problems. "
1748\& "if this is windows, they already arrived\en");
1749\& }
1750\&
1751\& ...
1752\& ev_stat passwd;
1753\&
1754\& ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1755\& ev_stat_start (loop, &passwd);
1756.Ve
1757.PP
1758Example: Like above, but additionally use a one-second delay so we do not
1759miss updates (however, frequent updates will delay processing, too, so
1760one might do the work both on \f(CW\*(C`ev_stat\*(C'\fR callback invocation \fIand\fR on
1761\&\f(CW\*(C`ev_timer\*(C'\fR callback invocation).
1762.PP
1763.Vb 2
1764\& static ev_stat passwd;
1765\& static ev_timer timer;
1766\&
1767\& static void
1768\& timer_cb (EV_P_ ev_timer *w, int revents)
1769\& {
1770\& ev_timer_stop (EV_A_ w);
1771\&
1772\& /* now it\*(Aqs one second after the most recent passwd change */
1773\& }
1774\&
1775\& static void
1776\& stat_cb (EV_P_ ev_stat *w, int revents)
1777\& {
1778\& /* reset the one\-second timer */
1779\& ev_timer_again (EV_A_ &timer);
1780\& }
1781\&
1782\& ...
1783\& ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
1784\& ev_stat_start (loop, &passwd);
1785\& ev_timer_init (&timer, timer_cb, 0., 1.01);
1786.Ve
758.ie n .Sh """ev_idle"" \- when you've got nothing better to do" 1787.ie n .Sh """ev_idle"" \- when you've got nothing better to do..."
759.el .Sh "\f(CWev_idle\fP \- when you've got nothing better to do" 1788.el .Sh "\f(CWev_idle\fP \- when you've got nothing better to do..."
760.IX Subsection "ev_idle - when you've got nothing better to do" 1789.IX Subsection "ev_idle - when you've got nothing better to do..."
761Idle watchers trigger events when there are no other events are pending 1790Idle watchers trigger events when no other events of the same or higher
762(prepare, check and other idle watchers do not count). That is, as long 1791priority are pending (prepare, check and other idle watchers do not
763as your process is busy handling sockets or timeouts (or even signals, 1792count).
764imagine) it will not be triggered. But when your process is idle all idle 1793.PP
765watchers are being called again and again, once per event loop iteration \- 1794That is, as long as your process is busy handling sockets or timeouts
1795(or even signals, imagine) of the same or higher priority it will not be
1796triggered. But when your process is idle (or only lower-priority watchers
1797are pending), the idle watchers are being called once per event loop
766until stopped, that is, or your process receives more events and becomes 1798iteration \- until stopped, that is, or your process receives more events
767busy. 1799and becomes busy again with higher priority stuff.
768.PP 1800.PP
769The most noteworthy effect is that as long as any idle watchers are 1801The most noteworthy effect is that as long as any idle watchers are
770active, the process will not block when waiting for new events. 1802active, the process will not block when waiting for new events.
771.PP 1803.PP
772Apart from keeping your process non-blocking (which is a useful 1804Apart from keeping your process non-blocking (which is a useful
773effect on its own sometimes), idle watchers are a good place to do 1805effect on its own sometimes), idle watchers are a good place to do
774\&\*(L"pseudo\-background processing\*(R", or delay processing stuff to after the 1806\&\*(L"pseudo-background processing\*(R", or delay processing stuff to after the
775event loop has handled all outstanding events. 1807event loop has handled all outstanding events.
1808.PP
1809\fIWatcher-Specific Functions and Data Members\fR
1810.IX Subsection "Watcher-Specific Functions and Data Members"
776.IP "ev_idle_init (ev_signal *, callback)" 4 1811.IP "ev_idle_init (ev_signal *, callback)" 4
777.IX Item "ev_idle_init (ev_signal *, callback)" 1812.IX Item "ev_idle_init (ev_signal *, callback)"
778Initialises and configures the idle watcher \- it has no parameters of any 1813Initialises and configures the idle watcher \- it has no parameters of any
779kind. There is a \f(CW\*(C`ev_idle_set\*(C'\fR macro, but using it is utterly pointless, 1814kind. There is a \f(CW\*(C`ev_idle_set\*(C'\fR macro, but using it is utterly pointless,
780believe me. 1815believe me.
1816.PP
1817\fIExamples\fR
1818.IX Subsection "Examples"
1819.PP
1820Example: Dynamically allocate an \f(CW\*(C`ev_idle\*(C'\fR watcher, start it, and in the
1821callback, free it. Also, use no error checking, as usual.
1822.PP
1823.Vb 7
1824\& static void
1825\& idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1826\& {
1827\& free (w);
1828\& // now do something you wanted to do when the program has
1829\& // no longer anything immediate to do.
1830\& }
1831\&
1832\& struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1833\& ev_idle_init (idle_watcher, idle_cb);
1834\& ev_idle_start (loop, idle_cb);
1835.Ve
781.ie n .Sh """ev_prepare""\fP and \f(CW""ev_check"" \- customise your event loop" 1836.ie n .Sh """ev_prepare""\fP and \f(CW""ev_check"" \- customise your event loop!"
782.el .Sh "\f(CWev_prepare\fP and \f(CWev_check\fP \- customise your event loop" 1837.el .Sh "\f(CWev_prepare\fP and \f(CWev_check\fP \- customise your event loop!"
783.IX Subsection "ev_prepare and ev_check - customise your event loop" 1838.IX Subsection "ev_prepare and ev_check - customise your event loop!"
784Prepare and check watchers are usually (but not always) used in tandem: 1839Prepare and check watchers are usually (but not always) used in tandem:
785prepare watchers get invoked before the process blocks and check watchers 1840prepare watchers get invoked before the process blocks and check watchers
786afterwards. 1841afterwards.
787.PP 1842.PP
1843You \fImust not\fR call \f(CW\*(C`ev_loop\*(C'\fR or similar functions that enter
1844the current event loop from either \f(CW\*(C`ev_prepare\*(C'\fR or \f(CW\*(C`ev_check\*(C'\fR
1845watchers. Other loops than the current one are fine, however. The
1846rationale behind this is that you do not need to check for recursion in
1847those watchers, i.e. the sequence will always be \f(CW\*(C`ev_prepare\*(C'\fR, blocking,
1848\&\f(CW\*(C`ev_check\*(C'\fR so if you have one watcher of each kind they will always be
1849called in pairs bracketing the blocking call.
1850.PP
788Their main purpose is to integrate other event mechanisms into libev. This 1851Their main purpose is to integrate other event mechanisms into libev and
789could be used, for example, to track variable changes, implement your own 1852their use is somewhat advanced. This could be used, for example, to track
790watchers, integrate net-snmp or a coroutine library and lots more. 1853variable changes, implement your own watchers, integrate net-snmp or a
1854coroutine library and lots more. They are also occasionally useful if
1855you cache some data and want to flush it before blocking (for example,
1856in X programs you might want to do an \f(CW\*(C`XFlush ()\*(C'\fR in an \f(CW\*(C`ev_prepare\*(C'\fR
1857watcher).
791.PP 1858.PP
792This is done by examining in each prepare call which file descriptors need 1859This is done by examining in each prepare call which file descriptors need
793to be watched by the other library, registering \f(CW\*(C`ev_io\*(C'\fR watchers for 1860to be watched by the other library, registering \f(CW\*(C`ev_io\*(C'\fR watchers for
794them and starting an \f(CW\*(C`ev_timer\*(C'\fR watcher for any timeouts (many libraries 1861them and starting an \f(CW\*(C`ev_timer\*(C'\fR watcher for any timeouts (many libraries
795provide just this functionality). Then, in the check watcher you check for 1862provide just this functionality). Then, in the check watcher you check for
… …
804are ready to run (it's actually more complicated: it only runs coroutines 1871are ready to run (it's actually more complicated: it only runs coroutines
805with priority higher than or equal to the event loop and one coroutine 1872with priority higher than or equal to the event loop and one coroutine
806of lower priority, but only once, using idle watchers to keep the event 1873of lower priority, but only once, using idle watchers to keep the event
807loop from blocking if lower-priority coroutines are active, thus mapping 1874loop from blocking if lower-priority coroutines are active, thus mapping
808low-priority coroutines to idle/background tasks). 1875low-priority coroutines to idle/background tasks).
1876.PP
1877It is recommended to give \f(CW\*(C`ev_check\*(C'\fR watchers highest (\f(CW\*(C`EV_MAXPRI\*(C'\fR)
1878priority, to ensure that they are being run before any other watchers
1879after the poll. Also, \f(CW\*(C`ev_check\*(C'\fR watchers (and \f(CW\*(C`ev_prepare\*(C'\fR watchers,
1880too) should not activate (\*(L"feed\*(R") events into libev. While libev fully
1881supports this, they will be called before other \f(CW\*(C`ev_check\*(C'\fR watchers
1882did their job. As \f(CW\*(C`ev_check\*(C'\fR watchers are often used to embed other
1883(non-libev) event loops those other event loops might be in an unusable
1884state until their \f(CW\*(C`ev_check\*(C'\fR watcher ran (always remind yourself to
1885coexist peacefully with others).
1886.PP
1887\fIWatcher-Specific Functions and Data Members\fR
1888.IX Subsection "Watcher-Specific Functions and Data Members"
809.IP "ev_prepare_init (ev_prepare *, callback)" 4 1889.IP "ev_prepare_init (ev_prepare *, callback)" 4
810.IX Item "ev_prepare_init (ev_prepare *, callback)" 1890.IX Item "ev_prepare_init (ev_prepare *, callback)"
811.PD 0 1891.PD 0
812.IP "ev_check_init (ev_check *, callback)" 4 1892.IP "ev_check_init (ev_check *, callback)" 4
813.IX Item "ev_check_init (ev_check *, callback)" 1893.IX Item "ev_check_init (ev_check *, callback)"
814.PD 1894.PD
815Initialises and configures the prepare or check watcher \- they have no 1895Initialises and configures the prepare or check watcher \- they have no
816parameters of any kind. There are \f(CW\*(C`ev_prepare_set\*(C'\fR and \f(CW\*(C`ev_check_set\*(C'\fR 1896parameters of any kind. There are \f(CW\*(C`ev_prepare_set\*(C'\fR and \f(CW\*(C`ev_check_set\*(C'\fR
817macros, but using them is utterly, utterly and completely pointless. 1897macros, but using them is utterly, utterly and completely pointless.
1898.PP
1899\fIExamples\fR
1900.IX Subsection "Examples"
1901.PP
1902There are a number of principal ways to embed other event loops or modules
1903into libev. Here are some ideas on how to include libadns into libev
1904(there is a Perl module named \f(CW\*(C`EV::ADNS\*(C'\fR that does this, which you could
1905use for an actually working example. Another Perl module named \f(CW\*(C`EV::Glib\*(C'\fR
1906embeds a Glib main context into libev, and finally, \f(CW\*(C`Glib::EV\*(C'\fR embeds \s-1EV\s0
1907into the Glib event loop).
1908.PP
1909Method 1: Add \s-1IO\s0 watchers and a timeout watcher in a prepare handler,
1910and in a check watcher, destroy them and call into libadns. What follows
1911is pseudo-code only of course. This requires you to either use a low
1912priority for the check watcher or use \f(CW\*(C`ev_clear_pending\*(C'\fR explicitly, as
1913the callbacks for the IO/timeout watchers might not have been called yet.
1914.PP
1915.Vb 2
1916\& static ev_io iow [nfd];
1917\& static ev_timer tw;
1918\&
1919\& static void
1920\& io_cb (ev_loop *loop, ev_io *w, int revents)
1921\& {
1922\& }
1923\&
1924\& // create io watchers for each fd and a timer before blocking
1925\& static void
1926\& adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1927\& {
1928\& int timeout = 3600000;
1929\& struct pollfd fds [nfd];
1930\& // actual code will need to loop here and realloc etc.
1931\& adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
1932\&
1933\& /* the callback is illegal, but won\*(Aqt be called as we stop during check */
1934\& ev_timer_init (&tw, 0, timeout * 1e\-3);
1935\& ev_timer_start (loop, &tw);
1936\&
1937\& // create one ev_io per pollfd
1938\& for (int i = 0; i < nfd; ++i)
1939\& {
1940\& ev_io_init (iow + i, io_cb, fds [i].fd,
1941\& ((fds [i].events & POLLIN ? EV_READ : 0)
1942\& | (fds [i].events & POLLOUT ? EV_WRITE : 0)));
1943\&
1944\& fds [i].revents = 0;
1945\& ev_io_start (loop, iow + i);
1946\& }
1947\& }
1948\&
1949\& // stop all watchers after blocking
1950\& static void
1951\& adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1952\& {
1953\& ev_timer_stop (loop, &tw);
1954\&
1955\& for (int i = 0; i < nfd; ++i)
1956\& {
1957\& // set the relevant poll flags
1958\& // could also call adns_processreadable etc. here
1959\& struct pollfd *fd = fds + i;
1960\& int revents = ev_clear_pending (iow + i);
1961\& if (revents & EV_READ ) fd\->revents |= fd\->events & POLLIN;
1962\& if (revents & EV_WRITE) fd\->revents |= fd\->events & POLLOUT;
1963\&
1964\& // now stop the watcher
1965\& ev_io_stop (loop, iow + i);
1966\& }
1967\&
1968\& adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1969\& }
1970.Ve
1971.PP
1972Method 2: This would be just like method 1, but you run \f(CW\*(C`adns_afterpoll\*(C'\fR
1973in the prepare watcher and would dispose of the check watcher.
1974.PP
1975Method 3: If the module to be embedded supports explicit event
1976notification (adns does), you can also make use of the actual watcher
1977callbacks, and only destroy/create the watchers in the prepare watcher.
1978.PP
1979.Vb 5
1980\& static void
1981\& timer_cb (EV_P_ ev_timer *w, int revents)
1982\& {
1983\& adns_state ads = (adns_state)w\->data;
1984\& update_now (EV_A);
1985\&
1986\& adns_processtimeouts (ads, &tv_now);
1987\& }
1988\&
1989\& static void
1990\& io_cb (EV_P_ ev_io *w, int revents)
1991\& {
1992\& adns_state ads = (adns_state)w\->data;
1993\& update_now (EV_A);
1994\&
1995\& if (revents & EV_READ ) adns_processreadable (ads, w\->fd, &tv_now);
1996\& if (revents & EV_WRITE) adns_processwriteable (ads, w\->fd, &tv_now);
1997\& }
1998\&
1999\& // do not ever call adns_afterpoll
2000.Ve
2001.PP
2002Method 4: Do not use a prepare or check watcher because the module you
2003want to embed is too inflexible to support it. Instead, youc na override
2004their poll function. The drawback with this solution is that the main
2005loop is now no longer controllable by \s-1EV\s0. The \f(CW\*(C`Glib::EV\*(C'\fR module does
2006this.
2007.PP
2008.Vb 4
2009\& static gint
2010\& event_poll_func (GPollFD *fds, guint nfds, gint timeout)
2011\& {
2012\& int got_events = 0;
2013\&
2014\& for (n = 0; n < nfds; ++n)
2015\& // create/start io watcher that sets the relevant bits in fds[n] and increment got_events
2016\&
2017\& if (timeout >= 0)
2018\& // create/start timer
2019\&
2020\& // poll
2021\& ev_loop (EV_A_ 0);
2022\&
2023\& // stop timer again
2024\& if (timeout >= 0)
2025\& ev_timer_stop (EV_A_ &to);
2026\&
2027\& // stop io watchers again \- their callbacks should have set
2028\& for (n = 0; n < nfds; ++n)
2029\& ev_io_stop (EV_A_ iow [n]);
2030\&
2031\& return got_events;
2032\& }
2033.Ve
2034.ie n .Sh """ev_embed"" \- when one backend isn't enough..."
2035.el .Sh "\f(CWev_embed\fP \- when one backend isn't enough..."
2036.IX Subsection "ev_embed - when one backend isn't enough..."
2037This is a rather advanced watcher type that lets you embed one event loop
2038into another (currently only \f(CW\*(C`ev_io\*(C'\fR events are supported in the embedded
2039loop, other types of watchers might be handled in a delayed or incorrect
2040fashion and must not be used).
2041.PP
2042There are primarily two reasons you would want that: work around bugs and
2043prioritise I/O.
2044.PP
2045As an example for a bug workaround, the kqueue backend might only support
2046sockets on some platform, so it is unusable as generic backend, but you
2047still want to make use of it because you have many sockets and it scales
2048so nicely. In this case, you would create a kqueue-based loop and embed it
2049into your default loop (which might use e.g. poll). Overall operation will
2050be a bit slower because first libev has to poll and then call kevent, but
2051at least you can use both at what they are best.
2052.PP
2053As for prioritising I/O: rarely you have the case where some fds have
2054to be watched and handled very quickly (with low latency), and even
2055priorities and idle watchers might have too much overhead. In this case
2056you would put all the high priority stuff in one loop and all the rest in
2057a second one, and embed the second one in the first.
2058.PP
2059As long as the watcher is active, the callback will be invoked every time
2060there might be events pending in the embedded loop. The callback must then
2061call \f(CW\*(C`ev_embed_sweep (mainloop, watcher)\*(C'\fR to make a single sweep and invoke
2062their callbacks (you could also start an idle watcher to give the embedded
2063loop strictly lower priority for example). You can also set the callback
2064to \f(CW0\fR, in which case the embed watcher will automatically execute the
2065embedded loop sweep.
2066.PP
2067As long as the watcher is started it will automatically handle events. The
2068callback will be invoked whenever some events have been handled. You can
2069set the callback to \f(CW0\fR to avoid having to specify one if you are not
2070interested in that.
2071.PP
2072Also, there have not currently been made special provisions for forking:
2073when you fork, you not only have to call \f(CW\*(C`ev_loop_fork\*(C'\fR on both loops,
2074but you will also have to stop and restart any \f(CW\*(C`ev_embed\*(C'\fR watchers
2075yourself.
2076.PP
2077Unfortunately, not all backends are embeddable, only the ones returned by
2078\&\f(CW\*(C`ev_embeddable_backends\*(C'\fR are, which, unfortunately, does not include any
2079portable one.
2080.PP
2081So when you want to use this feature you will always have to be prepared
2082that you cannot get an embeddable loop. The recommended way to get around
2083this is to have a separate variables for your embeddable loop, try to
2084create it, and if that fails, use the normal loop for everything.
2085.PP
2086\fIWatcher-Specific Functions and Data Members\fR
2087.IX Subsection "Watcher-Specific Functions and Data Members"
2088.IP "ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)" 4
2089.IX Item "ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)"
2090.PD 0
2091.IP "ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)" 4
2092.IX Item "ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)"
2093.PD
2094Configures the watcher to embed the given loop, which must be
2095embeddable. If the callback is \f(CW0\fR, then \f(CW\*(C`ev_embed_sweep\*(C'\fR will be
2096invoked automatically, otherwise it is the responsibility of the callback
2097to invoke it (it will continue to be called until the sweep has been done,
2098if you do not want thta, you need to temporarily stop the embed watcher).
2099.IP "ev_embed_sweep (loop, ev_embed *)" 4
2100.IX Item "ev_embed_sweep (loop, ev_embed *)"
2101Make a single, non-blocking sweep over the embedded loop. This works
2102similarly to \f(CW\*(C`ev_loop (embedded_loop, EVLOOP_NONBLOCK)\*(C'\fR, but in the most
2103apropriate way for embedded loops.
2104.IP "struct ev_loop *other [read\-only]" 4
2105.IX Item "struct ev_loop *other [read-only]"
2106The embedded event loop.
2107.PP
2108\fIExamples\fR
2109.IX Subsection "Examples"
2110.PP
2111Example: Try to get an embeddable event loop and embed it into the default
2112event loop. If that is not possible, use the default loop. The default
2113loop is stored in \f(CW\*(C`loop_hi\*(C'\fR, while the mebeddable loop is stored in
2114\&\f(CW\*(C`loop_lo\*(C'\fR (which is \f(CW\*(C`loop_hi\*(C'\fR in the acse no embeddable loop can be
2115used).
2116.PP
2117.Vb 3
2118\& struct ev_loop *loop_hi = ev_default_init (0);
2119\& struct ev_loop *loop_lo = 0;
2120\& struct ev_embed embed;
2121\&
2122\& // see if there is a chance of getting one that works
2123\& // (remember that a flags value of 0 means autodetection)
2124\& loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
2125\& ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
2126\& : 0;
2127\&
2128\& // if we got one, then embed it, otherwise default to loop_hi
2129\& if (loop_lo)
2130\& {
2131\& ev_embed_init (&embed, 0, loop_lo);
2132\& ev_embed_start (loop_hi, &embed);
2133\& }
2134\& else
2135\& loop_lo = loop_hi;
2136.Ve
2137.PP
2138Example: Check if kqueue is available but not recommended and create
2139a kqueue backend for use with sockets (which usually work with any
2140kqueue implementation). Store the kqueue/socket\-only event loop in
2141\&\f(CW\*(C`loop_socket\*(C'\fR. (One might optionally use \f(CW\*(C`EVFLAG_NOENV\*(C'\fR, too).
2142.PP
2143.Vb 3
2144\& struct ev_loop *loop = ev_default_init (0);
2145\& struct ev_loop *loop_socket = 0;
2146\& struct ev_embed embed;
2147\&
2148\& if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
2149\& if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
2150\& {
2151\& ev_embed_init (&embed, 0, loop_socket);
2152\& ev_embed_start (loop, &embed);
2153\& }
2154\&
2155\& if (!loop_socket)
2156\& loop_socket = loop;
2157\&
2158\& // now use loop_socket for all sockets, and loop for everything else
2159.Ve
2160.ie n .Sh """ev_fork"" \- the audacity to resume the event loop after a fork"
2161.el .Sh "\f(CWev_fork\fP \- the audacity to resume the event loop after a fork"
2162.IX Subsection "ev_fork - the audacity to resume the event loop after a fork"
2163Fork watchers are called when a \f(CW\*(C`fork ()\*(C'\fR was detected (usually because
2164whoever is a good citizen cared to tell libev about it by calling
2165\&\f(CW\*(C`ev_default_fork\*(C'\fR or \f(CW\*(C`ev_loop_fork\*(C'\fR). The invocation is done before the
2166event loop blocks next and before \f(CW\*(C`ev_check\*(C'\fR watchers are being called,
2167and only in the child after the fork. If whoever good citizen calling
2168\&\f(CW\*(C`ev_default_fork\*(C'\fR cheats and calls it in the wrong process, the fork
2169handlers will be invoked, too, of course.
2170.PP
2171\fIWatcher-Specific Functions and Data Members\fR
2172.IX Subsection "Watcher-Specific Functions and Data Members"
2173.IP "ev_fork_init (ev_signal *, callback)" 4
2174.IX Item "ev_fork_init (ev_signal *, callback)"
2175Initialises and configures the fork watcher \- it has no parameters of any
2176kind. There is a \f(CW\*(C`ev_fork_set\*(C'\fR macro, but using it is utterly pointless,
2177believe me.
818.SH "OTHER FUNCTIONS" 2178.SH "OTHER FUNCTIONS"
819.IX Header "OTHER FUNCTIONS" 2179.IX Header "OTHER FUNCTIONS"
820There are some other functions of possible interest. Described. Here. Now. 2180There are some other functions of possible interest. Described. Here. Now.
821.IP "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 4 2181.IP "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 4
822.IX Item "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 2182.IX Item "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)"
… …
846\& if (revents & EV_TIMEOUT) 2206\& if (revents & EV_TIMEOUT)
847\& /* doh, nothing entered */; 2207\& /* doh, nothing entered */;
848\& else if (revents & EV_READ) 2208\& else if (revents & EV_READ)
849\& /* stdin might have data for us, joy! */; 2209\& /* stdin might have data for us, joy! */;
850\& } 2210\& }
851.Ve 2211\&
852.Sp
853.Vb 1
854\& ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0); 2212\& ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
855.Ve 2213.Ve
856.IP "ev_feed_event (loop, watcher, int events)" 4 2214.IP "ev_feed_event (ev_loop *, watcher *, int revents)" 4
857.IX Item "ev_feed_event (loop, watcher, int events)" 2215.IX Item "ev_feed_event (ev_loop *, watcher *, int revents)"
858Feeds the given event set into the event loop, as if the specified event 2216Feeds the given event set into the event loop, as if the specified event
859had happened for the specified watcher (which must be a pointer to an 2217had happened for the specified watcher (which must be a pointer to an
860initialised but not necessarily started event watcher). 2218initialised but not necessarily started event watcher).
861.IP "ev_feed_fd_event (loop, int fd, int revents)" 4 2219.IP "ev_feed_fd_event (ev_loop *, int fd, int revents)" 4
862.IX Item "ev_feed_fd_event (loop, int fd, int revents)" 2220.IX Item "ev_feed_fd_event (ev_loop *, int fd, int revents)"
863Feed an event on the given fd, as if a file descriptor backend detected 2221Feed an event on the given fd, as if a file descriptor backend detected
864the given events it. 2222the given events it.
865.IP "ev_feed_signal_event (loop, int signum)" 4 2223.IP "ev_feed_signal_event (ev_loop *loop, int signum)" 4
866.IX Item "ev_feed_signal_event (loop, int signum)" 2224.IX Item "ev_feed_signal_event (ev_loop *loop, int signum)"
867Feed an event as if the given signal occured (loop must be the default loop!). 2225Feed an event as if the given signal occured (\f(CW\*(C`loop\*(C'\fR must be the default
2226loop!).
868.SH "LIBEVENT EMULATION" 2227.SH "LIBEVENT EMULATION"
869.IX Header "LIBEVENT EMULATION" 2228.IX Header "LIBEVENT EMULATION"
870Libev offers a compatibility emulation layer for libevent. It cannot 2229Libev offers a compatibility emulation layer for libevent. It cannot
871emulate the internals of libevent, so here are some usage hints: 2230emulate the internals of libevent, so here are some usage hints:
2231.IP "\(bu" 4
872.IP "* Use it by including <event.h>, as usual." 4 2232Use it by including <event.h>, as usual.
873.IX Item "Use it by including <event.h>, as usual." 2233.IP "\(bu" 4
874.PD 0 2234The following members are fully supported: ev_base, ev_callback,
875.IP "* The following members are fully supported: ev_base, ev_callback, ev_arg, ev_fd, ev_res, ev_events." 4 2235ev_arg, ev_fd, ev_res, ev_events.
876.IX Item "The following members are fully supported: ev_base, ev_callback, ev_arg, ev_fd, ev_res, ev_events." 2236.IP "\(bu" 4
877.IP "* Avoid using ev_flags and the EVLIST_*\-macros, while it is maintained by libev, it does not work exactly the same way as in libevent (consider it a private \s-1API\s0)." 4 2237Avoid using ev_flags and the EVLIST_*\-macros, while it is
878.IX Item "Avoid using ev_flags and the EVLIST_*-macros, while it is maintained by libev, it does not work exactly the same way as in libevent (consider it a private API)." 2238maintained by libev, it does not work exactly the same way as in libevent (consider
879.IP "* Priorities are not currently supported. Initialising priorities will fail and all watchers will have the same priority, even though there is an ev_pri field." 4 2239it a private \s-1API\s0).
880.IX Item "Priorities are not currently supported. Initialising priorities will fail and all watchers will have the same priority, even though there is an ev_pri field." 2240.IP "\(bu" 4
2241Priorities are not currently supported. Initialising priorities
2242will fail and all watchers will have the same priority, even though there
2243is an ev_pri field.
2244.IP "\(bu" 4
881.IP "* Other members are not supported." 4 2245Other members are not supported.
882.IX Item "Other members are not supported." 2246.IP "\(bu" 4
883.IP "* The libev emulation is \fInot\fR \s-1ABI\s0 compatible to libevent, you need to use the libev header file and library." 4 2247The libev emulation is \fInot\fR \s-1ABI\s0 compatible to libevent, you need
884.IX Item "The libev emulation is not ABI compatible to libevent, you need to use the libev header file and library." 2248to use the libev header file and library.
885.PD
886.SH "\*(C+ SUPPORT" 2249.SH "\*(C+ SUPPORT"
887.IX Header " SUPPORT" 2250.IX Header " SUPPORT"
888\&\s-1TBD\s0. 2251Libev comes with some simplistic wrapper classes for \*(C+ that mainly allow
2252you to use some convinience methods to start/stop watchers and also change
2253the callback model to a model using method callbacks on objects.
2254.PP
2255To use it,
2256.PP
2257.Vb 1
2258\& #include <ev++.h>
2259.Ve
2260.PP
2261This automatically includes \fIev.h\fR and puts all of its definitions (many
2262of them macros) into the global namespace. All \*(C+ specific things are
2263put into the \f(CW\*(C`ev\*(C'\fR namespace. It should support all the same embedding
2264options as \fIev.h\fR, most notably \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR.
2265.PP
2266Care has been taken to keep the overhead low. The only data member the \*(C+
2267classes add (compared to plain C\-style watchers) is the event loop pointer
2268that the watcher is associated with (or no additional members at all if
2269you disable \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR when embedding libev).
2270.PP
2271Currently, functions, and static and non-static member functions can be
2272used as callbacks. Other types should be easy to add as long as they only
2273need one additional pointer for context. If you need support for other
2274types of functors please contact the author (preferably after implementing
2275it).
2276.PP
2277Here is a list of things available in the \f(CW\*(C`ev\*(C'\fR namespace:
2278.ie n .IP """ev::READ""\fR, \f(CW""ev::WRITE"" etc." 4
2279.el .IP "\f(CWev::READ\fR, \f(CWev::WRITE\fR etc." 4
2280.IX Item "ev::READ, ev::WRITE etc."
2281These are just enum values with the same values as the \f(CW\*(C`EV_READ\*(C'\fR etc.
2282macros from \fIev.h\fR.
2283.ie n .IP """ev::tstamp""\fR, \f(CW""ev::now""" 4
2284.el .IP "\f(CWev::tstamp\fR, \f(CWev::now\fR" 4
2285.IX Item "ev::tstamp, ev::now"
2286Aliases to the same types/functions as with the \f(CW\*(C`ev_\*(C'\fR prefix.
2287.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
2288.el .IP "\f(CWev::io\fR, \f(CWev::timer\fR, \f(CWev::periodic\fR, \f(CWev::idle\fR, \f(CWev::sig\fR etc." 4
2289.IX Item "ev::io, ev::timer, ev::periodic, ev::idle, ev::sig etc."
2290For each \f(CW\*(C`ev_TYPE\*(C'\fR watcher in \fIev.h\fR there is a corresponding class of
2291the same name in the \f(CW\*(C`ev\*(C'\fR namespace, with the exception of \f(CW\*(C`ev_signal\*(C'\fR
2292which is called \f(CW\*(C`ev::sig\*(C'\fR to avoid clashes with the \f(CW\*(C`signal\*(C'\fR macro
2293defines by many implementations.
2294.Sp
2295All of those classes have these methods:
2296.RS 4
2297.IP "ev::TYPE::TYPE ()" 4
2298.IX Item "ev::TYPE::TYPE ()"
2299.PD 0
2300.IP "ev::TYPE::TYPE (struct ev_loop *)" 4
2301.IX Item "ev::TYPE::TYPE (struct ev_loop *)"
2302.IP "ev::TYPE::~TYPE" 4
2303.IX Item "ev::TYPE::~TYPE"
2304.PD
2305The constructor (optionally) takes an event loop to associate the watcher
2306with. If it is omitted, it will use \f(CW\*(C`EV_DEFAULT\*(C'\fR.
2307.Sp
2308The constructor calls \f(CW\*(C`ev_init\*(C'\fR for you, which means you have to call the
2309\&\f(CW\*(C`set\*(C'\fR method before starting it.
2310.Sp
2311It will not set a callback, however: You have to call the templated \f(CW\*(C`set\*(C'\fR
2312method to set a callback before you can start the watcher.
2313.Sp
2314(The reason why you have to use a method is a limitation in \*(C+ which does
2315not allow explicit template arguments for constructors).
2316.Sp
2317The destructor automatically stops the watcher if it is active.
2318.IP "w\->set<class, &class::method> (object *)" 4
2319.IX Item "w->set<class, &class::method> (object *)"
2320This method sets the callback method to call. The method has to have a
2321signature of \f(CW\*(C`void (*)(ev_TYPE &, int)\*(C'\fR, it receives the watcher as
2322first argument and the \f(CW\*(C`revents\*(C'\fR as second. The object must be given as
2323parameter and is stored in the \f(CW\*(C`data\*(C'\fR member of the watcher.
2324.Sp
2325This method synthesizes efficient thunking code to call your method from
2326the C callback that libev requires. If your compiler can inline your
2327callback (i.e. it is visible to it at the place of the \f(CW\*(C`set\*(C'\fR call and
2328your compiler is good :), then the method will be fully inlined into the
2329thunking function, making it as fast as a direct C callback.
2330.Sp
2331Example: simple class declaration and watcher initialisation
2332.Sp
2333.Vb 4
2334\& struct myclass
2335\& {
2336\& void io_cb (ev::io &w, int revents) { }
2337\& }
2338\&
2339\& myclass obj;
2340\& ev::io iow;
2341\& iow.set <myclass, &myclass::io_cb> (&obj);
2342.Ve
2343.IP "w\->set<function> (void *data = 0)" 4
2344.IX Item "w->set<function> (void *data = 0)"
2345Also sets a callback, but uses a static method or plain function as
2346callback. The optional \f(CW\*(C`data\*(C'\fR argument will be stored in the watcher's
2347\&\f(CW\*(C`data\*(C'\fR member and is free for you to use.
2348.Sp
2349The prototype of the \f(CW\*(C`function\*(C'\fR must be \f(CW\*(C`void (*)(ev::TYPE &w, int)\*(C'\fR.
2350.Sp
2351See the method\-\f(CW\*(C`set\*(C'\fR above for more details.
2352.Sp
2353Example:
2354.Sp
2355.Vb 2
2356\& static void io_cb (ev::io &w, int revents) { }
2357\& iow.set <io_cb> ();
2358.Ve
2359.IP "w\->set (struct ev_loop *)" 4
2360.IX Item "w->set (struct ev_loop *)"
2361Associates a different \f(CW\*(C`struct ev_loop\*(C'\fR with this watcher. You can only
2362do this when the watcher is inactive (and not pending either).
2363.IP "w\->set ([args])" 4
2364.IX Item "w->set ([args])"
2365Basically the same as \f(CW\*(C`ev_TYPE_set\*(C'\fR, with the same args. Must be
2366called at least once. Unlike the C counterpart, an active watcher gets
2367automatically stopped and restarted when reconfiguring it with this
2368method.
2369.IP "w\->start ()" 4
2370.IX Item "w->start ()"
2371Starts the watcher. Note that there is no \f(CW\*(C`loop\*(C'\fR argument, as the
2372constructor already stores the event loop.
2373.IP "w\->stop ()" 4
2374.IX Item "w->stop ()"
2375Stops the watcher if it is active. Again, no \f(CW\*(C`loop\*(C'\fR argument.
2376.ie n .IP "w\->again () (""ev::timer""\fR, \f(CW""ev::periodic"" only)" 4
2377.el .IP "w\->again () (\f(CWev::timer\fR, \f(CWev::periodic\fR only)" 4
2378.IX Item "w->again () (ev::timer, ev::periodic only)"
2379For \f(CW\*(C`ev::timer\*(C'\fR and \f(CW\*(C`ev::periodic\*(C'\fR, this invokes the corresponding
2380\&\f(CW\*(C`ev_TYPE_again\*(C'\fR function.
2381.ie n .IP "w\->sweep () (""ev::embed"" only)" 4
2382.el .IP "w\->sweep () (\f(CWev::embed\fR only)" 4
2383.IX Item "w->sweep () (ev::embed only)"
2384Invokes \f(CW\*(C`ev_embed_sweep\*(C'\fR.
2385.ie n .IP "w\->update () (""ev::stat"" only)" 4
2386.el .IP "w\->update () (\f(CWev::stat\fR only)" 4
2387.IX Item "w->update () (ev::stat only)"
2388Invokes \f(CW\*(C`ev_stat_stat\*(C'\fR.
2389.RE
2390.RS 4
2391.RE
2392.PP
2393Example: Define a class with an \s-1IO\s0 and idle watcher, start one of them in
2394the constructor.
2395.PP
2396.Vb 4
2397\& class myclass
2398\& {
2399\& ev::io io; void io_cb (ev::io &w, int revents);
2400\& ev:idle idle void idle_cb (ev::idle &w, int revents);
2401\&
2402\& myclass (int fd)
2403\& {
2404\& io .set <myclass, &myclass::io_cb > (this);
2405\& idle.set <myclass, &myclass::idle_cb> (this);
2406\&
2407\& io.start (fd, ev::READ);
2408\& }
2409\& };
2410.Ve
2411.SH "MACRO MAGIC"
2412.IX Header "MACRO MAGIC"
2413Libev can be compiled with a variety of options, the most fundamantal
2414of which is \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR. This option determines whether (most)
2415functions and callbacks have an initial \f(CW\*(C`struct ev_loop *\*(C'\fR argument.
2416.PP
2417To make it easier to write programs that cope with either variant, the
2418following macros are defined:
2419.ie n .IP """EV_A""\fR, \f(CW""EV_A_""" 4
2420.el .IP "\f(CWEV_A\fR, \f(CWEV_A_\fR" 4
2421.IX Item "EV_A, EV_A_"
2422This provides the loop \fIargument\fR for functions, if one is required (\*(L"ev
2423loop argument\*(R"). The \f(CW\*(C`EV_A\*(C'\fR form is used when this is the sole argument,
2424\&\f(CW\*(C`EV_A_\*(C'\fR is used when other arguments are following. Example:
2425.Sp
2426.Vb 3
2427\& ev_unref (EV_A);
2428\& ev_timer_add (EV_A_ watcher);
2429\& ev_loop (EV_A_ 0);
2430.Ve
2431.Sp
2432It assumes the variable \f(CW\*(C`loop\*(C'\fR of type \f(CW\*(C`struct ev_loop *\*(C'\fR is in scope,
2433which is often provided by the following macro.
2434.ie n .IP """EV_P""\fR, \f(CW""EV_P_""" 4
2435.el .IP "\f(CWEV_P\fR, \f(CWEV_P_\fR" 4
2436.IX Item "EV_P, EV_P_"
2437This provides the loop \fIparameter\fR for functions, if one is required (\*(L"ev
2438loop parameter\*(R"). The \f(CW\*(C`EV_P\*(C'\fR form is used when this is the sole parameter,
2439\&\f(CW\*(C`EV_P_\*(C'\fR is used when other parameters are following. Example:
2440.Sp
2441.Vb 2
2442\& // this is how ev_unref is being declared
2443\& static void ev_unref (EV_P);
2444\&
2445\& // this is how you can declare your typical callback
2446\& static void cb (EV_P_ ev_timer *w, int revents)
2447.Ve
2448.Sp
2449It declares a parameter \f(CW\*(C`loop\*(C'\fR of type \f(CW\*(C`struct ev_loop *\*(C'\fR, quite
2450suitable for use with \f(CW\*(C`EV_A\*(C'\fR.
2451.ie n .IP """EV_DEFAULT""\fR, \f(CW""EV_DEFAULT_""" 4
2452.el .IP "\f(CWEV_DEFAULT\fR, \f(CWEV_DEFAULT_\fR" 4
2453.IX Item "EV_DEFAULT, EV_DEFAULT_"
2454Similar to the other two macros, this gives you the value of the default
2455loop, if multiple loops are supported (\*(L"ev loop default\*(R").
2456.PP
2457Example: Declare and initialise a check watcher, utilising the above
2458macros so it will work regardless of whether multiple loops are supported
2459or not.
2460.PP
2461.Vb 5
2462\& static void
2463\& check_cb (EV_P_ ev_timer *w, int revents)
2464\& {
2465\& ev_check_stop (EV_A_ w);
2466\& }
2467\&
2468\& ev_check check;
2469\& ev_check_init (&check, check_cb);
2470\& ev_check_start (EV_DEFAULT_ &check);
2471\& ev_loop (EV_DEFAULT_ 0);
2472.Ve
2473.SH "EMBEDDING"
2474.IX Header "EMBEDDING"
2475Libev can (and often is) directly embedded into host
2476applications. Examples of applications that embed it include the Deliantra
2477Game Server, the \s-1EV\s0 perl module, the \s-1GNU\s0 Virtual Private Ethernet (gvpe)
2478and rxvt-unicode.
2479.PP
2480The goal is to enable you to just copy the necessary files into your
2481source directory without having to change even a single line in them, so
2482you can easily upgrade by simply copying (or having a checked-out copy of
2483libev somewhere in your source tree).
2484.Sh "\s-1FILESETS\s0"
2485.IX Subsection "FILESETS"
2486Depending on what features you need you need to include one or more sets of files
2487in your app.
2488.PP
2489\fI\s-1CORE\s0 \s-1EVENT\s0 \s-1LOOP\s0\fR
2490.IX Subsection "CORE EVENT LOOP"
2491.PP
2492To include only the libev core (all the \f(CW\*(C`ev_*\*(C'\fR functions), with manual
2493configuration (no autoconf):
2494.PP
2495.Vb 2
2496\& #define EV_STANDALONE 1
2497\& #include "ev.c"
2498.Ve
2499.PP
2500This will automatically include \fIev.h\fR, too, and should be done in a
2501single C source file only to provide the function implementations. To use
2502it, do the same for \fIev.h\fR in all files wishing to use this \s-1API\s0 (best
2503done by writing a wrapper around \fIev.h\fR that you can include instead and
2504where you can put other configuration options):
2505.PP
2506.Vb 2
2507\& #define EV_STANDALONE 1
2508\& #include "ev.h"
2509.Ve
2510.PP
2511Both header files and implementation files can be compiled with a \*(C+
2512compiler (at least, thats a stated goal, and breakage will be treated
2513as a bug).
2514.PP
2515You need the following files in your source tree, or in a directory
2516in your include path (e.g. in libev/ when using \-Ilibev):
2517.PP
2518.Vb 4
2519\& ev.h
2520\& ev.c
2521\& ev_vars.h
2522\& ev_wrap.h
2523\&
2524\& ev_win32.c required on win32 platforms only
2525\&
2526\& ev_select.c only when select backend is enabled (which is enabled by default)
2527\& ev_poll.c only when poll backend is enabled (disabled by default)
2528\& ev_epoll.c only when the epoll backend is enabled (disabled by default)
2529\& ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2530\& ev_port.c only when the solaris port backend is enabled (disabled by default)
2531.Ve
2532.PP
2533\&\fIev.c\fR includes the backend files directly when enabled, so you only need
2534to compile this single file.
2535.PP
2536\fI\s-1LIBEVENT\s0 \s-1COMPATIBILITY\s0 \s-1API\s0\fR
2537.IX Subsection "LIBEVENT COMPATIBILITY API"
2538.PP
2539To include the libevent compatibility \s-1API\s0, also include:
2540.PP
2541.Vb 1
2542\& #include "event.c"
2543.Ve
2544.PP
2545in the file including \fIev.c\fR, and:
2546.PP
2547.Vb 1
2548\& #include "event.h"
2549.Ve
2550.PP
2551in the files that want to use the libevent \s-1API\s0. This also includes \fIev.h\fR.
2552.PP
2553You need the following additional files for this:
2554.PP
2555.Vb 2
2556\& event.h
2557\& event.c
2558.Ve
2559.PP
2560\fI\s-1AUTOCONF\s0 \s-1SUPPORT\s0\fR
2561.IX Subsection "AUTOCONF SUPPORT"
2562.PP
2563Instead of using \f(CW\*(C`EV_STANDALONE=1\*(C'\fR and providing your config in
2564whatever way you want, you can also \f(CW\*(C`m4_include([libev.m4])\*(C'\fR in your
2565\&\fIconfigure.ac\fR and leave \f(CW\*(C`EV_STANDALONE\*(C'\fR undefined. \fIev.c\fR will then
2566include \fIconfig.h\fR and configure itself accordingly.
2567.PP
2568For this of course you need the m4 file:
2569.PP
2570.Vb 1
2571\& libev.m4
2572.Ve
2573.Sh "\s-1PREPROCESSOR\s0 \s-1SYMBOLS/MACROS\s0"
2574.IX Subsection "PREPROCESSOR SYMBOLS/MACROS"
2575Libev can be configured via a variety of preprocessor symbols you have to define
2576before including any of its files. The default is not to build for multiplicity
2577and only include the select backend.
2578.IP "\s-1EV_STANDALONE\s0" 4
2579.IX Item "EV_STANDALONE"
2580Must always be \f(CW1\fR if you do not use autoconf configuration, which
2581keeps libev from including \fIconfig.h\fR, and it also defines dummy
2582implementations for some libevent functions (such as logging, which is not
2583supported). It will also not define any of the structs usually found in
2584\&\fIevent.h\fR that are not directly supported by the libev core alone.
2585.IP "\s-1EV_USE_MONOTONIC\s0" 4
2586.IX Item "EV_USE_MONOTONIC"
2587If defined to be \f(CW1\fR, libev will try to detect the availability of the
2588monotonic clock option at both compiletime and runtime. Otherwise no use
2589of the monotonic clock option will be attempted. If you enable this, you
2590usually have to link against librt or something similar. Enabling it when
2591the functionality isn't available is safe, though, although you have
2592to make sure you link against any libraries where the \f(CW\*(C`clock_gettime\*(C'\fR
2593function is hiding in (often \fI\-lrt\fR).
2594.IP "\s-1EV_USE_REALTIME\s0" 4
2595.IX Item "EV_USE_REALTIME"
2596If defined to be \f(CW1\fR, libev will try to detect the availability of the
2597realtime clock option at compiletime (and assume its availability at
2598runtime if successful). Otherwise no use of the realtime clock option will
2599be attempted. This effectively replaces \f(CW\*(C`gettimeofday\*(C'\fR by \f(CW\*(C`clock_get
2600(CLOCK_REALTIME, ...)\*(C'\fR and will not normally affect correctness. See the
2601note about libraries in the description of \f(CW\*(C`EV_USE_MONOTONIC\*(C'\fR, though.
2602.IP "\s-1EV_USE_NANOSLEEP\s0" 4
2603.IX Item "EV_USE_NANOSLEEP"
2604If defined to be \f(CW1\fR, libev will assume that \f(CW\*(C`nanosleep ()\*(C'\fR is available
2605and will use it for delays. Otherwise it will use \f(CW\*(C`select ()\*(C'\fR.
2606.IP "\s-1EV_USE_SELECT\s0" 4
2607.IX Item "EV_USE_SELECT"
2608If undefined or defined to be \f(CW1\fR, libev will compile in support for the
2609\&\f(CW\*(C`select\*(C'\fR(2) backend. No attempt at autodetection will be done: if no
2610other method takes over, select will be it. Otherwise the select backend
2611will not be compiled in.
2612.IP "\s-1EV_SELECT_USE_FD_SET\s0" 4
2613.IX Item "EV_SELECT_USE_FD_SET"
2614If defined to \f(CW1\fR, then the select backend will use the system \f(CW\*(C`fd_set\*(C'\fR
2615structure. This is useful if libev doesn't compile due to a missing
2616\&\f(CW\*(C`NFDBITS\*(C'\fR or \f(CW\*(C`fd_mask\*(C'\fR definition or it misguesses the bitset layout on
2617exotic systems. This usually limits the range of file descriptors to some
2618low limit such as 1024 or might have other limitations (winsocket only
2619allows 64 sockets). The \f(CW\*(C`FD_SETSIZE\*(C'\fR macro, set before compilation, might
2620influence the size of the \f(CW\*(C`fd_set\*(C'\fR used.
2621.IP "\s-1EV_SELECT_IS_WINSOCKET\s0" 4
2622.IX Item "EV_SELECT_IS_WINSOCKET"
2623When defined to \f(CW1\fR, the select backend will assume that
2624select/socket/connect etc. don't understand file descriptors but
2625wants osf handles on win32 (this is the case when the select to
2626be used is the winsock select). This means that it will call
2627\&\f(CW\*(C`_get_osfhandle\*(C'\fR on the fd to convert it to an \s-1OS\s0 handle. Otherwise,
2628it is assumed that all these functions actually work on fds, even
2629on win32. Should not be defined on non\-win32 platforms.
2630.IP "\s-1EV_FD_TO_WIN32_HANDLE\s0" 4
2631.IX Item "EV_FD_TO_WIN32_HANDLE"
2632If \f(CW\*(C`EV_SELECT_IS_WINSOCKET\*(C'\fR is enabled, then libev needs a way to map
2633file descriptors to socket handles. When not defining this symbol (the
2634default), then libev will call \f(CW\*(C`_get_osfhandle\*(C'\fR, which is usually
2635correct. In some cases, programs use their own file descriptor management,
2636in which case they can provide this function to map fds to socket handles.
2637.IP "\s-1EV_USE_POLL\s0" 4
2638.IX Item "EV_USE_POLL"
2639If defined to be \f(CW1\fR, libev will compile in support for the \f(CW\*(C`poll\*(C'\fR(2)
2640backend. Otherwise it will be enabled on non\-win32 platforms. It
2641takes precedence over select.
2642.IP "\s-1EV_USE_EPOLL\s0" 4
2643.IX Item "EV_USE_EPOLL"
2644If defined to be \f(CW1\fR, libev will compile in support for the Linux
2645\&\f(CW\*(C`epoll\*(C'\fR(7) backend. Its availability will be detected at runtime,
2646otherwise another method will be used as fallback. This is the
2647preferred backend for GNU/Linux systems.
2648.IP "\s-1EV_USE_KQUEUE\s0" 4
2649.IX Item "EV_USE_KQUEUE"
2650If defined to be \f(CW1\fR, libev will compile in support for the \s-1BSD\s0 style
2651\&\f(CW\*(C`kqueue\*(C'\fR(2) backend. Its actual availability will be detected at runtime,
2652otherwise another method will be used as fallback. This is the preferred
2653backend for \s-1BSD\s0 and BSD-like systems, although on most BSDs kqueue only
2654supports some types of fds correctly (the only platform we found that
2655supports ptys for example was NetBSD), so kqueue might be compiled in, but
2656not be used unless explicitly requested. The best way to use it is to find
2657out whether kqueue supports your type of fd properly and use an embedded
2658kqueue loop.
2659.IP "\s-1EV_USE_PORT\s0" 4
2660.IX Item "EV_USE_PORT"
2661If defined to be \f(CW1\fR, libev will compile in support for the Solaris
266210 port style backend. Its availability will be detected at runtime,
2663otherwise another method will be used as fallback. This is the preferred
2664backend for Solaris 10 systems.
2665.IP "\s-1EV_USE_DEVPOLL\s0" 4
2666.IX Item "EV_USE_DEVPOLL"
2667reserved for future expansion, works like the \s-1USE\s0 symbols above.
2668.IP "\s-1EV_USE_INOTIFY\s0" 4
2669.IX Item "EV_USE_INOTIFY"
2670If defined to be \f(CW1\fR, libev will compile in support for the Linux inotify
2671interface to speed up \f(CW\*(C`ev_stat\*(C'\fR watchers. Its actual availability will
2672be detected at runtime.
2673.IP "\s-1EV_H\s0" 4
2674.IX Item "EV_H"
2675The name of the \fIev.h\fR header file used to include it. The default if
2676undefined is \f(CW"ev.h"\fR in \fIevent.h\fR, \fIev.c\fR and \fIev++.h\fR. This can be
2677used to virtually rename the \fIev.h\fR header file in case of conflicts.
2678.IP "\s-1EV_CONFIG_H\s0" 4
2679.IX Item "EV_CONFIG_H"
2680If \f(CW\*(C`EV_STANDALONE\*(C'\fR isn't \f(CW1\fR, this variable can be used to override
2681\&\fIev.c\fR's idea of where to find the \fIconfig.h\fR file, similarly to
2682\&\f(CW\*(C`EV_H\*(C'\fR, above.
2683.IP "\s-1EV_EVENT_H\s0" 4
2684.IX Item "EV_EVENT_H"
2685Similarly to \f(CW\*(C`EV_H\*(C'\fR, this macro can be used to override \fIevent.c\fR's idea
2686of how the \fIevent.h\fR header can be found, the default is \f(CW"event.h"\fR.
2687.IP "\s-1EV_PROTOTYPES\s0" 4
2688.IX Item "EV_PROTOTYPES"
2689If defined to be \f(CW0\fR, then \fIev.h\fR will not define any function
2690prototypes, but still define all the structs and other symbols. This is
2691occasionally useful if you want to provide your own wrapper functions
2692around libev functions.
2693.IP "\s-1EV_MULTIPLICITY\s0" 4
2694.IX Item "EV_MULTIPLICITY"
2695If undefined or defined to \f(CW1\fR, then all event-loop-specific functions
2696will have the \f(CW\*(C`struct ev_loop *\*(C'\fR as first argument, and you can create
2697additional independent event loops. Otherwise there will be no support
2698for multiple event loops and there is no first event loop pointer
2699argument. Instead, all functions act on the single default loop.
2700.IP "\s-1EV_MINPRI\s0" 4
2701.IX Item "EV_MINPRI"
2702.PD 0
2703.IP "\s-1EV_MAXPRI\s0" 4
2704.IX Item "EV_MAXPRI"
2705.PD
2706The range of allowed priorities. \f(CW\*(C`EV_MINPRI\*(C'\fR must be smaller or equal to
2707\&\f(CW\*(C`EV_MAXPRI\*(C'\fR, but otherwise there are no non-obvious limitations. You can
2708provide for more priorities by overriding those symbols (usually defined
2709to be \f(CW\*(C`\-2\*(C'\fR and \f(CW2\fR, respectively).
2710.Sp
2711When doing priority-based operations, libev usually has to linearly search
2712all the priorities, so having many of them (hundreds) uses a lot of space
2713and time, so using the defaults of five priorities (\-2 .. +2) is usually
2714fine.
2715.Sp
2716If your embedding app does not need any priorities, defining these both to
2717\&\f(CW0\fR will save some memory and cpu.
2718.IP "\s-1EV_PERIODIC_ENABLE\s0" 4
2719.IX Item "EV_PERIODIC_ENABLE"
2720If undefined or defined to be \f(CW1\fR, then periodic timers are supported. If
2721defined to be \f(CW0\fR, then they are not. Disabling them saves a few kB of
2722code.
2723.IP "\s-1EV_IDLE_ENABLE\s0" 4
2724.IX Item "EV_IDLE_ENABLE"
2725If undefined or defined to be \f(CW1\fR, then idle watchers are supported. If
2726defined to be \f(CW0\fR, then they are not. Disabling them saves a few kB of
2727code.
2728.IP "\s-1EV_EMBED_ENABLE\s0" 4
2729.IX Item "EV_EMBED_ENABLE"
2730If undefined or defined to be \f(CW1\fR, then embed watchers are supported. If
2731defined to be \f(CW0\fR, then they are not.
2732.IP "\s-1EV_STAT_ENABLE\s0" 4
2733.IX Item "EV_STAT_ENABLE"
2734If undefined or defined to be \f(CW1\fR, then stat watchers are supported. If
2735defined to be \f(CW0\fR, then they are not.
2736.IP "\s-1EV_FORK_ENABLE\s0" 4
2737.IX Item "EV_FORK_ENABLE"
2738If undefined or defined to be \f(CW1\fR, then fork watchers are supported. If
2739defined to be \f(CW0\fR, then they are not.
2740.IP "\s-1EV_MINIMAL\s0" 4
2741.IX Item "EV_MINIMAL"
2742If you need to shave off some kilobytes of code at the expense of some
2743speed, define this symbol to \f(CW1\fR. Currently only used for gcc to override
2744some inlining decisions, saves roughly 30% codesize of amd64.
2745.IP "\s-1EV_PID_HASHSIZE\s0" 4
2746.IX Item "EV_PID_HASHSIZE"
2747\&\f(CW\*(C`ev_child\*(C'\fR watchers use a small hash table to distribute workload by
2748pid. The default size is \f(CW16\fR (or \f(CW1\fR with \f(CW\*(C`EV_MINIMAL\*(C'\fR), usually more
2749than enough. If you need to manage thousands of children you might want to
2750increase this value (\fImust\fR be a power of two).
2751.IP "\s-1EV_INOTIFY_HASHSIZE\s0" 4
2752.IX Item "EV_INOTIFY_HASHSIZE"
2753\&\f(CW\*(C`ev_stat\*(C'\fR watchers use a small hash table to distribute workload by
2754inotify watch id. The default size is \f(CW16\fR (or \f(CW1\fR with \f(CW\*(C`EV_MINIMAL\*(C'\fR),
2755usually more than enough. If you need to manage thousands of \f(CW\*(C`ev_stat\*(C'\fR
2756watchers you might want to increase this value (\fImust\fR be a power of
2757two).
2758.IP "\s-1EV_COMMON\s0" 4
2759.IX Item "EV_COMMON"
2760By default, all watchers have a \f(CW\*(C`void *data\*(C'\fR member. By redefining
2761this macro to a something else you can include more and other types of
2762members. You have to define it each time you include one of the files,
2763though, and it must be identical each time.
2764.Sp
2765For example, the perl \s-1EV\s0 module uses something like this:
2766.Sp
2767.Vb 3
2768\& #define EV_COMMON \e
2769\& SV *self; /* contains this struct */ \e
2770\& SV *cb_sv, *fh /* note no trailing ";" */
2771.Ve
2772.IP "\s-1EV_CB_DECLARE\s0 (type)" 4
2773.IX Item "EV_CB_DECLARE (type)"
2774.PD 0
2775.IP "\s-1EV_CB_INVOKE\s0 (watcher, revents)" 4
2776.IX Item "EV_CB_INVOKE (watcher, revents)"
2777.IP "ev_set_cb (ev, cb)" 4
2778.IX Item "ev_set_cb (ev, cb)"
2779.PD
2780Can be used to change the callback member declaration in each watcher,
2781and the way callbacks are invoked and set. Must expand to a struct member
2782definition and a statement, respectively. See the \fIev.h\fR header file for
2783their default definitions. One possible use for overriding these is to
2784avoid the \f(CW\*(C`struct ev_loop *\*(C'\fR as first argument in all cases, or to use
2785method calls instead of plain function calls in \*(C+.
2786.Sh "\s-1EXPORTED\s0 \s-1API\s0 \s-1SYMBOLS\s0"
2787.IX Subsection "EXPORTED API SYMBOLS"
2788If you need to re-export the \s-1API\s0 (e.g. via a dll) and you need a list of
2789exported symbols, you can use the provided \fISymbol.*\fR files which list
2790all public symbols, one per line:
2791.PP
2792.Vb 2
2793\& Symbols.ev for libev proper
2794\& Symbols.event for the libevent emulation
2795.Ve
2796.PP
2797This can also be used to rename all public symbols to avoid clashes with
2798multiple versions of libev linked together (which is obviously bad in
2799itself, but sometimes it is inconvinient to avoid this).
2800.PP
2801A sed command like this will create wrapper \f(CW\*(C`#define\*(C'\fR's that you need to
2802include before including \fIev.h\fR:
2803.PP
2804.Vb 1
2805\& <Symbols.ev sed \-e "s/.*/#define & myprefix_&/" >wrap.h
2806.Ve
2807.PP
2808This would create a file \fIwrap.h\fR which essentially looks like this:
2809.PP
2810.Vb 4
2811\& #define ev_backend myprefix_ev_backend
2812\& #define ev_check_start myprefix_ev_check_start
2813\& #define ev_check_stop myprefix_ev_check_stop
2814\& ...
2815.Ve
2816.Sh "\s-1EXAMPLES\s0"
2817.IX Subsection "EXAMPLES"
2818For a real-world example of a program the includes libev
2819verbatim, you can have a look at the \s-1EV\s0 perl module
2820(<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
2821the \fIlibev/\fR subdirectory and includes them in the \fI\s-1EV/EVAPI\s0.h\fR (public
2822interface) and \fI\s-1EV\s0.xs\fR (implementation) files. Only the \fI\s-1EV\s0.xs\fR file
2823will be compiled. It is pretty complex because it provides its own header
2824file.
2825.PP
2826The usage in rxvt-unicode is simpler. It has a \fIev_cpp.h\fR header file
2827that everybody includes and which overrides some configure choices:
2828.PP
2829.Vb 9
2830\& #define EV_MINIMAL 1
2831\& #define EV_USE_POLL 0
2832\& #define EV_MULTIPLICITY 0
2833\& #define EV_PERIODIC_ENABLE 0
2834\& #define EV_STAT_ENABLE 0
2835\& #define EV_FORK_ENABLE 0
2836\& #define EV_CONFIG_H <config.h>
2837\& #define EV_MINPRI 0
2838\& #define EV_MAXPRI 0
2839\&
2840\& #include "ev++.h"
2841.Ve
2842.PP
2843And a \fIev_cpp.C\fR implementation file that contains libev proper and is compiled:
2844.PP
2845.Vb 2
2846\& #include "ev_cpp.h"
2847\& #include "ev.c"
2848.Ve
2849.SH "COMPLEXITIES"
2850.IX Header "COMPLEXITIES"
2851In this section the complexities of (many of) the algorithms used inside
2852libev will be explained. For complexity discussions about backends see the
2853documentation for \f(CW\*(C`ev_default_init\*(C'\fR.
2854.PP
2855All of the following are about amortised time: If an array needs to be
2856extended, libev needs to realloc and move the whole array, but this
2857happens asymptotically never with higher number of elements, so O(1) might
2858mean it might do a lengthy realloc operation in rare cases, but on average
2859it is much faster and asymptotically approaches constant time.
2860.IP "Starting and stopping timer/periodic watchers: O(log skipped_other_timers)" 4
2861.IX Item "Starting and stopping timer/periodic watchers: O(log skipped_other_timers)"
2862This means that, when you have a watcher that triggers in one hour and
2863there are 100 watchers that would trigger before that then inserting will
2864have to skip roughly seven (\f(CW\*(C`ld 100\*(C'\fR) of these watchers.
2865.IP "Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)" 4
2866.IX Item "Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)"
2867That means that changing a timer costs less than removing/adding them
2868as only the relative motion in the event queue has to be paid for.
2869.IP "Starting io/check/prepare/idle/signal/child watchers: O(1)" 4
2870.IX Item "Starting io/check/prepare/idle/signal/child watchers: O(1)"
2871These just add the watcher into an array or at the head of a list.
2872.IP "Stopping check/prepare/idle watchers: O(1)" 4
2873.IX Item "Stopping check/prepare/idle watchers: O(1)"
2874.PD 0
2875.IP "Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % \s-1EV_PID_HASHSIZE\s0))" 4
2876.IX Item "Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))"
2877.PD
2878These watchers are stored in lists then need to be walked to find the
2879correct watcher to remove. The lists are usually short (you don't usually
2880have many watchers waiting for the same fd or signal).
2881.IP "Finding the next timer in each loop iteration: O(1)" 4
2882.IX Item "Finding the next timer in each loop iteration: O(1)"
2883By virtue of using a binary heap, the next timer is always found at the
2884beginning of the storage array.
2885.IP "Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)" 4
2886.IX Item "Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)"
2887A change means an I/O watcher gets started or stopped, which requires
2888libev to recalculate its status (and possibly tell the kernel, depending
2889on backend and wether \f(CW\*(C`ev_io_set\*(C'\fR was used).
2890.IP "Activating one watcher (putting it into the pending state): O(1)" 4
2891.IX Item "Activating one watcher (putting it into the pending state): O(1)"
2892.PD 0
2893.IP "Priority handling: O(number_of_priorities)" 4
2894.IX Item "Priority handling: O(number_of_priorities)"
2895.PD
2896Priorities are implemented by allocating some space for each
2897priority. When doing priority-based operations, libev usually has to
2898linearly search all the priorities, but starting/stopping and activating
2899watchers becomes O(1) w.r.t. prioritiy handling.
2900.SH "Win32 platform limitations and workarounds"
2901.IX Header "Win32 platform limitations and workarounds"
2902Win32 doesn't support any of the standards (e.g. \s-1POSIX\s0) that libev
2903requires, and its I/O model is fundamentally incompatible with the \s-1POSIX\s0
2904model. Libev still offers limited functionality on this platform in
2905the form of the \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR backend, and only supports socket
2906descriptors. This only applies when using Win32 natively, not when using
2907e.g. cygwin.
2908.PP
2909There is no supported compilation method available on windows except
2910embedding it into other applications.
2911.PP
2912Due to the many, low, and arbitrary limits on the win32 platform and the
2913abysmal performance of winsockets, using a large number of sockets is not
2914recommended (and not reasonable). If your program needs to use more than
2915a hundred or so sockets, then likely it needs to use a totally different
2916implementation for windows, as libev offers the \s-1POSIX\s0 model, which cannot
2917be implemented efficiently on windows (microsoft monopoly games).
2918.IP "The winsocket select function" 4
2919.IX Item "The winsocket select function"
2920The winsocket \f(CW\*(C`select\*(C'\fR function doesn't follow \s-1POSIX\s0 in that it requires
2921socket \fIhandles\fR and not socket \fIfile descriptors\fR. This makes select
2922very inefficient, and also requires a mapping from file descriptors
2923to socket handles. See the discussion of the \f(CW\*(C`EV_SELECT_USE_FD_SET\*(C'\fR,
2924\&\f(CW\*(C`EV_SELECT_IS_WINSOCKET\*(C'\fR and \f(CW\*(C`EV_FD_TO_WIN32_HANDLE\*(C'\fR preprocessor
2925symbols for more info.
2926.Sp
2927The configuration for a \*(L"naked\*(R" win32 using the microsoft runtime
2928libraries and raw winsocket select is:
2929.Sp
2930.Vb 2
2931\& #define EV_USE_SELECT 1
2932\& #define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
2933.Ve
2934.Sp
2935Note that winsockets handling of fd sets is O(n), so you can easily get a
2936complexity in the O(nA\*^X) range when using win32.
2937.IP "Limited number of file descriptors" 4
2938.IX Item "Limited number of file descriptors"
2939Windows has numerous arbitrary (and low) limits on things. Early versions
2940of winsocket's select only supported waiting for a max. of \f(CW64\fR handles
2941(probably owning to the fact that all windows kernels can only wait for
2942\&\f(CW64\fR things at the same time internally; microsoft recommends spawning a
2943chain of threads and wait for 63 handles and the previous thread in each).
2944.Sp
2945Newer versions support more handles, but you need to define \f(CW\*(C`FD_SETSIZE\*(C'\fR
2946to some high number (e.g. \f(CW2048\fR) before compiling the winsocket select
2947call (which might be in libev or elsewhere, for example, perl does its own
2948select emulation on windows).
2949.Sp
2950Another limit is the number of file descriptors in the microsoft runtime
2951libraries, which by default is \f(CW64\fR (there must be a hidden \fI64\fR fetish
2952or something like this inside microsoft). You can increase this by calling
2953\&\f(CW\*(C`_setmaxstdio\*(C'\fR, which can increase this limit to \f(CW2048\fR (another
2954arbitrary limit), but is broken in many versions of the microsoft runtime
2955libraries.
2956.Sp
2957This might get you to about \f(CW512\fR or \f(CW2048\fR sockets (depending on
2958windows version and/or the phase of the moon). To get more, you need to
2959wrap all I/O functions and provide your own fd management, but the cost of
2960calling select (O(nA\*^X)) will likely make this unworkable.
889.SH "AUTHOR" 2961.SH "AUTHOR"
890.IX Header "AUTHOR" 2962.IX Header "AUTHOR"
891Marc Lehmann <libev@schmorp.de>. 2963Marc Lehmann <libev@schmorp.de>.
2964.SH "POD ERRORS"
2965.IX Header "POD ERRORS"
2966Hey! \fBThe above document had some coding errors, which are explained below:\fR
2967.IP "Around line 2686:" 4
2968.IX Item "Around line 2686:"
2969You forgot a '=back' before '=head2'

Diff Legend

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