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

Comparing libev/ev.3 (file contents):
Revision 1.9 by root, Fri Nov 23 16:17:12 2007 UTC vs.
Revision 1.63 by root, Wed Apr 2 15:23:11 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 "2007-11-23" "perl v5.8.8" "User Contributed Perl Documentation" 135.TH EV 1 "2008-04-02" "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 2
150\& // a single header file is required
151\& #include <ev.h>
152\&
153\& // every watcher type has its own typedef\*(Aqd struct
154\& // with the name ev_<type>
155\& ev_io stdin_watcher;
156\& ev_timer timeout_watcher;
157\&
158\& // all watcher callbacks have a similar signature
159\& // this callback is called when data is readable on stdin
160\& static void
161\& stdin_cb (EV_P_ struct ev_io *w, int revents)
162\& {
163\& puts ("stdin ready");
164\& // for one\-shot events, one must manually stop the watcher
165\& // with its corresponding stop function.
166\& ev_io_stop (EV_A_ w);
167\&
168\& // this causes all nested ev_loop\*(Aqs to stop iterating
169\& ev_unloop (EV_A_ EVUNLOOP_ALL);
170\& }
171\&
172\& // another callback, this time for a time\-out
173\& static void
174\& timeout_cb (EV_P_ struct ev_timer *w, int revents)
175\& {
176\& puts ("timeout");
177\& // this causes the innermost ev_loop to stop iterating
178\& ev_unloop (EV_A_ EVUNLOOP_ONE);
179\& }
180\&
181\& int
182\& main (void)
183\& {
184\& // use the default event loop unless you have special needs
185\& struct ev_loop *loop = ev_default_loop (0);
186\&
187\& // initialise an io watcher, then start it
188\& // this one will watch for stdin to become readable
189\& ev_io_init (&stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
190\& ev_io_start (loop, &stdin_watcher);
191\&
192\& // initialise a timer watcher, then start it
193\& // simple non\-repeating 5.5 second timeout
194\& ev_timer_init (&timeout_watcher, timeout_cb, 5.5, 0.);
195\& ev_timer_start (loop, &timeout_watcher);
196\&
197\& // now wait for events to arrive
198\& ev_loop (loop, 0);
199\&
200\& // unloop was called, so exit
201\& return 0;
202\& }
203.Ve
140.SH "DESCRIPTION" 204.SH "DESCRIPTION"
141.IX Header "DESCRIPTION" 205.IX Header "DESCRIPTION"
206The newest version of this document is also available as an html-formatted
207web page you might find easier to navigate when reading it for the first
208time: <http://cvs.schmorp.de/libev/ev.html>.
209.PP
142Libev is an event loop: you register interest in certain events (such as a 210Libev 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 211file descriptor being readable or a timeout occurring), and it will manage
144these event sources and provide your program with events. 212these event sources and provide your program with events.
145.PP 213.PP
146To do this, it must take more or less complete control over your process 214To 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 215(or thread) by executing the \fIevent loop\fR handler, and will then
148communicate events via a callback mechanism. 216communicate events via a callback mechanism.
149.PP 217.PP
150You register interest in certain events by registering so-called \fIevent 218You register interest in certain events by registering so-called \fIevent
151watchers\fR, which are relatively small C structures you initialise with the 219watchers\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 220details of the event, and then hand it over to libev by \fIstarting\fR the
153watcher. 221watcher.
154.SH "FEATURES" 222.Sh "\s-1FEATURES\s0"
155.IX Header "FEATURES" 223.IX Subsection "FEATURES"
156Libev supports select, poll, the linux-specific epoll and the bsd-specific 224Libev 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 225BSD-specific \f(CW\*(C`kqueue\*(C'\fR and the Solaris-specific event port mechanisms
158timers with customised rescheduling, signal events, process status change 226for 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 227(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 228with customised rescheduling (\f(CW\*(C`ev_periodic\*(C'\fR), synchronous signals
161fast (see this benchmark comparing 229(\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). 230watchers dealing with the event loop mechanism itself (\f(CW\*(C`ev_idle\*(C'\fR,
231\&\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
232file watchers (\f(CW\*(C`ev_stat\*(C'\fR) and even limited support for fork events
233(\f(CW\*(C`ev_fork\*(C'\fR).
234.PP
235It also is quite fast (see this
236benchmark comparing it to libevent
237for example).
163.SH "CONVENTIONS" 238.Sh "\s-1CONVENTIONS\s0"
164.IX Header "CONVENTIONS" 239.IX Subsection "CONVENTIONS"
165Libev is very configurable. In this manual the default configuration 240Libev is very configurable. In this manual the default (and most common)
166will be described, which supports multiple event loops. For more info 241configuration will be described, which supports multiple event loops. For
167about various configuration options please have a look at the file 242more info about various configuration options please have a look at
168\&\fI\s-1README\s0.embed\fR in the libev distribution. If libev was configured without 243\&\fB\s-1EMBED\s0\fR section in this manual. If libev was configured without support
169support for multiple event loops, then all functions taking an initial 244for multiple event loops, then all functions taking an initial argument of
170argument of name \f(CW\*(C`loop\*(C'\fR (which is always of type \f(CW\*(C`struct ev_loop *\*(C'\fR) 245name \f(CW\*(C`loop\*(C'\fR (which is always of type \f(CW\*(C`struct ev_loop *\*(C'\fR) will not have
171will not have this argument. 246this argument.
172.SH "TIME REPRESENTATION" 247.Sh "\s-1TIME\s0 \s-1REPRESENTATION\s0"
173.IX Header "TIME REPRESENTATION" 248.IX Subsection "TIME REPRESENTATION"
174Libev represents time as a single floating point number, representing the 249Libev represents time as a single floating point number, representing the
175(fractional) number of seconds since the (\s-1POSIX\s0) epoch (somewhere near 250(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 251the 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 252called \f(CW\*(C`ev_tstamp\*(C'\fR, which is what you should use too. It usually aliases
178to the \f(CW\*(C`double\*(C'\fR type in C, and when you need to do any calculations on 253to the \f(CW\*(C`double\*(C'\fR type in C, and when you need to do any calculations on
179it, you should treat it as such. 254it, you should treat it as some floatingpoint value. Unlike the name
255component \f(CW\*(C`stamp\*(C'\fR might indicate, it is also used for time differences
256throughout libev.
180.SH "GLOBAL FUNCTIONS" 257.SH "GLOBAL FUNCTIONS"
181.IX Header "GLOBAL FUNCTIONS" 258.IX Header "GLOBAL FUNCTIONS"
182These functions can be called anytime, even before initialising the 259These functions can be called anytime, even before initialising the
183library in any way. 260library in any way.
184.IP "ev_tstamp ev_time ()" 4 261.IP "ev_tstamp ev_time ()" 4
185.IX Item "ev_tstamp ev_time ()" 262.IX Item "ev_tstamp ev_time ()"
186Returns the current time as libev would use it. Please note that the 263Returns the current time as libev would use it. Please note that the
187\&\f(CW\*(C`ev_now\*(C'\fR function is usually faster and also often returns the timestamp 264\&\f(CW\*(C`ev_now\*(C'\fR function is usually faster and also often returns the timestamp
188you actually want to know. 265you actually want to know.
266.IP "ev_sleep (ev_tstamp interval)" 4
267.IX Item "ev_sleep (ev_tstamp interval)"
268Sleep for the given interval: The current thread will be blocked until
269either it is interrupted or the given time interval has passed. Basically
270this is a subsecond-resolution \f(CW\*(C`sleep ()\*(C'\fR.
189.IP "int ev_version_major ()" 4 271.IP "int ev_version_major ()" 4
190.IX Item "int ev_version_major ()" 272.IX Item "int ev_version_major ()"
191.PD 0 273.PD 0
192.IP "int ev_version_minor ()" 4 274.IP "int ev_version_minor ()" 4
193.IX Item "int ev_version_minor ()" 275.IX Item "int ev_version_minor ()"
194.PD 276.PD
195You can find out the major and minor version numbers of the library 277You can find out the major and minor \s-1ABI\s0 version numbers of the library
196you linked against by calling the functions \f(CW\*(C`ev_version_major\*(C'\fR and 278you linked against by calling the functions \f(CW\*(C`ev_version_major\*(C'\fR and
197\&\f(CW\*(C`ev_version_minor\*(C'\fR. If you want, you can compare against the global 279\&\f(CW\*(C`ev_version_minor\*(C'\fR. If you want, you can compare against the global
198symbols \f(CW\*(C`EV_VERSION_MAJOR\*(C'\fR and \f(CW\*(C`EV_VERSION_MINOR\*(C'\fR, which specify the 280symbols \f(CW\*(C`EV_VERSION_MAJOR\*(C'\fR and \f(CW\*(C`EV_VERSION_MINOR\*(C'\fR, which specify the
199version of the library your program was compiled against. 281version of the library your program was compiled against.
200.Sp 282.Sp
283These version numbers refer to the \s-1ABI\s0 version of the library, not the
284release version.
285.Sp
201Usually, it's a good idea to terminate if the major versions mismatch, 286Usually, it's a good idea to terminate if the major versions mismatch,
202as this indicates an incompatible change. Minor versions are usually 287as this indicates an incompatible change. Minor versions are usually
203compatible to older versions, so a larger minor version alone is usually 288compatible to older versions, so a larger minor version alone is usually
204not a problem. 289not a problem.
205.Sp 290.Sp
206Example: make sure we haven't accidentally been linked against the wrong 291Example: Make sure we haven't accidentally been linked against the wrong
207version: 292version.
208.Sp 293.Sp
209.Vb 3 294.Vb 3
210\& assert (("libev version mismatch", 295\& assert (("libev version mismatch",
211\& ev_version_major () == EV_VERSION_MAJOR 296\& ev_version_major () == EV_VERSION_MAJOR
212\& && ev_version_minor () >= EV_VERSION_MINOR)); 297\& && ev_version_minor () >= EV_VERSION_MINOR));
… …
231recommended for this platform. This set is often smaller than the one 316recommended for this platform. This set is often smaller than the one
232returned by \f(CW\*(C`ev_supported_backends\*(C'\fR, as for example kqueue is broken on 317returned by \f(CW\*(C`ev_supported_backends\*(C'\fR, as for example kqueue is broken on
233most BSDs and will not be autodetected unless you explicitly request it 318most BSDs and will not be autodetected unless you explicitly request it
234(assuming you know what you are doing). This is the set of backends that 319(assuming you know what you are doing). This is the set of backends that
235libev will probe for if you specify no backends explicitly. 320libev will probe for if you specify no backends explicitly.
321.IP "unsigned int ev_embeddable_backends ()" 4
322.IX Item "unsigned int ev_embeddable_backends ()"
323Returns the set of backends that are embeddable in other event loops. This
324is the theoretical, all-platform, value. To find which backends
325might be supported on the current system, you would need to look at
326\&\f(CW\*(C`ev_embeddable_backends () & ev_supported_backends ()\*(C'\fR, likewise for
327recommended ones.
328.Sp
329See the description of \f(CW\*(C`ev_embed\*(C'\fR watchers for more info.
236.IP "ev_set_allocator (void *(*cb)(void *ptr, long size))" 4 330.IP "ev_set_allocator (void *(*cb)(void *ptr, long size))" 4
237.IX Item "ev_set_allocator (void *(*cb)(void *ptr, long size))" 331.IX Item "ev_set_allocator (void *(*cb)(void *ptr, long size))"
238Sets the allocation function to use (the prototype is similar to the 332Sets the allocation function to use (the prototype is similar \- the
239realloc C function, the semantics are identical). It is used to allocate 333semantics is identical \- to the realloc C function). It is used to
240and free memory (no surprises here). If it returns zero when memory 334allocate and free memory (no surprises here). If it returns zero when
241needs to be allocated, the library might abort or take some potentially 335memory needs to be allocated, the library might abort or take some
242destructive action. The default is your system realloc function. 336potentially destructive action. The default is your system realloc
337function.
243.Sp 338.Sp
244You could override this function in high-availability programs to, say, 339You could override this function in high-availability programs to, say,
245free some memory if it cannot allocate memory, to use a special allocator, 340free some memory if it cannot allocate memory, to use a special allocator,
246or even to sleep a while and retry until some memory is available. 341or even to sleep a while and retry until some memory is available.
247.Sp 342.Sp
248Example: replace the libev allocator with one that waits a bit and then 343Example: Replace the libev allocator with one that waits a bit and then
249retries: better than mine). 344retries).
250.Sp 345.Sp
251.Vb 6 346.Vb 6
252\& static void * 347\& static void *
253\& persistent_realloc (void *ptr, long size) 348\& persistent_realloc (void *ptr, size_t size)
254\& { 349\& {
255\& for (;;) 350\& for (;;)
256\& { 351\& {
257\& void *newptr = realloc (ptr, size); 352\& void *newptr = realloc (ptr, size);
258.Ve 353\&
259.Sp
260.Vb 2
261\& if (newptr) 354\& if (newptr)
262\& return newptr; 355\& return newptr;
263.Ve 356\&
264.Sp
265.Vb 3
266\& sleep (60); 357\& sleep (60);
267\& } 358\& }
268\& } 359\& }
269.Ve 360\&
270.Sp
271.Vb 2
272\& ... 361\& ...
273\& ev_set_allocator (persistent_realloc); 362\& ev_set_allocator (persistent_realloc);
274.Ve 363.Ve
275.IP "ev_set_syserr_cb (void (*cb)(const char *msg));" 4 364.IP "ev_set_syserr_cb (void (*cb)(const char *msg));" 4
276.IX Item "ev_set_syserr_cb (void (*cb)(const char *msg));" 365.IX Item "ev_set_syserr_cb (void (*cb)(const char *msg));"
… …
280callback is set, then libev will expect it to remedy the sitution, no 369callback is set, then libev will expect it to remedy the sitution, no
281matter what, when it returns. That is, libev will generally retry the 370matter what, when it returns. That is, libev will generally retry the
282requested operation, or, if the condition doesn't go away, do bad stuff 371requested operation, or, if the condition doesn't go away, do bad stuff
283(such as abort). 372(such as abort).
284.Sp 373.Sp
285Example: do the same thing as libev does internally: 374Example: This is basically the same thing that libev does internally, too.
286.Sp 375.Sp
287.Vb 6 376.Vb 6
288\& static void 377\& static void
289\& fatal_error (const char *msg) 378\& fatal_error (const char *msg)
290\& { 379\& {
291\& perror (msg); 380\& perror (msg);
292\& abort (); 381\& abort ();
293\& } 382\& }
294.Ve 383\&
295.Sp
296.Vb 2
297\& ... 384\& ...
298\& ev_set_syserr_cb (fatal_error); 385\& ev_set_syserr_cb (fatal_error);
299.Ve 386.Ve
300.SH "FUNCTIONS CONTROLLING THE EVENT LOOP" 387.SH "FUNCTIONS CONTROLLING THE EVENT LOOP"
301.IX Header "FUNCTIONS CONTROLLING THE EVENT LOOP" 388.IX Header "FUNCTIONS CONTROLLING THE EVENT LOOP"
… …
316false. If it already was initialised it simply returns it (and ignores the 403false. If it already was initialised it simply returns it (and ignores the
317flags. If that is troubling you, check \f(CW\*(C`ev_backend ()\*(C'\fR afterwards). 404flags. If that is troubling you, check \f(CW\*(C`ev_backend ()\*(C'\fR afterwards).
318.Sp 405.Sp
319If you don't know what event loop to use, use the one returned from this 406If you don't know what event loop to use, use the one returned from this
320function. 407function.
408.Sp
409Note that this function is \fInot\fR thread-safe, so if you want to use it
410from multiple threads, you have to lock (note also that this is unlikely,
411as loops cannot bes hared easily between threads anyway).
412.Sp
413The default loop is the only loop that can handle \f(CW\*(C`ev_signal\*(C'\fR and
414\&\f(CW\*(C`ev_child\*(C'\fR watchers, and to do this, it always registers a handler
415for \f(CW\*(C`SIGCHLD\*(C'\fR. If this is a problem for your app you can either
416create a dynamic loop with \f(CW\*(C`ev_loop_new\*(C'\fR that doesn't do that, or you
417can simply overwrite the \f(CW\*(C`SIGCHLD\*(C'\fR signal handler \fIafter\fR calling
418\&\f(CW\*(C`ev_default_init\*(C'\fR.
321.Sp 419.Sp
322The flags argument can be used to specify special behaviour or specific 420The flags argument can be used to specify special behaviour or specific
323backends to use, and is usually specified as \f(CW0\fR (or \f(CW\*(C`EVFLAG_AUTO\*(C'\fR). 421backends to use, and is usually specified as \f(CW0\fR (or \f(CW\*(C`EVFLAG_AUTO\*(C'\fR).
324.Sp 422.Sp
325The following flags are supported: 423The following flags are supported:
… …
336or setgid) then libev will \fInot\fR look at the environment variable 434or setgid) then libev will \fInot\fR look at the environment variable
337\&\f(CW\*(C`LIBEV_FLAGS\*(C'\fR. Otherwise (the default), this environment variable will 435\&\f(CW\*(C`LIBEV_FLAGS\*(C'\fR. Otherwise (the default), this environment variable will
338override the flags completely if it is found in the environment. This is 436override the flags completely if it is found in the environment. This is
339useful to try out specific backends to test their performance, or to work 437useful to try out specific backends to test their performance, or to work
340around bugs. 438around bugs.
439.ie n .IP """EVFLAG_FORKCHECK""" 4
440.el .IP "\f(CWEVFLAG_FORKCHECK\fR" 4
441.IX Item "EVFLAG_FORKCHECK"
442Instead of calling \f(CW\*(C`ev_default_fork\*(C'\fR or \f(CW\*(C`ev_loop_fork\*(C'\fR manually after
443a fork, you can also make libev check for a fork in each iteration by
444enabling this flag.
445.Sp
446This works by calling \f(CW\*(C`getpid ()\*(C'\fR on every iteration of the loop,
447and thus this might slow down your event loop if you do a lot of loop
448iterations and little real work, but is usually not noticeable (on my
449GNU/Linux system for example, \f(CW\*(C`getpid\*(C'\fR is actually a simple 5\-insn sequence
450without a syscall and thus \fIvery\fR fast, but my GNU/Linux system also has
451\&\f(CW\*(C`pthread_atfork\*(C'\fR which is even faster).
452.Sp
453The big advantage of this flag is that you can forget about fork (and
454forget about forgetting to tell libev about forking) when you use this
455flag.
456.Sp
457This flag setting cannot be overriden or specified in the \f(CW\*(C`LIBEV_FLAGS\*(C'\fR
458environment variable.
341.ie n .IP """EVBACKEND_SELECT"" (value 1, portable select backend)" 4 459.ie n .IP """EVBACKEND_SELECT"" (value 1, portable select backend)" 4
342.el .IP "\f(CWEVBACKEND_SELECT\fR (value 1, portable select backend)" 4 460.el .IP "\f(CWEVBACKEND_SELECT\fR (value 1, portable select backend)" 4
343.IX Item "EVBACKEND_SELECT (value 1, portable select backend)" 461.IX Item "EVBACKEND_SELECT (value 1, portable select backend)"
344This is your standard \fIselect\fR\|(2) backend. Not \fIcompletely\fR standard, as 462This is your standard \fIselect\fR\|(2) backend. Not \fIcompletely\fR standard, as
345libev tries to roll its own fd_set with no limits on the number of fds, 463libev tries to roll its own fd_set with no limits on the number of fds,
346but if that fails, expect a fairly low limit on the number of fds when 464but if that fails, expect a fairly low limit on the number of fds when
347using this backend. It doesn't scale too well (O(highest_fd)), but its usually 465using this backend. It doesn't scale too well (O(highest_fd)), but its
348the fastest backend for a low number of fds. 466usually the fastest backend for a low number of (low-numbered :) fds.
467.Sp
468To get good performance out of this backend you need a high amount of
469parallelity (most of the file descriptors should be busy). If you are
470writing a server, you should \f(CW\*(C`accept ()\*(C'\fR in a loop to accept as many
471connections as possible during one iteration. You might also want to have
472a look at \f(CW\*(C`ev_set_io_collect_interval ()\*(C'\fR to increase the amount of
473readyness notifications you get per iteration.
349.ie n .IP """EVBACKEND_POLL"" (value 2, poll backend, available everywhere except on windows)" 4 474.ie n .IP """EVBACKEND_POLL"" (value 2, poll backend, available everywhere except on windows)" 4
350.el .IP "\f(CWEVBACKEND_POLL\fR (value 2, poll backend, available everywhere except on windows)" 4 475.el .IP "\f(CWEVBACKEND_POLL\fR (value 2, poll backend, available everywhere except on windows)" 4
351.IX Item "EVBACKEND_POLL (value 2, poll backend, available everywhere except on windows)" 476.IX Item "EVBACKEND_POLL (value 2, poll backend, available everywhere except on windows)"
352And this is your standard \fIpoll\fR\|(2) backend. It's more complicated than 477And this is your standard \fIpoll\fR\|(2) backend. It's more complicated
353select, but handles sparse fds better and has no artificial limit on the 478than select, but handles sparse fds better and has no artificial
354number of fds you can use (except it will slow down considerably with a 479limit on the number of fds you can use (except it will slow down
355lot of inactive fds). It scales similarly to select, i.e. O(total_fds). 480considerably with a lot of inactive fds). It scales similarly to select,
481i.e. O(total_fds). See the entry for \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR, above, for
482performance tips.
356.ie n .IP """EVBACKEND_EPOLL"" (value 4, Linux)" 4 483.ie n .IP """EVBACKEND_EPOLL"" (value 4, Linux)" 4
357.el .IP "\f(CWEVBACKEND_EPOLL\fR (value 4, Linux)" 4 484.el .IP "\f(CWEVBACKEND_EPOLL\fR (value 4, Linux)" 4
358.IX Item "EVBACKEND_EPOLL (value 4, Linux)" 485.IX Item "EVBACKEND_EPOLL (value 4, Linux)"
359For few fds, this backend is a bit little slower than poll and select, 486For few fds, this backend is a bit little slower than poll and select,
360but it scales phenomenally better. While poll and select usually scale like 487but it scales phenomenally better. While poll and select usually scale
361O(total_fds) where n is the total number of fds (or the highest fd), epoll scales 488like O(total_fds) where n is the total number of fds (or the highest fd),
362either O(1) or O(active_fds). 489epoll scales either O(1) or O(active_fds). The epoll design has a number
490of shortcomings, such as silently dropping events in some hard-to-detect
491cases and rewiring a syscall per fd change, no fork support and bad
492support for dup.
363.Sp 493.Sp
364While stopping and starting an I/O watcher in the same iteration will 494While stopping, setting and starting an I/O watcher in the same iteration
365result in some caching, there is still a syscall per such incident 495will result in some caching, there is still a syscall per such incident
366(because the fd could point to a different file description now), so its 496(because the fd could point to a different file description now), so its
367best to avoid that. Also, \fIdup()\fRed file descriptors might not work very 497best to avoid that. Also, \f(CW\*(C`dup ()\*(C'\fR'ed file descriptors might not work
368well if you register events for both fds. 498very well if you register events for both fds.
369.Sp 499.Sp
370Please note that epoll sometimes generates spurious notifications, so you 500Please note that epoll sometimes generates spurious notifications, so you
371need to use non-blocking I/O or other means to avoid blocking when no data 501need to use non-blocking I/O or other means to avoid blocking when no data
372(or space) is available. 502(or space) is available.
503.Sp
504Best performance from this backend is achieved by not unregistering all
505watchers for a file descriptor until it has been closed, if possible, i.e.
506keep at least one watcher active per fd at all times.
507.Sp
508While nominally embeddeble in other event loops, this feature is broken in
509all kernel versions tested so far.
373.ie n .IP """EVBACKEND_KQUEUE"" (value 8, most \s-1BSD\s0 clones)" 4 510.ie n .IP """EVBACKEND_KQUEUE"" (value 8, most \s-1BSD\s0 clones)" 4
374.el .IP "\f(CWEVBACKEND_KQUEUE\fR (value 8, most \s-1BSD\s0 clones)" 4 511.el .IP "\f(CWEVBACKEND_KQUEUE\fR (value 8, most \s-1BSD\s0 clones)" 4
375.IX Item "EVBACKEND_KQUEUE (value 8, most BSD clones)" 512.IX Item "EVBACKEND_KQUEUE (value 8, most BSD clones)"
376Kqueue deserves special mention, as at the time of this writing, it 513Kqueue deserves special mention, as at the time of this writing, it
377was broken on all BSDs except NetBSD (usually it doesn't work with 514was broken on all BSDs except NetBSD (usually it doesn't work reliably
378anything but sockets and pipes, except on Darwin, where of course its 515with anything but sockets and pipes, except on Darwin, where of course
379completely useless). For this reason its not being \*(L"autodetected\*(R" 516it's completely useless). For this reason it's not being \*(L"autodetected\*(R"
380unless you explicitly specify it explicitly in the flags (i.e. using 517unless you explicitly specify it explicitly in the flags (i.e. using
381\&\f(CW\*(C`EVBACKEND_KQUEUE\*(C'\fR). 518\&\f(CW\*(C`EVBACKEND_KQUEUE\*(C'\fR) or libev was compiled on a known-to-be-good (\-enough)
519system like NetBSD.
520.Sp
521You still can embed kqueue into a normal poll or select backend and use it
522only for sockets (after having made sure that sockets work with kqueue on
523the target platform). See \f(CW\*(C`ev_embed\*(C'\fR watchers for more info.
382.Sp 524.Sp
383It scales in the same way as the epoll backend, but the interface to the 525It scales in the same way as the epoll backend, but the interface to the
384kernel is more efficient (which says nothing about its actual speed, of 526kernel is more efficient (which says nothing about its actual speed, of
385course). While starting and stopping an I/O watcher does not cause an 527course). While stopping, setting and starting an I/O watcher does never
386extra syscall as with epoll, it still adds up to four event changes per 528cause an extra syscall as with \f(CW\*(C`EVBACKEND_EPOLL\*(C'\fR, it still adds up to
387incident, so its best to avoid that. 529two event changes per incident, support for \f(CW\*(C`fork ()\*(C'\fR is very bad and it
530drops fds silently in similarly hard-to-detect cases.
531.Sp
532This backend usually performs well under most conditions.
533.Sp
534While nominally embeddable in other event loops, this doesn't work
535everywhere, so you might need to test for this. And since it is broken
536almost everywhere, you should only use it when you have a lot of sockets
537(for which it usually works), by embedding it into another event loop
538(e.g. \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or \f(CW\*(C`EVBACKEND_POLL\*(C'\fR) and using it only for
539sockets.
388.ie n .IP """EVBACKEND_DEVPOLL"" (value 16, Solaris 8)" 4 540.ie n .IP """EVBACKEND_DEVPOLL"" (value 16, Solaris 8)" 4
389.el .IP "\f(CWEVBACKEND_DEVPOLL\fR (value 16, Solaris 8)" 4 541.el .IP "\f(CWEVBACKEND_DEVPOLL\fR (value 16, Solaris 8)" 4
390.IX Item "EVBACKEND_DEVPOLL (value 16, Solaris 8)" 542.IX Item "EVBACKEND_DEVPOLL (value 16, Solaris 8)"
391This is not implemented yet (and might never be). 543This is not implemented yet (and might never be, unless you send me an
544implementation). According to reports, \f(CW\*(C`/dev/poll\*(C'\fR only supports sockets
545and is not embeddable, which would limit the usefulness of this backend
546immensely.
392.ie n .IP """EVBACKEND_PORT"" (value 32, Solaris 10)" 4 547.ie n .IP """EVBACKEND_PORT"" (value 32, Solaris 10)" 4
393.el .IP "\f(CWEVBACKEND_PORT\fR (value 32, Solaris 10)" 4 548.el .IP "\f(CWEVBACKEND_PORT\fR (value 32, Solaris 10)" 4
394.IX Item "EVBACKEND_PORT (value 32, Solaris 10)" 549.IX Item "EVBACKEND_PORT (value 32, Solaris 10)"
395This uses the Solaris 10 port mechanism. As with everything on Solaris, 550This uses the Solaris 10 event port mechanism. As with everything on Solaris,
396it's really slow, but it still scales very well (O(active_fds)). 551it's really slow, but it still scales very well (O(active_fds)).
397.Sp 552.Sp
398Please note that solaris ports can result in a lot of spurious 553Please note that solaris event ports can deliver a lot of spurious
399notifications, so you need to use non-blocking I/O or other means to avoid 554notifications, so you need to use non-blocking I/O or other means to avoid
400blocking when no data (or space) is available. 555blocking when no data (or space) is available.
556.Sp
557While this backend scales well, it requires one system call per active
558file descriptor per loop iteration. For small and medium numbers of file
559descriptors a \*(L"slow\*(R" \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or \f(CW\*(C`EVBACKEND_POLL\*(C'\fR backend
560might perform better.
561.Sp
562On the positive side, ignoring the spurious readyness notifications, this
563backend actually performed to specification in all tests and is fully
564embeddable, which is a rare feat among the OS-specific backends.
401.ie n .IP """EVBACKEND_ALL""" 4 565.ie n .IP """EVBACKEND_ALL""" 4
402.el .IP "\f(CWEVBACKEND_ALL\fR" 4 566.el .IP "\f(CWEVBACKEND_ALL\fR" 4
403.IX Item "EVBACKEND_ALL" 567.IX Item "EVBACKEND_ALL"
404Try all backends (even potentially broken ones that wouldn't be tried 568Try all backends (even potentially broken ones that wouldn't be tried
405with \f(CW\*(C`EVFLAG_AUTO\*(C'\fR). Since this is a mask, you can do stuff such as 569with \f(CW\*(C`EVFLAG_AUTO\*(C'\fR). Since this is a mask, you can do stuff such as
406\&\f(CW\*(C`EVBACKEND_ALL & ~EVBACKEND_KQUEUE\*(C'\fR. 570\&\f(CW\*(C`EVBACKEND_ALL & ~EVBACKEND_KQUEUE\*(C'\fR.
571.Sp
572It is definitely not recommended to use this flag.
407.RE 573.RE
408.RS 4 574.RS 4
409.Sp 575.Sp
410If one or more of these are ored into the flags value, then only these 576If one or more of these are ored into the flags value, then only these
411backends will be tried (in the reverse order as given here). If none are 577backends will be tried (in the reverse order as listed here). If none are
412specified, most compiled-in backend will be tried, usually in reverse 578specified, all backends in \f(CW\*(C`ev_recommended_backends ()\*(C'\fR will be tried.
413order of their flag values :)
414.Sp 579.Sp
415The most typical usage is like this: 580The most typical usage is like this:
416.Sp 581.Sp
417.Vb 2 582.Vb 2
418\& if (!ev_default_loop (0)) 583\& if (!ev_default_loop (0))
… …
439Similar to \f(CW\*(C`ev_default_loop\*(C'\fR, but always creates a new event loop that is 604Similar to \f(CW\*(C`ev_default_loop\*(C'\fR, but always creates a new event loop that is
440always distinct from the default loop. Unlike the default loop, it cannot 605always distinct from the default loop. Unlike the default loop, it cannot
441handle signal and child watchers, and attempts to do so will be greeted by 606handle signal and child watchers, and attempts to do so will be greeted by
442undefined behaviour (or a failed assertion if assertions are enabled). 607undefined behaviour (or a failed assertion if assertions are enabled).
443.Sp 608.Sp
609Note that this function \fIis\fR thread-safe, and the recommended way to use
610libev with threads is indeed to create one loop per thread, and using the
611default loop in the \*(L"main\*(R" or \*(L"initial\*(R" thread.
612.Sp
444Example: try to create a event loop that uses epoll and nothing else. 613Example: Try to create a event loop that uses epoll and nothing else.
445.Sp 614.Sp
446.Vb 3 615.Vb 3
447\& struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV); 616\& struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
448\& if (!epoller) 617\& if (!epoller)
449\& fatal ("no epoll found here, maybe it hides under your chair"); 618\& fatal ("no epoll found here, maybe it hides under your chair");
450.Ve 619.Ve
451.IP "ev_default_destroy ()" 4 620.IP "ev_default_destroy ()" 4
452.IX Item "ev_default_destroy ()" 621.IX Item "ev_default_destroy ()"
453Destroys the default loop again (frees all memory and kernel state 622Destroys the default loop again (frees all memory and kernel state
454etc.). This stops all registered event watchers (by not touching them in 623etc.). None of the active event watchers will be stopped in the normal
455any way whatsoever, although you cannot rely on this :). 624sense, so e.g. \f(CW\*(C`ev_is_active\*(C'\fR might still return true. It is your
625responsibility to either stop all watchers cleanly yoursef \fIbefore\fR
626calling this function, or cope with the fact afterwards (which is usually
627the easiest thing, you can just ignore the watchers and/or \f(CW\*(C`free ()\*(C'\fR them
628for example).
629.Sp
630Note that certain global state, such as signal state, will not be freed by
631this function, and related watchers (such as signal and child watchers)
632would need to be stopped manually.
633.Sp
634In general it is not advisable to call this function except in the
635rare occasion where you really need to free e.g. the signal handling
636pipe fds. If you need dynamically allocated loops it is better to use
637\&\f(CW\*(C`ev_loop_new\*(C'\fR and \f(CW\*(C`ev_loop_destroy\*(C'\fR).
456.IP "ev_loop_destroy (loop)" 4 638.IP "ev_loop_destroy (loop)" 4
457.IX Item "ev_loop_destroy (loop)" 639.IX Item "ev_loop_destroy (loop)"
458Like \f(CW\*(C`ev_default_destroy\*(C'\fR, but destroys an event loop created by an 640Like \f(CW\*(C`ev_default_destroy\*(C'\fR, but destroys an event loop created by an
459earlier call to \f(CW\*(C`ev_loop_new\*(C'\fR. 641earlier call to \f(CW\*(C`ev_loop_new\*(C'\fR.
460.IP "ev_default_fork ()" 4 642.IP "ev_default_fork ()" 4
461.IX Item "ev_default_fork ()" 643.IX Item "ev_default_fork ()"
644This function sets a flag that causes subsequent \f(CW\*(C`ev_loop\*(C'\fR iterations
462This function reinitialises the kernel state for backends that have 645to reinitialise the kernel state for backends that have one. Despite the
463one. Despite the name, you can call it anytime, but it makes most sense 646name, you can call it anytime, but it makes most sense after forking, in
464after forking, in either the parent or child process (or both, but that 647the child process (or both child and parent, but that again makes little
465again makes little sense). 648sense). You \fImust\fR call it in the child before using any of the libev
649functions, and it will only take effect at the next \f(CW\*(C`ev_loop\*(C'\fR iteration.
466.Sp 650.Sp
467You \fImust\fR call this function in the child process after forking if and 651On the other hand, you only need to call this function in the child
468only if you want to use the event library in both processes. If you just 652process if and only if you want to use the event library in the child. If
469fork+exec, you don't have to call it. 653you just fork+exec, you don't have to call it at all.
470.Sp 654.Sp
471The function itself is quite fast and it's usually not a problem to call 655The function itself is quite fast and it's usually not a problem to call
472it just in case after a fork. To make this easy, the function will fit in 656it just in case after a fork. To make this easy, the function will fit in
473quite nicely into a call to \f(CW\*(C`pthread_atfork\*(C'\fR: 657quite nicely into a call to \f(CW\*(C`pthread_atfork\*(C'\fR:
474.Sp 658.Sp
475.Vb 1 659.Vb 1
476\& pthread_atfork (0, 0, ev_default_fork); 660\& pthread_atfork (0, 0, ev_default_fork);
477.Ve 661.Ve
478.Sp
479At the moment, \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR and \f(CW\*(C`EVBACKEND_POLL\*(C'\fR are safe to use
480without calling this function, so if you force one of those backends you
481do not need to care.
482.IP "ev_loop_fork (loop)" 4 662.IP "ev_loop_fork (loop)" 4
483.IX Item "ev_loop_fork (loop)" 663.IX Item "ev_loop_fork (loop)"
484Like \f(CW\*(C`ev_default_fork\*(C'\fR, but acts on an event loop created by 664Like \f(CW\*(C`ev_default_fork\*(C'\fR, but acts on an event loop created by
485\&\f(CW\*(C`ev_loop_new\*(C'\fR. Yes, you have to call this on every allocated event loop 665\&\f(CW\*(C`ev_loop_new\*(C'\fR. Yes, you have to call this on every allocated event loop
486after fork, and how you do this is entirely your own problem. 666after fork, and how you do this is entirely your own problem.
667.IP "int ev_is_default_loop (loop)" 4
668.IX Item "int ev_is_default_loop (loop)"
669Returns true when the given loop actually is the default loop, false otherwise.
670.IP "unsigned int ev_loop_count (loop)" 4
671.IX Item "unsigned int ev_loop_count (loop)"
672Returns the count of loop iterations for the loop, which is identical to
673the number of times libev did poll for new events. It starts at \f(CW0\fR and
674happily wraps around with enough iterations.
675.Sp
676This value can sometimes be useful as a generation counter of sorts (it
677\&\*(L"ticks\*(R" the number of loop iterations), as it roughly corresponds with
678\&\f(CW\*(C`ev_prepare\*(C'\fR and \f(CW\*(C`ev_check\*(C'\fR calls.
487.IP "unsigned int ev_backend (loop)" 4 679.IP "unsigned int ev_backend (loop)" 4
488.IX Item "unsigned int ev_backend (loop)" 680.IX Item "unsigned int ev_backend (loop)"
489Returns one of the \f(CW\*(C`EVBACKEND_*\*(C'\fR flags indicating the event backend in 681Returns one of the \f(CW\*(C`EVBACKEND_*\*(C'\fR flags indicating the event backend in
490use. 682use.
491.IP "ev_tstamp ev_now (loop)" 4 683.IP "ev_tstamp ev_now (loop)" 4
492.IX Item "ev_tstamp ev_now (loop)" 684.IX Item "ev_tstamp ev_now (loop)"
493Returns the current \*(L"event loop time\*(R", which is the time the event loop 685Returns the current \*(L"event loop time\*(R", which is the time the event loop
494received events and started processing them. This timestamp does not 686received events and started processing them. This timestamp does not
495change as long as callbacks are being processed, and this is also the base 687change as long as callbacks are being processed, and this is also the base
496time used for relative timers. You can treat it as the timestamp of the 688time used for relative timers. You can treat it as the timestamp of the
497event occuring (or more correctly, libev finding out about it). 689event occurring (or more correctly, libev finding out about it).
498.IP "ev_loop (loop, int flags)" 4 690.IP "ev_loop (loop, int flags)" 4
499.IX Item "ev_loop (loop, int flags)" 691.IX Item "ev_loop (loop, int flags)"
500Finally, this is it, the event handler. This function usually is called 692Finally, this is it, the event handler. This function usually is called
501after you initialised all your watchers and you want to start handling 693after you initialised all your watchers and you want to start handling
502events. 694events.
… …
522libev watchers. However, a pair of \f(CW\*(C`ev_prepare\*(C'\fR/\f(CW\*(C`ev_check\*(C'\fR watchers is 714libev watchers. However, a pair of \f(CW\*(C`ev_prepare\*(C'\fR/\f(CW\*(C`ev_check\*(C'\fR watchers is
523usually a better approach for this kind of thing. 715usually a better approach for this kind of thing.
524.Sp 716.Sp
525Here are the gory details of what \f(CW\*(C`ev_loop\*(C'\fR does: 717Here are the gory details of what \f(CW\*(C`ev_loop\*(C'\fR does:
526.Sp 718.Sp
527.Vb 18 719.Vb 10
528\& * If there are no active watchers (reference count is zero), return. 720\& \- Before the first iteration, call any pending watchers.
529\& - Queue prepare watchers and then call all outstanding watchers. 721\& * If EVFLAG_FORKCHECK was used, check for a fork.
722\& \- If a fork was detected, queue and call all fork watchers.
723\& \- Queue and call all prepare watchers.
530\& - If we have been forked, recreate the kernel state. 724\& \- If we have been forked, recreate the kernel state.
531\& - Update the kernel state with all outstanding changes. 725\& \- Update the kernel state with all outstanding changes.
532\& - Update the "event loop time". 726\& \- Update the "event loop time".
533\& - Calculate for how long to block. 727\& \- Calculate for how long to sleep or block, if at all
728\& (active idle watchers, EVLOOP_NONBLOCK or not having
729\& any active watchers at all will result in not sleeping).
730\& \- Sleep if the I/O and timer collect interval say so.
534\& - Block the process, waiting for any events. 731\& \- Block the process, waiting for any events.
535\& - Queue all outstanding I/O (fd) events. 732\& \- Queue all outstanding I/O (fd) events.
536\& - Update the "event loop time" and do time jump handling. 733\& \- Update the "event loop time" and do time jump handling.
537\& - Queue all outstanding timers. 734\& \- Queue all outstanding timers.
538\& - Queue all outstanding periodics. 735\& \- Queue all outstanding periodics.
539\& - If no events are pending now, queue all idle watchers. 736\& \- If no events are pending now, queue all idle watchers.
540\& - Queue all check watchers. 737\& \- Queue all check watchers.
541\& - Call all queued watchers in reverse order (i.e. check watchers first). 738\& \- Call all queued watchers in reverse order (i.e. check watchers first).
542\& Signals and child watchers are implemented as I/O watchers, and will 739\& Signals and child watchers are implemented as I/O watchers, and will
543\& be handled here by queueing them when their watcher gets executed. 740\& be handled here by queueing them when their watcher gets executed.
544\& - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK 741\& \- If ev_unloop has been called, or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
545\& were used, return, otherwise continue with step *. 742\& were used, or there are no active watchers, return, otherwise
743\& continue with step *.
546.Ve 744.Ve
547.Sp 745.Sp
548Example: queue some jobs and then loop until no events are outsanding 746Example: Queue some jobs and then loop until no events are outstanding
549anymore. 747anymore.
550.Sp 748.Sp
551.Vb 4 749.Vb 4
552\& ... queue jobs here, make sure they register event watchers as long 750\& ... queue jobs here, make sure they register event watchers as long
553\& ... as they still have work to do (even an idle watcher will do..) 751\& ... as they still have work to do (even an idle watcher will do..)
… …
558.IX Item "ev_unloop (loop, how)" 756.IX Item "ev_unloop (loop, how)"
559Can be used to make a call to \f(CW\*(C`ev_loop\*(C'\fR return early (but only after it 757Can be used to make a call to \f(CW\*(C`ev_loop\*(C'\fR return early (but only after it
560has processed all outstanding events). The \f(CW\*(C`how\*(C'\fR argument must be either 758has processed all outstanding events). The \f(CW\*(C`how\*(C'\fR argument must be either
561\&\f(CW\*(C`EVUNLOOP_ONE\*(C'\fR, which will make the innermost \f(CW\*(C`ev_loop\*(C'\fR call return, or 759\&\f(CW\*(C`EVUNLOOP_ONE\*(C'\fR, which will make the innermost \f(CW\*(C`ev_loop\*(C'\fR call return, or
562\&\f(CW\*(C`EVUNLOOP_ALL\*(C'\fR, which will make all nested \f(CW\*(C`ev_loop\*(C'\fR calls return. 760\&\f(CW\*(C`EVUNLOOP_ALL\*(C'\fR, which will make all nested \f(CW\*(C`ev_loop\*(C'\fR calls return.
761.Sp
762This \*(L"unloop state\*(R" will be cleared when entering \f(CW\*(C`ev_loop\*(C'\fR again.
563.IP "ev_ref (loop)" 4 763.IP "ev_ref (loop)" 4
564.IX Item "ev_ref (loop)" 764.IX Item "ev_ref (loop)"
565.PD 0 765.PD 0
566.IP "ev_unref (loop)" 4 766.IP "ev_unref (loop)" 4
567.IX Item "ev_unref (loop)" 767.IX Item "ev_unref (loop)"
… …
573returning, \fIev_unref()\fR after starting, and \fIev_ref()\fR before stopping it. For 773returning, \fIev_unref()\fR after starting, and \fIev_ref()\fR before stopping it. For
574example, libev itself uses this for its internal signal pipe: It is not 774example, libev itself uses this for its internal signal pipe: It is not
575visible to the libev user and should not keep \f(CW\*(C`ev_loop\*(C'\fR from exiting if 775visible to the libev user and should not keep \f(CW\*(C`ev_loop\*(C'\fR from exiting if
576no event watchers registered by it are active. It is also an excellent 776no event watchers registered by it are active. It is also an excellent
577way to do this for generic recurring timers or from within third-party 777way to do this for generic recurring timers or from within third-party
578libraries. Just remember to \fIunref after start\fR and \fIref before stop\fR. 778libraries. Just remember to \fIunref after start\fR and \fIref before stop\fR
779(but only if the watcher wasn't active before, or was active before,
780respectively).
579.Sp 781.Sp
580Example: create a signal watcher, but keep it from keeping \f(CW\*(C`ev_loop\*(C'\fR 782Example: Create a signal watcher, but keep it from keeping \f(CW\*(C`ev_loop\*(C'\fR
581running when nothing else is active. 783running when nothing else is active.
582.Sp 784.Sp
583.Vb 4 785.Vb 4
584\& struct dv_signal exitsig; 786\& struct ev_signal exitsig;
585\& ev_signal_init (&exitsig, sig_cb, SIGINT); 787\& ev_signal_init (&exitsig, sig_cb, SIGINT);
586\& ev_signal_start (myloop, &exitsig); 788\& ev_signal_start (loop, &exitsig);
587\& evf_unref (myloop); 789\& evf_unref (loop);
588.Ve 790.Ve
589.Sp 791.Sp
590Example: for some weird reason, unregister the above signal handler again. 792Example: For some weird reason, unregister the above signal handler again.
591.Sp 793.Sp
592.Vb 2 794.Vb 2
593\& ev_ref (myloop); 795\& ev_ref (loop);
594\& ev_signal_stop (myloop, &exitsig); 796\& ev_signal_stop (loop, &exitsig);
595.Ve 797.Ve
798.IP "ev_set_io_collect_interval (loop, ev_tstamp interval)" 4
799.IX Item "ev_set_io_collect_interval (loop, ev_tstamp interval)"
800.PD 0
801.IP "ev_set_timeout_collect_interval (loop, ev_tstamp interval)" 4
802.IX Item "ev_set_timeout_collect_interval (loop, ev_tstamp interval)"
803.PD
804These advanced functions influence the time that libev will spend waiting
805for events. Both are by default \f(CW0\fR, meaning that libev will try to
806invoke timer/periodic callbacks and I/O callbacks with minimum latency.
807.Sp
808Setting these to a higher value (the \f(CW\*(C`interval\*(C'\fR \fImust\fR be >= \f(CW0\fR)
809allows libev to delay invocation of I/O and timer/periodic callbacks to
810increase efficiency of loop iterations.
811.Sp
812The background is that sometimes your program runs just fast enough to
813handle one (or very few) event(s) per loop iteration. While this makes
814the program responsive, it also wastes a lot of \s-1CPU\s0 time to poll for new
815events, especially with backends like \f(CW\*(C`select ()\*(C'\fR which have a high
816overhead for the actual polling but can deliver many events at once.
817.Sp
818By setting a higher \fIio collect interval\fR you allow libev to spend more
819time collecting I/O events, so you can handle more events per iteration,
820at the cost of increasing latency. Timeouts (both \f(CW\*(C`ev_periodic\*(C'\fR and
821\&\f(CW\*(C`ev_timer\*(C'\fR) will be not affected. Setting this to a non-null value will
822introduce an additional \f(CW\*(C`ev_sleep ()\*(C'\fR call into most loop iterations.
823.Sp
824Likewise, by setting a higher \fItimeout collect interval\fR you allow libev
825to spend more time collecting timeouts, at the expense of increased
826latency (the watcher callback will be called later). \f(CW\*(C`ev_io\*(C'\fR watchers
827will not be affected. Setting this to a non-null value will not introduce
828any overhead in libev.
829.Sp
830Many (busy) programs can usually benefit by setting the io collect
831interval to a value near \f(CW0.1\fR or so, which is often enough for
832interactive servers (of course not for games), likewise for timeouts. It
833usually doesn't make much sense to set it to a lower value than \f(CW0.01\fR,
834as this approsaches the timing granularity of most systems.
596.SH "ANATOMY OF A WATCHER" 835.SH "ANATOMY OF A WATCHER"
597.IX Header "ANATOMY OF A WATCHER" 836.IX Header "ANATOMY OF A WATCHER"
598A watcher is a structure that you create and register to record your 837A watcher is a structure that you create and register to record your
599interest in some event. For instance, if you want to wait for \s-1STDIN\s0 to 838interest in some event. For instance, if you want to wait for \s-1STDIN\s0 to
600become readable, you would create an \f(CW\*(C`ev_io\*(C'\fR watcher for that: 839become readable, you would create an \f(CW\*(C`ev_io\*(C'\fR watcher for that:
… …
603\& static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents) 842\& static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
604\& { 843\& {
605\& ev_io_stop (w); 844\& ev_io_stop (w);
606\& ev_unloop (loop, EVUNLOOP_ALL); 845\& ev_unloop (loop, EVUNLOOP_ALL);
607\& } 846\& }
608.Ve 847\&
609.PP
610.Vb 6
611\& struct ev_loop *loop = ev_default_loop (0); 848\& struct ev_loop *loop = ev_default_loop (0);
612\& struct ev_io stdin_watcher; 849\& struct ev_io stdin_watcher;
613\& ev_init (&stdin_watcher, my_cb); 850\& ev_init (&stdin_watcher, my_cb);
614\& ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ); 851\& ev_io_set (&stdin_watcher, STDIN_FILENO, EV_READ);
615\& ev_io_start (loop, &stdin_watcher); 852\& ev_io_start (loop, &stdin_watcher);
… …
636*)\*(C'\fR), and you can stop watching for events at any time by calling the 873*)\*(C'\fR), and you can stop watching for events at any time by calling the
637corresponding stop function (\f(CW\*(C`ev_<type>_stop (loop, watcher *)\*(C'\fR. 874corresponding stop function (\f(CW\*(C`ev_<type>_stop (loop, watcher *)\*(C'\fR.
638.PP 875.PP
639As long as your watcher is active (has been started but not stopped) you 876As long as your watcher is active (has been started but not stopped) you
640must not touch the values stored in it. Most specifically you must never 877must not touch the values stored in it. Most specifically you must never
641reinitialise it or call its set macro. 878reinitialise it or call its \f(CW\*(C`set\*(C'\fR macro.
642.PP
643You can check whether an event is active by calling the \f(CW\*(C`ev_is_active
644(watcher *)\*(C'\fR macro. To see whether an event is outstanding (but the
645callback for it has not been called yet) you can use the \f(CW\*(C`ev_is_pending
646(watcher *)\*(C'\fR macro.
647.PP 879.PP
648Each and every callback receives the event loop pointer as first, the 880Each and every callback receives the event loop pointer as first, the
649registered watcher structure as second, and a bitset of received events as 881registered watcher structure as second, and a bitset of received events as
650third argument. 882third argument.
651.PP 883.PP
… …
676The signal specified in the \f(CW\*(C`ev_signal\*(C'\fR watcher has been received by a thread. 908The signal specified in the \f(CW\*(C`ev_signal\*(C'\fR watcher has been received by a thread.
677.ie n .IP """EV_CHILD""" 4 909.ie n .IP """EV_CHILD""" 4
678.el .IP "\f(CWEV_CHILD\fR" 4 910.el .IP "\f(CWEV_CHILD\fR" 4
679.IX Item "EV_CHILD" 911.IX Item "EV_CHILD"
680The pid specified in the \f(CW\*(C`ev_child\*(C'\fR watcher has received a status change. 912The pid specified in the \f(CW\*(C`ev_child\*(C'\fR watcher has received a status change.
913.ie n .IP """EV_STAT""" 4
914.el .IP "\f(CWEV_STAT\fR" 4
915.IX Item "EV_STAT"
916The path specified in the \f(CW\*(C`ev_stat\*(C'\fR watcher changed its attributes somehow.
681.ie n .IP """EV_IDLE""" 4 917.ie n .IP """EV_IDLE""" 4
682.el .IP "\f(CWEV_IDLE\fR" 4 918.el .IP "\f(CWEV_IDLE\fR" 4
683.IX Item "EV_IDLE" 919.IX Item "EV_IDLE"
684The \f(CW\*(C`ev_idle\*(C'\fR watcher has determined that you have nothing better to do. 920The \f(CW\*(C`ev_idle\*(C'\fR watcher has determined that you have nothing better to do.
685.ie n .IP """EV_PREPARE""" 4 921.ie n .IP """EV_PREPARE""" 4
… …
695\&\f(CW\*(C`ev_loop\*(C'\fR has gathered them, but before it invokes any callbacks for any 931\&\f(CW\*(C`ev_loop\*(C'\fR has gathered them, but before it invokes any callbacks for any
696received events. Callbacks of both watcher types can start and stop as 932received events. Callbacks of both watcher types can start and stop as
697many watchers as they want, and all of them will be taken into account 933many watchers as they want, and all of them will be taken into account
698(for example, a \f(CW\*(C`ev_prepare\*(C'\fR watcher might start an idle watcher to keep 934(for example, a \f(CW\*(C`ev_prepare\*(C'\fR watcher might start an idle watcher to keep
699\&\f(CW\*(C`ev_loop\*(C'\fR from blocking). 935\&\f(CW\*(C`ev_loop\*(C'\fR from blocking).
936.ie n .IP """EV_EMBED""" 4
937.el .IP "\f(CWEV_EMBED\fR" 4
938.IX Item "EV_EMBED"
939The embedded event loop specified in the \f(CW\*(C`ev_embed\*(C'\fR watcher needs attention.
940.ie n .IP """EV_FORK""" 4
941.el .IP "\f(CWEV_FORK\fR" 4
942.IX Item "EV_FORK"
943The event loop has been resumed in the child process after fork (see
944\&\f(CW\*(C`ev_fork\*(C'\fR).
945.ie n .IP """EV_ASYNC""" 4
946.el .IP "\f(CWEV_ASYNC\fR" 4
947.IX Item "EV_ASYNC"
948The given async watcher has been asynchronously notified (see \f(CW\*(C`ev_async\*(C'\fR).
700.ie n .IP """EV_ERROR""" 4 949.ie n .IP """EV_ERROR""" 4
701.el .IP "\f(CWEV_ERROR\fR" 4 950.el .IP "\f(CWEV_ERROR\fR" 4
702.IX Item "EV_ERROR" 951.IX Item "EV_ERROR"
703An unspecified error has occured, the watcher has been stopped. This might 952An unspecified error has occured, the watcher has been stopped. This might
704happen because the watcher could not be properly started because libev 953happen because the watcher could not be properly started because libev
… …
709Libev will usually signal a few \*(L"dummy\*(R" events together with an error, 958Libev will usually signal a few \*(L"dummy\*(R" events together with an error,
710for example it might indicate that a fd is readable or writable, and if 959for example it might indicate that a fd is readable or writable, and if
711your callbacks is well-written it can just attempt the operation and cope 960your callbacks is well-written it can just attempt the operation and cope
712with the error from \fIread()\fR or \fIwrite()\fR. This will not work in multithreaded 961with the error from \fIread()\fR or \fIwrite()\fR. This will not work in multithreaded
713programs, though, so beware. 962programs, though, so beware.
963.Sh "\s-1GENERIC\s0 \s-1WATCHER\s0 \s-1FUNCTIONS\s0"
964.IX Subsection "GENERIC WATCHER FUNCTIONS"
965In the following description, \f(CW\*(C`TYPE\*(C'\fR stands for the watcher type,
966e.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.
967.ie n .IP """ev_init"" (ev_TYPE *watcher, callback)" 4
968.el .IP "\f(CWev_init\fR (ev_TYPE *watcher, callback)" 4
969.IX Item "ev_init (ev_TYPE *watcher, callback)"
970This macro initialises the generic portion of a watcher. The contents
971of the watcher object can be arbitrary (so \f(CW\*(C`malloc\*(C'\fR will do). Only
972the generic parts of the watcher are initialised, you \fIneed\fR to call
973the type-specific \f(CW\*(C`ev_TYPE_set\*(C'\fR macro afterwards to initialise the
974type-specific parts. For each type there is also a \f(CW\*(C`ev_TYPE_init\*(C'\fR macro
975which rolls both calls into one.
976.Sp
977You can reinitialise a watcher at any time as long as it has been stopped
978(or never started) and there are no pending events outstanding.
979.Sp
980The callback is always of type \f(CW\*(C`void (*)(ev_loop *loop, ev_TYPE *watcher,
981int revents)\*(C'\fR.
982.ie n .IP """ev_TYPE_set"" (ev_TYPE *, [args])" 4
983.el .IP "\f(CWev_TYPE_set\fR (ev_TYPE *, [args])" 4
984.IX Item "ev_TYPE_set (ev_TYPE *, [args])"
985This macro initialises the type-specific parts of a watcher. You need to
986call \f(CW\*(C`ev_init\*(C'\fR at least once before you call this macro, but you can
987call \f(CW\*(C`ev_TYPE_set\*(C'\fR any number of times. You must not, however, call this
988macro on a watcher that is active (it can be pending, however, which is a
989difference to the \f(CW\*(C`ev_init\*(C'\fR macro).
990.Sp
991Although some watcher types do not have type-specific arguments
992(e.g. \f(CW\*(C`ev_prepare\*(C'\fR) you still need to call its \f(CW\*(C`set\*(C'\fR macro.
993.ie n .IP """ev_TYPE_init"" (ev_TYPE *watcher, callback, [args])" 4
994.el .IP "\f(CWev_TYPE_init\fR (ev_TYPE *watcher, callback, [args])" 4
995.IX Item "ev_TYPE_init (ev_TYPE *watcher, callback, [args])"
996This convinience macro rolls both \f(CW\*(C`ev_init\*(C'\fR and \f(CW\*(C`ev_TYPE_set\*(C'\fR macro
997calls into a single call. This is the most convinient method to initialise
998a watcher. The same limitations apply, of course.
999.ie n .IP """ev_TYPE_start"" (loop *, ev_TYPE *watcher)" 4
1000.el .IP "\f(CWev_TYPE_start\fR (loop *, ev_TYPE *watcher)" 4
1001.IX Item "ev_TYPE_start (loop *, ev_TYPE *watcher)"
1002Starts (activates) the given watcher. Only active watchers will receive
1003events. If the watcher is already active nothing will happen.
1004.ie n .IP """ev_TYPE_stop"" (loop *, ev_TYPE *watcher)" 4
1005.el .IP "\f(CWev_TYPE_stop\fR (loop *, ev_TYPE *watcher)" 4
1006.IX Item "ev_TYPE_stop (loop *, ev_TYPE *watcher)"
1007Stops the given watcher again (if active) and clears the pending
1008status. It is possible that stopped watchers are pending (for example,
1009non-repeating timers are being stopped when they become pending), but
1010\&\f(CW\*(C`ev_TYPE_stop\*(C'\fR ensures that the watcher is neither active nor pending. If
1011you want to free or reuse the memory used by the watcher it is therefore a
1012good idea to always call its \f(CW\*(C`ev_TYPE_stop\*(C'\fR function.
1013.IP "bool ev_is_active (ev_TYPE *watcher)" 4
1014.IX Item "bool ev_is_active (ev_TYPE *watcher)"
1015Returns a true value iff the watcher is active (i.e. it has been started
1016and not yet been stopped). As long as a watcher is active you must not modify
1017it.
1018.IP "bool ev_is_pending (ev_TYPE *watcher)" 4
1019.IX Item "bool ev_is_pending (ev_TYPE *watcher)"
1020Returns a true value iff the watcher is pending, (i.e. it has outstanding
1021events but its callback has not yet been invoked). As long as a watcher
1022is pending (but not active) you must not call an init function on it (but
1023\&\f(CW\*(C`ev_TYPE_set\*(C'\fR is safe), you must not change its priority, and you must
1024make sure the watcher is available to libev (e.g. you cannot \f(CW\*(C`free ()\*(C'\fR
1025it).
1026.IP "callback ev_cb (ev_TYPE *watcher)" 4
1027.IX Item "callback ev_cb (ev_TYPE *watcher)"
1028Returns the callback currently set on the watcher.
1029.IP "ev_cb_set (ev_TYPE *watcher, callback)" 4
1030.IX Item "ev_cb_set (ev_TYPE *watcher, callback)"
1031Change the callback. You can change the callback at virtually any time
1032(modulo threads).
1033.IP "ev_set_priority (ev_TYPE *watcher, priority)" 4
1034.IX Item "ev_set_priority (ev_TYPE *watcher, priority)"
1035.PD 0
1036.IP "int ev_priority (ev_TYPE *watcher)" 4
1037.IX Item "int ev_priority (ev_TYPE *watcher)"
1038.PD
1039Set and query the priority of the watcher. The priority is a small
1040integer between \f(CW\*(C`EV_MAXPRI\*(C'\fR (default: \f(CW2\fR) and \f(CW\*(C`EV_MINPRI\*(C'\fR
1041(default: \f(CW\*(C`\-2\*(C'\fR). Pending watchers with higher priority will be invoked
1042before watchers with lower priority, but priority will not keep watchers
1043from being executed (except for \f(CW\*(C`ev_idle\*(C'\fR watchers).
1044.Sp
1045This means that priorities are \fIonly\fR used for ordering callback
1046invocation after new events have been received. This is useful, for
1047example, to reduce latency after idling, or more often, to bind two
1048watchers on the same event and make sure one is called first.
1049.Sp
1050If you need to suppress invocation when higher priority events are pending
1051you need to look at \f(CW\*(C`ev_idle\*(C'\fR watchers, which provide this functionality.
1052.Sp
1053You \fImust not\fR change the priority of a watcher as long as it is active or
1054pending.
1055.Sp
1056The default priority used by watchers when no priority has been set is
1057always \f(CW0\fR, which is supposed to not be too high and not be too low :).
1058.Sp
1059Setting a priority outside the range of \f(CW\*(C`EV_MINPRI\*(C'\fR to \f(CW\*(C`EV_MAXPRI\*(C'\fR is
1060fine, as long as you do not mind that the priority value you query might
1061or might not have been adjusted to be within valid range.
1062.IP "ev_invoke (loop, ev_TYPE *watcher, int revents)" 4
1063.IX Item "ev_invoke (loop, ev_TYPE *watcher, int revents)"
1064Invoke the \f(CW\*(C`watcher\*(C'\fR with the given \f(CW\*(C`loop\*(C'\fR and \f(CW\*(C`revents\*(C'\fR. Neither
1065\&\f(CW\*(C`loop\*(C'\fR nor \f(CW\*(C`revents\*(C'\fR need to be valid as long as the watcher callback
1066can deal with that fact.
1067.IP "int ev_clear_pending (loop, ev_TYPE *watcher)" 4
1068.IX Item "int ev_clear_pending (loop, ev_TYPE *watcher)"
1069If the watcher is pending, this function returns clears its pending status
1070and returns its \f(CW\*(C`revents\*(C'\fR bitset (as if its callback was invoked). If the
1071watcher isn't pending it does nothing and returns \f(CW0\fR.
714.Sh "\s-1ASSOCIATING\s0 \s-1CUSTOM\s0 \s-1DATA\s0 \s-1WITH\s0 A \s-1WATCHER\s0" 1072.Sh "\s-1ASSOCIATING\s0 \s-1CUSTOM\s0 \s-1DATA\s0 \s-1WITH\s0 A \s-1WATCHER\s0"
715.IX Subsection "ASSOCIATING CUSTOM DATA WITH A WATCHER" 1073.IX Subsection "ASSOCIATING CUSTOM DATA WITH A WATCHER"
716Each watcher has, by default, a member \f(CW\*(C`void *data\*(C'\fR that you can change 1074Each watcher has, by default, a member \f(CW\*(C`void *data\*(C'\fR that you can change
717and read at any time, libev will completely ignore it. This can be used 1075and read at any time, libev will completely ignore it. This can be used
718to associate arbitrary data with your watcher. If you need more data and 1076to associate arbitrary data with your watcher. If you need more data and
… …
739\& struct my_io *w = (struct my_io *)w_; 1097\& struct my_io *w = (struct my_io *)w_;
740\& ... 1098\& ...
741\& } 1099\& }
742.Ve 1100.Ve
743.PP 1101.PP
744More interesting and less C\-conformant ways of catsing your callback type 1102More interesting and less C\-conformant ways of casting your callback type
745have been omitted.... 1103instead have been omitted.
1104.PP
1105Another common scenario is having some data structure with multiple
1106watchers:
1107.PP
1108.Vb 6
1109\& struct my_biggy
1110\& {
1111\& int some_data;
1112\& ev_timer t1;
1113\& ev_timer t2;
1114\& }
1115.Ve
1116.PP
1117In this case getting the pointer to \f(CW\*(C`my_biggy\*(C'\fR is a bit more complicated,
1118you need to use \f(CW\*(C`offsetof\*(C'\fR:
1119.PP
1120.Vb 1
1121\& #include <stddef.h>
1122\&
1123\& static void
1124\& t1_cb (EV_P_ struct ev_timer *w, int revents)
1125\& {
1126\& struct my_biggy big = (struct my_biggy *
1127\& (((char *)w) \- offsetof (struct my_biggy, t1));
1128\& }
1129\&
1130\& static void
1131\& t2_cb (EV_P_ struct ev_timer *w, int revents)
1132\& {
1133\& struct my_biggy big = (struct my_biggy *
1134\& (((char *)w) \- offsetof (struct my_biggy, t2));
1135\& }
1136.Ve
746.SH "WATCHER TYPES" 1137.SH "WATCHER TYPES"
747.IX Header "WATCHER TYPES" 1138.IX Header "WATCHER TYPES"
748This section describes each watcher in detail, but will not repeat 1139This section describes each watcher in detail, but will not repeat
749information given in the last section. 1140information given in the last section. Any initialisation/set macros,
1141functions and members specific to the watcher type are explained.
1142.PP
1143Members are additionally marked with either \fI[read\-only]\fR, meaning that,
1144while the watcher is active, you can look at the member and expect some
1145sensible content, but you must not modify it (you can modify it while the
1146watcher is stopped to your hearts content), or \fI[read\-write]\fR, which
1147means you can expect it to have some sensible content while the watcher
1148is active, but you can also modify it. Modifying it may not do something
1149sensible or take immediate effect (or do anything at all), but libev will
1150not crash or malfunction in any way.
750.ie n .Sh """ev_io"" \- is this file descriptor readable or writable" 1151.ie n .Sh """ev_io"" \- is this file descriptor readable or writable?"
751.el .Sh "\f(CWev_io\fP \- is this file descriptor readable or writable" 1152.el .Sh "\f(CWev_io\fP \- is this file descriptor readable or writable?"
752.IX Subsection "ev_io - is this file descriptor readable or writable" 1153.IX Subsection "ev_io - is this file descriptor readable or writable?"
753I/O watchers check whether a file descriptor is readable or writable 1154I/O watchers check whether a file descriptor is readable or writable
754in each iteration of the event loop (This behaviour is called 1155in each iteration of the event loop, or, more precisely, when reading
755level-triggering because you keep receiving events as long as the 1156would not block the process and writing would at least be able to write
756condition persists. Remember you can stop the watcher if you don't want to 1157some data. This behaviour is called level-triggering because you keep
757act on the event and neither want to receive future events). 1158receiving events as long as the condition persists. Remember you can stop
1159the watcher if you don't want to act on the event and neither want to
1160receive future events.
758.PP 1161.PP
759In general you can register as many read and/or write event watchers per 1162In general you can register as many read and/or write event watchers per
760fd as you want (as long as you don't confuse yourself). Setting all file 1163fd as you want (as long as you don't confuse yourself). Setting all file
761descriptors to non-blocking mode is also usually a good idea (but not 1164descriptors to non-blocking mode is also usually a good idea (but not
762required if you know what you are doing). 1165required if you know what you are doing).
763.PP 1166.PP
764You have to be careful with dup'ed file descriptors, though. Some backends
765(the linux epoll backend is a notable example) cannot handle dup'ed file
766descriptors correctly if you register interest in two or more fds pointing
767to the same underlying file/socket etc. description (that is, they share
768the same underlying \*(L"file open\*(R").
769.PP
770If you must do this, then force the use of a known-to-be-good backend 1167If you must do this, then force the use of a known-to-be-good backend
771(at the time of this writing, this includes only \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR and 1168(at the time of this writing, this includes only \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR and
772\&\f(CW\*(C`EVBACKEND_POLL\*(C'\fR). 1169\&\f(CW\*(C`EVBACKEND_POLL\*(C'\fR).
1170.PP
1171Another thing you have to watch out for is that it is quite easy to
1172receive \*(L"spurious\*(R" readyness notifications, that is your callback might
1173be called with \f(CW\*(C`EV_READ\*(C'\fR but a subsequent \f(CW\*(C`read\*(C'\fR(2) will actually block
1174because there is no data. Not only are some backends known to create a
1175lot of those (for example solaris ports), it is very easy to get into
1176this situation even with a relatively standard program structure. Thus
1177it is best to always use non-blocking I/O: An extra \f(CW\*(C`read\*(C'\fR(2) returning
1178\&\f(CW\*(C`EAGAIN\*(C'\fR is far preferable to a program hanging until some data arrives.
1179.PP
1180If you cannot run the fd in non-blocking mode (for example you should not
1181play around with an Xlib connection), then you have to seperately re-test
1182whether a file descriptor is really ready with a known-to-be good interface
1183such as poll (fortunately in our Xlib example, Xlib already does this on
1184its own, so its quite safe to use).
1185.PP
1186\fIThe special problem of disappearing file descriptors\fR
1187.IX Subsection "The special problem of disappearing file descriptors"
1188.PP
1189Some backends (e.g. kqueue, epoll) need to be told about closing a file
1190descriptor (either by calling \f(CW\*(C`close\*(C'\fR explicitly or by any other means,
1191such as \f(CW\*(C`dup\*(C'\fR). The reason is that you register interest in some file
1192descriptor, but when it goes away, the operating system will silently drop
1193this interest. If another file descriptor with the same number then is
1194registered with libev, there is no efficient way to see that this is, in
1195fact, a different file descriptor.
1196.PP
1197To avoid having to explicitly tell libev about such cases, libev follows
1198the following policy: Each time \f(CW\*(C`ev_io_set\*(C'\fR is being called, libev
1199will assume that this is potentially a new file descriptor, otherwise
1200it is assumed that the file descriptor stays the same. That means that
1201you \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
1202descriptor even if the file descriptor number itself did not change.
1203.PP
1204This is how one would do it normally anyway, the important point is that
1205the libev application should not optimise around libev but should leave
1206optimisations to libev.
1207.PP
1208\fIThe special problem of dup'ed file descriptors\fR
1209.IX Subsection "The special problem of dup'ed file descriptors"
1210.PP
1211Some backends (e.g. epoll), cannot register events for file descriptors,
1212but only events for the underlying file descriptions. That means when you
1213have \f(CW\*(C`dup ()\*(C'\fR'ed file descriptors or weirder constellations, and register
1214events for them, only one file descriptor might actually receive events.
1215.PP
1216There is no workaround possible except not registering events
1217for potentially \f(CW\*(C`dup ()\*(C'\fR'ed file descriptors, or to resort to
1218\&\f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or \f(CW\*(C`EVBACKEND_POLL\*(C'\fR.
1219.PP
1220\fIThe special problem of fork\fR
1221.IX Subsection "The special problem of fork"
1222.PP
1223Some backends (epoll, kqueue) do not support \f(CW\*(C`fork ()\*(C'\fR at all or exhibit
1224useless behaviour. Libev fully supports fork, but needs to be told about
1225it in the child.
1226.PP
1227To support fork in your programs, you either have to call
1228\&\f(CW\*(C`ev_default_fork ()\*(C'\fR or \f(CW\*(C`ev_loop_fork ()\*(C'\fR after a fork in the child,
1229enable \f(CW\*(C`EVFLAG_FORKCHECK\*(C'\fR, or resort to \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or
1230\&\f(CW\*(C`EVBACKEND_POLL\*(C'\fR.
1231.PP
1232\fIThe special problem of \s-1SIGPIPE\s0\fR
1233.IX Subsection "The special problem of SIGPIPE"
1234.PP
1235While not really specific to libev, it is easy to forget about \s-1SIGPIPE:\s0
1236when reading from a pipe whose other end has been closed, your program
1237gets send a \s-1SIGPIPE\s0, which, by default, aborts your program. For most
1238programs this is sensible behaviour, for daemons, this is usually
1239undesirable.
1240.PP
1241So when you encounter spurious, unexplained daemon exits, make sure you
1242ignore \s-1SIGPIPE\s0 (and maybe make sure you log the exit status of your daemon
1243somewhere, as that would have given you a big clue).
1244.PP
1245\fIWatcher-Specific Functions\fR
1246.IX Subsection "Watcher-Specific Functions"
773.IP "ev_io_init (ev_io *, callback, int fd, int events)" 4 1247.IP "ev_io_init (ev_io *, callback, int fd, int events)" 4
774.IX Item "ev_io_init (ev_io *, callback, int fd, int events)" 1248.IX Item "ev_io_init (ev_io *, callback, int fd, int events)"
775.PD 0 1249.PD 0
776.IP "ev_io_set (ev_io *, int fd, int events)" 4 1250.IP "ev_io_set (ev_io *, int fd, int events)" 4
777.IX Item "ev_io_set (ev_io *, int fd, int events)" 1251.IX Item "ev_io_set (ev_io *, int fd, int events)"
778.PD 1252.PD
779Configures an \f(CW\*(C`ev_io\*(C'\fR watcher. The fd is the file descriptor to rceeive 1253Configures an \f(CW\*(C`ev_io\*(C'\fR watcher. The \f(CW\*(C`fd\*(C'\fR is the file descriptor to
780events 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 | 1254rceeive events for and events is either \f(CW\*(C`EV_READ\*(C'\fR, \f(CW\*(C`EV_WRITE\*(C'\fR or
781EV_WRITE\*(C'\fR to receive the given events. 1255\&\f(CW\*(C`EV_READ | EV_WRITE\*(C'\fR to receive the given events.
782.Sp 1256.IP "int fd [read\-only]" 4
783Please note that most of the more scalable backend mechanisms (for example 1257.IX Item "int fd [read-only]"
784epoll and solaris ports) can result in spurious readyness notifications 1258The file descriptor being watched.
785for file descriptors, so you practically need to use non-blocking I/O (and 1259.IP "int events [read\-only]" 4
786treat callback invocation as hint only), or retest separately with a safe 1260.IX Item "int events [read-only]"
787interface before doing I/O (XLib can do this), or force the use of either 1261The events being watched.
788\&\f(CW\*(C`EVBACKEND_SELECT\*(C'\fR or \f(CW\*(C`EVBACKEND_POLL\*(C'\fR, which don't suffer from this
789problem. Also note that it is quite easy to have your callback invoked
790when the readyness condition is no longer valid even when employing
791typical ways of handling events, so its a good idea to use non-blocking
792I/O unconditionally.
793.PP 1262.PP
1263\fIExamples\fR
1264.IX Subsection "Examples"
1265.PP
794Example: call \f(CW\*(C`stdin_readable_cb\*(C'\fR when \s-1STDIN_FILENO\s0 has become, well 1266Example: Call \f(CW\*(C`stdin_readable_cb\*(C'\fR when \s-1STDIN_FILENO\s0 has become, well
795readable, but only once. Since it is likely line\-buffered, you could 1267readable, but only once. Since it is likely line-buffered, you could
796attempt to read a whole line in the callback: 1268attempt to read a whole line in the callback.
797.PP 1269.PP
798.Vb 6 1270.Vb 6
799\& static void 1271\& static void
800\& stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents) 1272\& stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
801\& { 1273\& {
802\& ev_io_stop (loop, w); 1274\& ev_io_stop (loop, w);
803\& .. read from stdin here (or from w->fd) and haqndle any I/O errors 1275\& .. read from stdin here (or from w\->fd) and haqndle any I/O errors
804\& } 1276\& }
805.Ve 1277\&
806.PP
807.Vb 6
808\& ... 1278\& ...
809\& struct ev_loop *loop = ev_default_init (0); 1279\& struct ev_loop *loop = ev_default_init (0);
810\& struct ev_io stdin_readable; 1280\& struct ev_io stdin_readable;
811\& ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ); 1281\& ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
812\& ev_io_start (loop, &stdin_readable); 1282\& ev_io_start (loop, &stdin_readable);
813\& ev_loop (loop, 0); 1283\& ev_loop (loop, 0);
814.Ve 1284.Ve
815.ie n .Sh """ev_timer"" \- relative and optionally recurring timeouts" 1285.ie n .Sh """ev_timer"" \- relative and optionally repeating timeouts"
816.el .Sh "\f(CWev_timer\fP \- relative and optionally recurring timeouts" 1286.el .Sh "\f(CWev_timer\fP \- relative and optionally repeating timeouts"
817.IX Subsection "ev_timer - relative and optionally recurring timeouts" 1287.IX Subsection "ev_timer - relative and optionally repeating timeouts"
818Timer watchers are simple relative timers that generate an event after a 1288Timer watchers are simple relative timers that generate an event after a
819given time, and optionally repeating in regular intervals after that. 1289given time, and optionally repeating in regular intervals after that.
820.PP 1290.PP
821The timers are based on real time, that is, if you register an event that 1291The timers are based on real time, that is, if you register an event that
822times out after an hour and you reset your system clock to last years 1292times out after an hour and you reset your system clock to last years
… …
829of the event triggering whatever timeout you are modifying/starting. If 1299of the event triggering whatever timeout you are modifying/starting. If
830you suspect event processing to be delayed and you \fIneed\fR to base the timeout 1300you suspect event processing to be delayed and you \fIneed\fR to base the timeout
831on the current time, use something like this to adjust for this: 1301on the current time, use something like this to adjust for this:
832.PP 1302.PP
833.Vb 1 1303.Vb 1
834\& ev_timer_set (&timer, after + ev_now () - ev_time (), 0.); 1304\& ev_timer_set (&timer, after + ev_now () \- ev_time (), 0.);
835.Ve 1305.Ve
836.PP 1306.PP
837The callback is guarenteed to be invoked only when its timeout has passed, 1307The callback is guarenteed to be invoked only when its timeout has passed,
838but if multiple timers become ready during the same loop iteration then 1308but if multiple timers become ready during the same loop iteration then
839order of execution is undefined. 1309order of execution is undefined.
1310.PP
1311\fIWatcher-Specific Functions and Data Members\fR
1312.IX Subsection "Watcher-Specific Functions and Data Members"
840.IP "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)" 4 1313.IP "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)" 4
841.IX Item "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)" 1314.IX Item "ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)"
842.PD 0 1315.PD 0
843.IP "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)" 4 1316.IP "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)" 4
844.IX Item "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)" 1317.IX Item "ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)"
… …
851The timer itself will do a best-effort at avoiding drift, that is, if you 1324The timer itself will do a best-effort at avoiding drift, that is, if you
852configure a timer to trigger every 10 seconds, then it will trigger at 1325configure a timer to trigger every 10 seconds, then it will trigger at
853exactly 10 second intervals. If, however, your program cannot keep up with 1326exactly 10 second intervals. If, however, your program cannot keep up with
854the timer (because it takes longer than those 10 seconds to do stuff) the 1327the timer (because it takes longer than those 10 seconds to do stuff) the
855timer will not fire more than once per event loop iteration. 1328timer will not fire more than once per event loop iteration.
856.IP "ev_timer_again (loop)" 4 1329.IP "ev_timer_again (loop, ev_timer *)" 4
857.IX Item "ev_timer_again (loop)" 1330.IX Item "ev_timer_again (loop, ev_timer *)"
858This will act as if the timer timed out and restart it again if it is 1331This will act as if the timer timed out and restart it again if it is
859repeating. The exact semantics are: 1332repeating. The exact semantics are:
860.Sp 1333.Sp
1334If the timer is pending, its pending status is cleared.
1335.Sp
861If the timer is started but nonrepeating, stop it. 1336If the timer is started but nonrepeating, stop it (as if it timed out).
862.Sp 1337.Sp
863If the timer is repeating, either start it if necessary (with the repeat 1338If the timer is repeating, either start it if necessary (with the
864value), or reset the running timer to the repeat value. 1339\&\f(CW\*(C`repeat\*(C'\fR value), or reset the running timer to the \f(CW\*(C`repeat\*(C'\fR value.
865.Sp 1340.Sp
866This sounds a bit complicated, but here is a useful and typical 1341This sounds a bit complicated, but here is a useful and typical
867example: Imagine you have a tcp connection and you want a so-called idle 1342example: Imagine you have a tcp connection and you want a so-called idle
868timeout, that is, you want to be called when there have been, say, 60 1343timeout, that is, you want to be called when there have been, say, 60
869seconds of inactivity on the socket. The easiest way to do this is to 1344seconds of inactivity on the socket. The easiest way to do this is to
870configure an \f(CW\*(C`ev_timer\*(C'\fR with after=repeat=60 and calling ev_timer_again each 1345configure an \f(CW\*(C`ev_timer\*(C'\fR with a \f(CW\*(C`repeat\*(C'\fR value of \f(CW60\fR and then call
871time you successfully read or write some data. If you go into an idle 1346\&\f(CW\*(C`ev_timer_again\*(C'\fR each time you successfully read or write some data. If
872state where you do not expect data to travel on the socket, you can stop 1347you go into an idle state where you do not expect data to travel on the
873the timer, and again will automatically restart it if need be. 1348socket, you can \f(CW\*(C`ev_timer_stop\*(C'\fR the timer, and \f(CW\*(C`ev_timer_again\*(C'\fR will
1349automatically restart it if need be.
1350.Sp
1351That means you can ignore the \f(CW\*(C`after\*(C'\fR value and \f(CW\*(C`ev_timer_start\*(C'\fR
1352altogether and only ever use the \f(CW\*(C`repeat\*(C'\fR value and \f(CW\*(C`ev_timer_again\*(C'\fR:
1353.Sp
1354.Vb 8
1355\& ev_timer_init (timer, callback, 0., 5.);
1356\& ev_timer_again (loop, timer);
1357\& ...
1358\& timer\->again = 17.;
1359\& ev_timer_again (loop, timer);
1360\& ...
1361\& timer\->again = 10.;
1362\& ev_timer_again (loop, timer);
1363.Ve
1364.Sp
1365This is more slightly efficient then stopping/starting the timer each time
1366you want to modify its timeout value.
1367.IP "ev_tstamp repeat [read\-write]" 4
1368.IX Item "ev_tstamp repeat [read-write]"
1369The current \f(CW\*(C`repeat\*(C'\fR value. Will be used each time the watcher times out
1370or \f(CW\*(C`ev_timer_again\*(C'\fR is called and determines the next timeout (if any),
1371which is also when any modifications are taken into account.
874.PP 1372.PP
1373\fIExamples\fR
1374.IX Subsection "Examples"
1375.PP
875Example: create a timer that fires after 60 seconds. 1376Example: Create a timer that fires after 60 seconds.
876.PP 1377.PP
877.Vb 5 1378.Vb 5
878\& static void 1379\& static void
879\& one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents) 1380\& one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
880\& { 1381\& {
881\& .. one minute over, w is actually stopped right here 1382\& .. one minute over, w is actually stopped right here
882\& } 1383\& }
883.Ve 1384\&
884.PP
885.Vb 3
886\& struct ev_timer mytimer; 1385\& struct ev_timer mytimer;
887\& ev_timer_init (&mytimer, one_minute_cb, 60., 0.); 1386\& ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
888\& ev_timer_start (loop, &mytimer); 1387\& ev_timer_start (loop, &mytimer);
889.Ve 1388.Ve
890.PP 1389.PP
891Example: create a timeout timer that times out after 10 seconds of 1390Example: Create a timeout timer that times out after 10 seconds of
892inactivity. 1391inactivity.
893.PP 1392.PP
894.Vb 5 1393.Vb 5
895\& static void 1394\& static void
896\& timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents) 1395\& timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
897\& { 1396\& {
898\& .. ten seconds without any activity 1397\& .. ten seconds without any activity
899\& } 1398\& }
900.Ve 1399\&
901.PP
902.Vb 4
903\& struct ev_timer mytimer; 1400\& struct ev_timer mytimer;
904\& ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */ 1401\& ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
905\& ev_timer_again (&mytimer); /* start timer */ 1402\& ev_timer_again (&mytimer); /* start timer */
906\& ev_loop (loop, 0); 1403\& ev_loop (loop, 0);
907.Ve 1404\&
908.PP
909.Vb 3
910\& // and in some piece of code that gets executed on any "activity": 1405\& // and in some piece of code that gets executed on any "activity":
911\& // reset the timeout to start ticking again at 10 seconds 1406\& // reset the timeout to start ticking again at 10 seconds
912\& ev_timer_again (&mytimer); 1407\& ev_timer_again (&mytimer);
913.Ve 1408.Ve
914.ie n .Sh """ev_periodic"" \- to cron or not to cron" 1409.ie n .Sh """ev_periodic"" \- to cron or not to cron?"
915.el .Sh "\f(CWev_periodic\fP \- to cron or not to cron" 1410.el .Sh "\f(CWev_periodic\fP \- to cron or not to cron?"
916.IX Subsection "ev_periodic - to cron or not to cron" 1411.IX Subsection "ev_periodic - to cron or not to cron?"
917Periodic watchers are also timers of a kind, but they are very versatile 1412Periodic watchers are also timers of a kind, but they are very versatile
918(and unfortunately a bit complex). 1413(and unfortunately a bit complex).
919.PP 1414.PP
920Unlike \f(CW\*(C`ev_timer\*(C'\fR's, they are not based on real time (or relative time) 1415Unlike \f(CW\*(C`ev_timer\*(C'\fR's, they are not based on real time (or relative time)
921but on wallclock time (absolute time). You can tell a periodic watcher 1416but on wallclock time (absolute time). You can tell a periodic watcher
922to trigger \*(L"at\*(R" some specific point in time. For example, if you tell a 1417to trigger \*(L"at\*(R" some specific point in time. For example, if you tell a
923periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now () 1418periodic watcher to trigger in 10 seconds (by specifiying e.g. \f(CW\*(C`ev_now ()
924+ 10.>) and then reset your system clock to the last year, then it will 1419+ 10.\*(C'\fR) and then reset your system clock to the last year, then it will
925take a year to trigger the event (unlike an \f(CW\*(C`ev_timer\*(C'\fR, which would trigger 1420take a year to trigger the event (unlike an \f(CW\*(C`ev_timer\*(C'\fR, which would trigger
926roughly 10 seconds later and of course not if you reset your system time 1421roughly 10 seconds later).
927again).
928.PP 1422.PP
929They can also be used to implement vastly more complex timers, such as 1423They can also be used to implement vastly more complex timers, such as
930triggering an event on eahc midnight, local time. 1424triggering an event on each midnight, local time or other, complicated,
1425rules.
931.PP 1426.PP
932As with timers, the callback is guarenteed to be invoked only when the 1427As with timers, the callback is guarenteed to be invoked only when the
933time (\f(CW\*(C`at\*(C'\fR) has been passed, but if multiple periodic timers become ready 1428time (\f(CW\*(C`at\*(C'\fR) has been passed, but if multiple periodic timers become ready
934during the same loop iteration then order of execution is undefined. 1429during the same loop iteration then order of execution is undefined.
1430.PP
1431\fIWatcher-Specific Functions and Data Members\fR
1432.IX Subsection "Watcher-Specific Functions and Data Members"
935.IP "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 4 1433.IP "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 4
936.IX Item "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)" 1434.IX Item "ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)"
937.PD 0 1435.PD 0
938.IP "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)" 4 1436.IP "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)" 4
939.IX Item "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)" 1437.IX Item "ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)"
940.PD 1438.PD
941Lots of arguments, lets sort it out... There are basically three modes of 1439Lots of arguments, lets sort it out... There are basically three modes of
942operation, and we will explain them from simplest to complex: 1440operation, and we will explain them from simplest to complex:
943.RS 4 1441.RS 4
1442.IP "\(bu" 4
944.IP "* absolute timer (interval = reschedule_cb = 0)" 4 1443absolute timer (at = time, interval = reschedule_cb = 0)
945.IX Item "absolute timer (interval = reschedule_cb = 0)" 1444.Sp
946In this configuration the watcher triggers an event at the wallclock time 1445In this configuration the watcher triggers an event at the wallclock time
947\&\f(CW\*(C`at\*(C'\fR and doesn't repeat. It will not adjust when a time jump occurs, 1446\&\f(CW\*(C`at\*(C'\fR and doesn't repeat. It will not adjust when a time jump occurs,
948that is, if it is to be run at January 1st 2011 then it will run when the 1447that is, if it is to be run at January 1st 2011 then it will run when the
949system time reaches or surpasses this time. 1448system time reaches or surpasses this time.
1449.IP "\(bu" 4
950.IP "* non-repeating interval timer (interval > 0, reschedule_cb = 0)" 4 1450repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
951.IX Item "non-repeating interval timer (interval > 0, reschedule_cb = 0)" 1451.Sp
952In this mode the watcher will always be scheduled to time out at the next 1452In this mode the watcher will always be scheduled to time out at the next
953\&\f(CW\*(C`at + N * interval\*(C'\fR time (for some integer N) and then repeat, regardless 1453\&\f(CW\*(C`at + N * interval\*(C'\fR time (for some integer N, which can also be negative)
954of any time jumps. 1454and then repeat, regardless of any time jumps.
955.Sp 1455.Sp
956This can be used to create timers that do not drift with respect to system 1456This can be used to create timers that do not drift with respect to system
957time: 1457time:
958.Sp 1458.Sp
959.Vb 1 1459.Vb 1
… …
966by 3600. 1466by 3600.
967.Sp 1467.Sp
968Another way to think about it (for the mathematically inclined) is that 1468Another way to think about it (for the mathematically inclined) is that
969\&\f(CW\*(C`ev_periodic\*(C'\fR will try to run the callback in this mode at the next possible 1469\&\f(CW\*(C`ev_periodic\*(C'\fR will try to run the callback in this mode at the next possible
970time where \f(CW\*(C`time = at (mod interval)\*(C'\fR, regardless of any time jumps. 1470time where \f(CW\*(C`time = at (mod interval)\*(C'\fR, regardless of any time jumps.
971.IP "* manual reschedule mode (reschedule_cb = callback)" 4 1471.Sp
972.IX Item "manual reschedule mode (reschedule_cb = callback)" 1472For numerical stability it is preferable that the \f(CW\*(C`at\*(C'\fR value is near
1473\&\f(CW\*(C`ev_now ()\*(C'\fR (the current time), but there is no range requirement for
1474this value.
1475.IP "\(bu" 4
1476manual reschedule mode (at and interval ignored, reschedule_cb = callback)
1477.Sp
973In this mode the values for \f(CW\*(C`interval\*(C'\fR and \f(CW\*(C`at\*(C'\fR are both being 1478In this mode the values for \f(CW\*(C`interval\*(C'\fR and \f(CW\*(C`at\*(C'\fR are both being
974ignored. Instead, each time the periodic watcher gets scheduled, the 1479ignored. Instead, each time the periodic watcher gets scheduled, the
975reschedule callback will be called with the watcher as first, and the 1480reschedule callback will be called with the watcher as first, and the
976current time as second argument. 1481current time as second argument.
977.Sp 1482.Sp
978\&\s-1NOTE:\s0 \fIThis callback \s-1MUST\s0 \s-1NOT\s0 stop or destroy any periodic watcher, 1483\&\s-1NOTE:\s0 \fIThis callback \s-1MUST\s0 \s-1NOT\s0 stop or destroy any periodic watcher,
979ever, or make any event loop modifications\fR. If you need to stop it, 1484ever, or make any event loop modifications\fR. If you need to stop it,
980return \f(CW\*(C`now + 1e30\*(C'\fR (or so, fudge fudge) and stop it afterwards (e.g. by 1485return \f(CW\*(C`now + 1e30\*(C'\fR (or so, fudge fudge) and stop it afterwards (e.g. by
981starting a prepare watcher). 1486starting an \f(CW\*(C`ev_prepare\*(C'\fR watcher, which is legal).
982.Sp 1487.Sp
983Its prototype is \f(CW\*(C`ev_tstamp (*reschedule_cb)(struct ev_periodic *w, 1488Its prototype is \f(CW\*(C`ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
984ev_tstamp now)\*(C'\fR, e.g.: 1489ev_tstamp now)\*(C'\fR, e.g.:
985.Sp 1490.Sp
986.Vb 4 1491.Vb 4
… …
1010.IX Item "ev_periodic_again (loop, ev_periodic *)" 1515.IX Item "ev_periodic_again (loop, ev_periodic *)"
1011Simply stops and restarts the periodic watcher again. This is only useful 1516Simply stops and restarts the periodic watcher again. This is only useful
1012when you changed some parameters or the reschedule callback would return 1517when you changed some parameters or the reschedule callback would return
1013a different time than the last time it was called (e.g. in a crond like 1518a different time than the last time it was called (e.g. in a crond like
1014program when the crontabs have changed). 1519program when the crontabs have changed).
1520.IP "ev_tstamp offset [read\-write]" 4
1521.IX Item "ev_tstamp offset [read-write]"
1522When repeating, this contains the offset value, otherwise this is the
1523absolute point in time (the \f(CW\*(C`at\*(C'\fR value passed to \f(CW\*(C`ev_periodic_set\*(C'\fR).
1524.Sp
1525Can be modified any time, but changes only take effect when the periodic
1526timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being called.
1527.IP "ev_tstamp interval [read\-write]" 4
1528.IX Item "ev_tstamp interval [read-write]"
1529The current interval value. Can be modified any time, but changes only
1530take effect when the periodic timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being
1531called.
1532.IP "ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read\-write]" 4
1533.IX Item "ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]"
1534The current reschedule callback, or \f(CW0\fR, if this functionality is
1535switched off. Can be changed any time, but changes only take effect when
1536the periodic timer fires or \f(CW\*(C`ev_periodic_again\*(C'\fR is being called.
1537.IP "ev_tstamp at [read\-only]" 4
1538.IX Item "ev_tstamp at [read-only]"
1539When active, contains the absolute time that the watcher is supposed to
1540trigger next.
1015.PP 1541.PP
1542\fIExamples\fR
1543.IX Subsection "Examples"
1544.PP
1016Example: call a callback every hour, or, more precisely, whenever the 1545Example: Call a callback every hour, or, more precisely, whenever the
1017system clock is divisible by 3600. The callback invocation times have 1546system clock is divisible by 3600. The callback invocation times have
1018potentially a lot of jittering, but good long-term stability. 1547potentially a lot of jittering, but good long-term stability.
1019.PP 1548.PP
1020.Vb 5 1549.Vb 5
1021\& static void 1550\& static void
1022\& clock_cb (struct ev_loop *loop, struct ev_io *w, int revents) 1551\& clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1023\& { 1552\& {
1024\& ... its now a full hour (UTC, or TAI or whatever your clock follows) 1553\& ... its now a full hour (UTC, or TAI or whatever your clock follows)
1025\& } 1554\& }
1026.Ve 1555\&
1027.PP
1028.Vb 3
1029\& struct ev_periodic hourly_tick; 1556\& struct ev_periodic hourly_tick;
1030\& ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0); 1557\& ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
1031\& ev_periodic_start (loop, &hourly_tick); 1558\& ev_periodic_start (loop, &hourly_tick);
1032.Ve 1559.Ve
1033.PP 1560.PP
1034Example: the same as above, but use a reschedule callback to do it: 1561Example: The same as above, but use a reschedule callback to do it:
1035.PP 1562.PP
1036.Vb 1 1563.Vb 1
1037\& #include <math.h> 1564\& #include <math.h>
1038.Ve 1565\&
1039.PP
1040.Vb 5
1041\& static ev_tstamp 1566\& static ev_tstamp
1042\& my_scheduler_cb (struct ev_periodic *w, ev_tstamp now) 1567\& my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1043\& { 1568\& {
1044\& return fmod (now, 3600.) + 3600.; 1569\& return fmod (now, 3600.) + 3600.;
1045\& } 1570\& }
1046.Ve 1571\&
1047.PP
1048.Vb 1
1049\& ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb); 1572\& ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1050.Ve 1573.Ve
1051.PP 1574.PP
1052Example: call a callback every hour, starting now: 1575Example: Call a callback every hour, starting now:
1053.PP 1576.PP
1054.Vb 4 1577.Vb 4
1055\& struct ev_periodic hourly_tick; 1578\& struct ev_periodic hourly_tick;
1056\& ev_periodic_init (&hourly_tick, clock_cb, 1579\& ev_periodic_init (&hourly_tick, clock_cb,
1057\& fmod (ev_now (loop), 3600.), 3600., 0); 1580\& fmod (ev_now (loop), 3600.), 3600., 0);
1058\& ev_periodic_start (loop, &hourly_tick); 1581\& ev_periodic_start (loop, &hourly_tick);
1059.Ve 1582.Ve
1060.ie n .Sh """ev_signal"" \- signal me when a signal gets signalled" 1583.ie n .Sh """ev_signal"" \- signal me when a signal gets signalled!"
1061.el .Sh "\f(CWev_signal\fP \- signal me when a signal gets signalled" 1584.el .Sh "\f(CWev_signal\fP \- signal me when a signal gets signalled!"
1062.IX Subsection "ev_signal - signal me when a signal gets signalled" 1585.IX Subsection "ev_signal - signal me when a signal gets signalled!"
1063Signal watchers will trigger an event when the process receives a specific 1586Signal watchers will trigger an event when the process receives a specific
1064signal one or more times. Even though signals are very asynchronous, libev 1587signal one or more times. Even though signals are very asynchronous, libev
1065will try it's best to deliver signals synchronously, i.e. as part of the 1588will try it's best to deliver signals synchronously, i.e. as part of the
1066normal event processing, like any other event. 1589normal event processing, like any other event.
1067.PP 1590.PP
… …
1069first watcher gets started will libev actually register a signal watcher 1592first watcher gets started will libev actually register a signal watcher
1070with the kernel (thus it coexists with your own signal handlers as long 1593with the kernel (thus it coexists with your own signal handlers as long
1071as you don't register any with libev). Similarly, when the last signal 1594as you don't register any with libev). Similarly, when the last signal
1072watcher for a signal is stopped libev will reset the signal handler to 1595watcher for a signal is stopped libev will reset the signal handler to
1073\&\s-1SIG_DFL\s0 (regardless of what it was set to before). 1596\&\s-1SIG_DFL\s0 (regardless of what it was set to before).
1597.PP
1598If possible and supported, libev will install its handlers with
1599\&\f(CW\*(C`SA_RESTART\*(C'\fR behaviour enabled, so syscalls should not be unduly
1600interrupted. If you have a problem with syscalls getting interrupted by
1601signals you can block all signals in an \f(CW\*(C`ev_check\*(C'\fR watcher and unblock
1602them in an \f(CW\*(C`ev_prepare\*(C'\fR watcher.
1603.PP
1604\fIWatcher-Specific Functions and Data Members\fR
1605.IX Subsection "Watcher-Specific Functions and Data Members"
1074.IP "ev_signal_init (ev_signal *, callback, int signum)" 4 1606.IP "ev_signal_init (ev_signal *, callback, int signum)" 4
1075.IX Item "ev_signal_init (ev_signal *, callback, int signum)" 1607.IX Item "ev_signal_init (ev_signal *, callback, int signum)"
1076.PD 0 1608.PD 0
1077.IP "ev_signal_set (ev_signal *, int signum)" 4 1609.IP "ev_signal_set (ev_signal *, int signum)" 4
1078.IX Item "ev_signal_set (ev_signal *, int signum)" 1610.IX Item "ev_signal_set (ev_signal *, int signum)"
1079.PD 1611.PD
1080Configures the watcher to trigger on the given signal number (usually one 1612Configures the watcher to trigger on the given signal number (usually one
1081of the \f(CW\*(C`SIGxxx\*(C'\fR constants). 1613of the \f(CW\*(C`SIGxxx\*(C'\fR constants).
1614.IP "int signum [read\-only]" 4
1615.IX Item "int signum [read-only]"
1616The signal the watcher watches out for.
1617.PP
1618\fIExamples\fR
1619.IX Subsection "Examples"
1620.PP
1621Example: Try to exit cleanly on \s-1SIGINT\s0 and \s-1SIGTERM\s0.
1622.PP
1623.Vb 5
1624\& static void
1625\& sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1626\& {
1627\& ev_unloop (loop, EVUNLOOP_ALL);
1628\& }
1629\&
1630\& struct ev_signal signal_watcher;
1631\& ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
1632\& ev_signal_start (loop, &sigint_cb);
1633.Ve
1082.ie n .Sh """ev_child"" \- wait for pid status changes" 1634.ie n .Sh """ev_child"" \- watch out for process status changes"
1083.el .Sh "\f(CWev_child\fP \- wait for pid status changes" 1635.el .Sh "\f(CWev_child\fP \- watch out for process status changes"
1084.IX Subsection "ev_child - wait for pid status changes" 1636.IX Subsection "ev_child - watch out for process status changes"
1085Child watchers trigger when your process receives a \s-1SIGCHLD\s0 in response to 1637Child watchers trigger when your process receives a \s-1SIGCHLD\s0 in response to
1086some child status changes (most typically when a child of yours dies). 1638some child status changes (most typically when a child of yours dies). It
1639is permissible to install a child watcher \fIafter\fR the child has been
1640forked (which implies it might have already exited), as long as the event
1641loop isn't entered (or is continued from a watcher).
1642.PP
1643Only the default event loop is capable of handling signals, and therefore
1644you can only rgeister child watchers in the default event loop.
1645.PP
1646\fIProcess Interaction\fR
1647.IX Subsection "Process Interaction"
1648.PP
1649Libev grabs \f(CW\*(C`SIGCHLD\*(C'\fR as soon as the default event loop is
1650initialised. This is necessary to guarantee proper behaviour even if
1651the first child watcher is started after the child exits. The occurance
1652of \f(CW\*(C`SIGCHLD\*(C'\fR is recorded asynchronously, but child reaping is done
1653synchronously as part of the event loop processing. Libev always reaps all
1654children, even ones not watched.
1655.PP
1656\fIOverriding the Built-In Processing\fR
1657.IX Subsection "Overriding the Built-In Processing"
1658.PP
1659Libev offers no special support for overriding the built-in child
1660processing, but if your application collides with libev's default child
1661handler, you can override it easily by installing your own handler for
1662\&\f(CW\*(C`SIGCHLD\*(C'\fR after initialising the default loop, and making sure the
1663default loop never gets destroyed. You are encouraged, however, to use an
1664event-based approach to child reaping and thus use libev's support for
1665that, so other libev users can use \f(CW\*(C`ev_child\*(C'\fR watchers freely.
1666.PP
1667\fIWatcher-Specific Functions and Data Members\fR
1668.IX Subsection "Watcher-Specific Functions and Data Members"
1087.IP "ev_child_init (ev_child *, callback, int pid)" 4 1669.IP "ev_child_init (ev_child *, callback, int pid, int trace)" 4
1088.IX Item "ev_child_init (ev_child *, callback, int pid)" 1670.IX Item "ev_child_init (ev_child *, callback, int pid, int trace)"
1089.PD 0 1671.PD 0
1090.IP "ev_child_set (ev_child *, int pid)" 4 1672.IP "ev_child_set (ev_child *, int pid, int trace)" 4
1091.IX Item "ev_child_set (ev_child *, int pid)" 1673.IX Item "ev_child_set (ev_child *, int pid, int trace)"
1092.PD 1674.PD
1093Configures the watcher to wait for status changes of process \f(CW\*(C`pid\*(C'\fR (or 1675Configures the watcher to wait for status changes of process \f(CW\*(C`pid\*(C'\fR (or
1094\&\fIany\fR process if \f(CW\*(C`pid\*(C'\fR is specified as \f(CW0\fR). The callback can look 1676\&\fIany\fR process if \f(CW\*(C`pid\*(C'\fR is specified as \f(CW0\fR). The callback can look
1095at the \f(CW\*(C`rstatus\*(C'\fR member of the \f(CW\*(C`ev_child\*(C'\fR watcher structure to see 1677at the \f(CW\*(C`rstatus\*(C'\fR member of the \f(CW\*(C`ev_child\*(C'\fR watcher structure to see
1096the status word (use the macros from \f(CW\*(C`sys/wait.h\*(C'\fR and see your systems 1678the status word (use the macros from \f(CW\*(C`sys/wait.h\*(C'\fR and see your systems
1097\&\f(CW\*(C`waitpid\*(C'\fR documentation). The \f(CW\*(C`rpid\*(C'\fR member contains the pid of the 1679\&\f(CW\*(C`waitpid\*(C'\fR documentation). The \f(CW\*(C`rpid\*(C'\fR member contains the pid of the
1098process causing the status change. 1680process causing the status change. \f(CW\*(C`trace\*(C'\fR must be either \f(CW0\fR (only
1681activate the watcher when the process terminates) or \f(CW1\fR (additionally
1682activate the watcher when the process is stopped or continued).
1683.IP "int pid [read\-only]" 4
1684.IX Item "int pid [read-only]"
1685The process id this watcher watches out for, or \f(CW0\fR, meaning any process id.
1686.IP "int rpid [read\-write]" 4
1687.IX Item "int rpid [read-write]"
1688The process id that detected a status change.
1689.IP "int rstatus [read\-write]" 4
1690.IX Item "int rstatus [read-write]"
1691The process exit/trace status caused by \f(CW\*(C`rpid\*(C'\fR (see your systems
1692\&\f(CW\*(C`waitpid\*(C'\fR and \f(CW\*(C`sys/wait.h\*(C'\fR documentation for details).
1099.PP 1693.PP
1100Example: try to exit cleanly on \s-1SIGINT\s0 and \s-1SIGTERM\s0. 1694\fIExamples\fR
1695.IX Subsection "Examples"
1101.PP 1696.PP
1697Example: \f(CW\*(C`fork()\*(C'\fR a new process and install a child handler to wait for
1698its completion.
1699.PP
1102.Vb 5 1700.Vb 1
1701\& ev_child cw;
1702\&
1103\& static void 1703\& static void
1104\& sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents) 1704\& child_cb (EV_P_ struct ev_child *w, int revents)
1105\& { 1705\& {
1106\& ev_unloop (loop, EVUNLOOP_ALL); 1706\& ev_child_stop (EV_A_ w);
1707\& printf ("process %d exited with status %x\en", w\->rpid, w\->rstatus);
1107\& } 1708\& }
1709\&
1710\& pid_t pid = fork ();
1711\&
1712\& if (pid < 0)
1713\& // error
1714\& else if (pid == 0)
1715\& {
1716\& // the forked child executes here
1717\& exit (1);
1718\& }
1719\& else
1720\& {
1721\& ev_child_init (&cw, child_cb, pid, 0);
1722\& ev_child_start (EV_DEFAULT_ &cw);
1723\& }
1108.Ve 1724.Ve
1725.ie n .Sh """ev_stat"" \- did the file attributes just change?"
1726.el .Sh "\f(CWev_stat\fP \- did the file attributes just change?"
1727.IX Subsection "ev_stat - did the file attributes just change?"
1728This watches a filesystem path for attribute changes. That is, it calls
1729\&\f(CW\*(C`stat\*(C'\fR regularly (or when the \s-1OS\s0 says it changed) and sees if it changed
1730compared to the last time, invoking the callback if it did.
1109.PP 1731.PP
1732The path does not need to exist: changing from \*(L"path exists\*(R" to \*(L"path does
1733not exist\*(R" is a status change like any other. The condition \*(L"path does
1734not exist\*(R" is signified by the \f(CW\*(C`st_nlink\*(C'\fR field being zero (which is
1735otherwise always forced to be at least one) and all the other fields of
1736the stat buffer having unspecified contents.
1737.PP
1738The path \fIshould\fR be absolute and \fImust not\fR end in a slash. If it is
1739relative and your working directory changes, the behaviour is undefined.
1740.PP
1741Since there is no standard to do this, the portable implementation simply
1742calls \f(CW\*(C`stat (2)\*(C'\fR regularly on the path to see if it changed somehow. You
1743can specify a recommended polling interval for this case. If you specify
1744a polling interval of \f(CW0\fR (highly recommended!) then a \fIsuitable,
1745unspecified default\fR value will be used (which you can expect to be around
1746five seconds, although this might change dynamically). Libev will also
1747impose a minimum interval which is currently around \f(CW0.1\fR, but thats
1748usually overkill.
1749.PP
1750This watcher type is not meant for massive numbers of stat watchers,
1751as even with OS-supported change notifications, this can be
1752resource-intensive.
1753.PP
1754At the time of this writing, only the Linux inotify interface is
1755implemented (implementing kqueue support is left as an exercise for the
1756reader). Inotify will be used to give hints only and should not change the
1757semantics of \f(CW\*(C`ev_stat\*(C'\fR watchers, which means that libev sometimes needs
1758to fall back to regular polling again even with inotify, but changes are
1759usually detected immediately, and if the file exists there will be no
1760polling.
1761.PP
1762\fI\s-1ABI\s0 Issues (Largefile Support)\fR
1763.IX Subsection "ABI Issues (Largefile Support)"
1764.PP
1765Libev by default (unless the user overrides this) uses the default
1766compilation environment, which means that on systems with optionally
1767disabled large file support, you get the 32 bit version of the stat
1768structure. When using the library from programs that change the \s-1ABI\s0 to
1769use 64 bit file offsets the programs will fail. In that case you have to
1770compile libev with the same flags to get binary compatibility. This is
1771obviously the case with any flags that change the \s-1ABI\s0, but the problem is
1772most noticably with ev_stat and largefile support.
1773.PP
1774\fIInotify\fR
1775.IX Subsection "Inotify"
1776.PP
1777When \f(CW\*(C`inotify (7)\*(C'\fR support has been compiled into libev (generally only
1778available on Linux) and present at runtime, it will be used to speed up
1779change detection where possible. The inotify descriptor will be created lazily
1780when the first \f(CW\*(C`ev_stat\*(C'\fR watcher is being started.
1781.PP
1782Inotify presense does not change the semantics of \f(CW\*(C`ev_stat\*(C'\fR watchers
1783except that changes might be detected earlier, and in some cases, to avoid
1784making regular \f(CW\*(C`stat\*(C'\fR calls. Even in the presense of inotify support
1785there are many cases where libev has to resort to regular \f(CW\*(C`stat\*(C'\fR polling.
1786.PP
1787(There is no support for kqueue, as apparently it cannot be used to
1788implement this functionality, due to the requirement of having a file
1789descriptor open on the object at all times).
1790.PP
1791\fIThe special problem of stat time resolution\fR
1792.IX Subsection "The special problem of stat time resolution"
1793.PP
1794The \f(CW\*(C`stat ()\*(C'\fR syscall only supports full-second resolution portably, and
1795even on systems where the resolution is higher, many filesystems still
1796only support whole seconds.
1797.PP
1798That means that, if the time is the only thing that changes, you might
1799miss updates: on the first update, \f(CW\*(C`ev_stat\*(C'\fR detects a change and calls
1800your callback, which does something. When there is another update within
1801the same second, \f(CW\*(C`ev_stat\*(C'\fR will be unable to detect it.
1802.PP
1803The solution to this is to delay acting on a change for a second (or till
1804the next second boundary), using a roughly one-second delay \f(CW\*(C`ev_timer\*(C'\fR
1805(\f(CW\*(C`ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)\*(C'\fR). The \f(CW.01\fR
1806is added to work around small timing inconsistencies of some operating
1807systems.
1808.PP
1809\fIWatcher-Specific Functions and Data Members\fR
1810.IX Subsection "Watcher-Specific Functions and Data Members"
1811.IP "ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)" 4
1812.IX Item "ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)"
1813.PD 0
1814.IP "ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)" 4
1815.IX Item "ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)"
1816.PD
1817Configures the watcher to wait for status changes of the given
1818\&\f(CW\*(C`path\*(C'\fR. The \f(CW\*(C`interval\*(C'\fR is a hint on how quickly a change is expected to
1819be detected and should normally be specified as \f(CW0\fR to let libev choose
1820a suitable value. The memory pointed to by \f(CW\*(C`path\*(C'\fR must point to the same
1821path for as long as the watcher is active.
1822.Sp
1823The callback will be receive \f(CW\*(C`EV_STAT\*(C'\fR when a change was detected,
1824relative to the attributes at the time the watcher was started (or the
1825last change was detected).
1826.IP "ev_stat_stat (loop, ev_stat *)" 4
1827.IX Item "ev_stat_stat (loop, ev_stat *)"
1828Updates the stat buffer immediately with new values. If you change the
1829watched path in your callback, you could call this fucntion to avoid
1830detecting this change (while introducing a race condition). Can also be
1831useful simply to find out the new values.
1832.IP "ev_statdata attr [read\-only]" 4
1833.IX Item "ev_statdata attr [read-only]"
1834The most-recently detected attributes of the file. Although the type is of
1835\&\f(CW\*(C`ev_statdata\*(C'\fR, this is usually the (or one of the) \f(CW\*(C`struct stat\*(C'\fR types
1836suitable for your system. If the \f(CW\*(C`st_nlink\*(C'\fR member is \f(CW0\fR, then there
1837was some error while \f(CW\*(C`stat\*(C'\fRing the file.
1838.IP "ev_statdata prev [read\-only]" 4
1839.IX Item "ev_statdata prev [read-only]"
1840The previous attributes of the file. The callback gets invoked whenever
1841\&\f(CW\*(C`prev\*(C'\fR != \f(CW\*(C`attr\*(C'\fR.
1842.IP "ev_tstamp interval [read\-only]" 4
1843.IX Item "ev_tstamp interval [read-only]"
1844The specified interval.
1845.IP "const char *path [read\-only]" 4
1846.IX Item "const char *path [read-only]"
1847The filesystem path that is being watched.
1848.PP
1849\fIExamples\fR
1850.IX Subsection "Examples"
1851.PP
1852Example: Watch \f(CW\*(C`/etc/passwd\*(C'\fR for attribute changes.
1853.PP
1854.Vb 10
1855\& static void
1856\& passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1857\& {
1858\& /* /etc/passwd changed in some way */
1859\& if (w\->attr.st_nlink)
1860\& {
1861\& printf ("passwd current size %ld\en", (long)w\->attr.st_size);
1862\& printf ("passwd current atime %ld\en", (long)w\->attr.st_mtime);
1863\& printf ("passwd current mtime %ld\en", (long)w\->attr.st_mtime);
1864\& }
1865\& else
1866\& /* you shalt not abuse printf for puts */
1867\& puts ("wow, /etc/passwd is not there, expect problems. "
1868\& "if this is windows, they already arrived\en");
1869\& }
1870\&
1871\& ...
1872\& ev_stat passwd;
1873\&
1874\& ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1875\& ev_stat_start (loop, &passwd);
1876.Ve
1877.PP
1878Example: Like above, but additionally use a one-second delay so we do not
1879miss updates (however, frequent updates will delay processing, too, so
1880one might do the work both on \f(CW\*(C`ev_stat\*(C'\fR callback invocation \fIand\fR on
1881\&\f(CW\*(C`ev_timer\*(C'\fR callback invocation).
1882.PP
1110.Vb 3 1883.Vb 2
1111\& struct ev_signal signal_watcher; 1884\& static ev_stat passwd;
1112\& ev_signal_init (&signal_watcher, sigint_cb, SIGINT); 1885\& static ev_timer timer;
1113\& ev_signal_start (loop, &sigint_cb); 1886\&
1887\& static void
1888\& timer_cb (EV_P_ ev_timer *w, int revents)
1889\& {
1890\& ev_timer_stop (EV_A_ w);
1891\&
1892\& /* now it\*(Aqs one second after the most recent passwd change */
1893\& }
1894\&
1895\& static void
1896\& stat_cb (EV_P_ ev_stat *w, int revents)
1897\& {
1898\& /* reset the one\-second timer */
1899\& ev_timer_again (EV_A_ &timer);
1900\& }
1901\&
1902\& ...
1903\& ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
1904\& ev_stat_start (loop, &passwd);
1905\& ev_timer_init (&timer, timer_cb, 0., 1.01);
1114.Ve 1906.Ve
1115.ie n .Sh """ev_idle"" \- when you've got nothing better to do" 1907.ie n .Sh """ev_idle"" \- when you've got nothing better to do..."
1116.el .Sh "\f(CWev_idle\fP \- when you've got nothing better to do" 1908.el .Sh "\f(CWev_idle\fP \- when you've got nothing better to do..."
1117.IX Subsection "ev_idle - when you've got nothing better to do" 1909.IX Subsection "ev_idle - when you've got nothing better to do..."
1118Idle watchers trigger events when there are no other events are pending 1910Idle watchers trigger events when no other events of the same or higher
1119(prepare, check and other idle watchers do not count). That is, as long 1911priority are pending (prepare, check and other idle watchers do not
1120as your process is busy handling sockets or timeouts (or even signals, 1912count).
1121imagine) it will not be triggered. But when your process is idle all idle 1913.PP
1122watchers are being called again and again, once per event loop iteration \- 1914That is, as long as your process is busy handling sockets or timeouts
1915(or even signals, imagine) of the same or higher priority it will not be
1916triggered. But when your process is idle (or only lower-priority watchers
1917are pending), the idle watchers are being called once per event loop
1123until stopped, that is, or your process receives more events and becomes 1918iteration \- until stopped, that is, or your process receives more events
1124busy. 1919and becomes busy again with higher priority stuff.
1125.PP 1920.PP
1126The most noteworthy effect is that as long as any idle watchers are 1921The most noteworthy effect is that as long as any idle watchers are
1127active, the process will not block when waiting for new events. 1922active, the process will not block when waiting for new events.
1128.PP 1923.PP
1129Apart from keeping your process non-blocking (which is a useful 1924Apart from keeping your process non-blocking (which is a useful
1130effect on its own sometimes), idle watchers are a good place to do 1925effect on its own sometimes), idle watchers are a good place to do
1131\&\*(L"pseudo\-background processing\*(R", or delay processing stuff to after the 1926\&\*(L"pseudo-background processing\*(R", or delay processing stuff to after the
1132event loop has handled all outstanding events. 1927event loop has handled all outstanding events.
1928.PP
1929\fIWatcher-Specific Functions and Data Members\fR
1930.IX Subsection "Watcher-Specific Functions and Data Members"
1133.IP "ev_idle_init (ev_signal *, callback)" 4 1931.IP "ev_idle_init (ev_signal *, callback)" 4
1134.IX Item "ev_idle_init (ev_signal *, callback)" 1932.IX Item "ev_idle_init (ev_signal *, callback)"
1135Initialises and configures the idle watcher \- it has no parameters of any 1933Initialises and configures the idle watcher \- it has no parameters of any
1136kind. There is a \f(CW\*(C`ev_idle_set\*(C'\fR macro, but using it is utterly pointless, 1934kind. There is a \f(CW\*(C`ev_idle_set\*(C'\fR macro, but using it is utterly pointless,
1137believe me. 1935believe me.
1138.PP 1936.PP
1937\fIExamples\fR
1938.IX Subsection "Examples"
1939.PP
1139Example: dynamically allocate an \f(CW\*(C`ev_idle\*(C'\fR, start it, and in the 1940Example: Dynamically allocate an \f(CW\*(C`ev_idle\*(C'\fR watcher, start it, and in the
1140callback, free it. Alos, use no error checking, as usual. 1941callback, free it. Also, use no error checking, as usual.
1141.PP 1942.PP
1142.Vb 7 1943.Vb 7
1143\& static void 1944\& static void
1144\& idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents) 1945\& idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1145\& { 1946\& {
1146\& free (w); 1947\& free (w);
1147\& // now do something you wanted to do when the program has 1948\& // now do something you wanted to do when the program has
1148\& // no longer asnything immediate to do. 1949\& // no longer anything immediate to do.
1149\& } 1950\& }
1150.Ve 1951\&
1151.PP
1152.Vb 3
1153\& struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle)); 1952\& struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1154\& ev_idle_init (idle_watcher, idle_cb); 1953\& ev_idle_init (idle_watcher, idle_cb);
1155\& ev_idle_start (loop, idle_cb); 1954\& ev_idle_start (loop, idle_cb);
1156.Ve 1955.Ve
1157.ie n .Sh """ev_prepare""\fP and \f(CW""ev_check"" \- customise your event loop" 1956.ie n .Sh """ev_prepare""\fP and \f(CW""ev_check"" \- customise your event loop!"
1158.el .Sh "\f(CWev_prepare\fP and \f(CWev_check\fP \- customise your event loop" 1957.el .Sh "\f(CWev_prepare\fP and \f(CWev_check\fP \- customise your event loop!"
1159.IX Subsection "ev_prepare and ev_check - customise your event loop" 1958.IX Subsection "ev_prepare and ev_check - customise your event loop!"
1160Prepare and check watchers are usually (but not always) used in tandem: 1959Prepare and check watchers are usually (but not always) used in tandem:
1161prepare watchers get invoked before the process blocks and check watchers 1960prepare watchers get invoked before the process blocks and check watchers
1162afterwards. 1961afterwards.
1163.PP 1962.PP
1963You \fImust not\fR call \f(CW\*(C`ev_loop\*(C'\fR or similar functions that enter
1964the current event loop from either \f(CW\*(C`ev_prepare\*(C'\fR or \f(CW\*(C`ev_check\*(C'\fR
1965watchers. Other loops than the current one are fine, however. The
1966rationale behind this is that you do not need to check for recursion in
1967those watchers, i.e. the sequence will always be \f(CW\*(C`ev_prepare\*(C'\fR, blocking,
1968\&\f(CW\*(C`ev_check\*(C'\fR so if you have one watcher of each kind they will always be
1969called in pairs bracketing the blocking call.
1970.PP
1164Their main purpose is to integrate other event mechanisms into libev. This 1971Their main purpose is to integrate other event mechanisms into libev and
1165could be used, for example, to track variable changes, implement your own 1972their use is somewhat advanced. This could be used, for example, to track
1166watchers, integrate net-snmp or a coroutine library and lots more. 1973variable changes, implement your own watchers, integrate net-snmp or a
1974coroutine library and lots more. They are also occasionally useful if
1975you cache some data and want to flush it before blocking (for example,
1976in X programs you might want to do an \f(CW\*(C`XFlush ()\*(C'\fR in an \f(CW\*(C`ev_prepare\*(C'\fR
1977watcher).
1167.PP 1978.PP
1168This is done by examining in each prepare call which file descriptors need 1979This is done by examining in each prepare call which file descriptors need
1169to be watched by the other library, registering \f(CW\*(C`ev_io\*(C'\fR watchers for 1980to be watched by the other library, registering \f(CW\*(C`ev_io\*(C'\fR watchers for
1170them and starting an \f(CW\*(C`ev_timer\*(C'\fR watcher for any timeouts (many libraries 1981them and starting an \f(CW\*(C`ev_timer\*(C'\fR watcher for any timeouts (many libraries
1171provide just this functionality). Then, in the check watcher you check for 1982provide just this functionality). Then, in the check watcher you check for
… …
1180are ready to run (it's actually more complicated: it only runs coroutines 1991are ready to run (it's actually more complicated: it only runs coroutines
1181with priority higher than or equal to the event loop and one coroutine 1992with priority higher than or equal to the event loop and one coroutine
1182of lower priority, but only once, using idle watchers to keep the event 1993of lower priority, but only once, using idle watchers to keep the event
1183loop from blocking if lower-priority coroutines are active, thus mapping 1994loop from blocking if lower-priority coroutines are active, thus mapping
1184low-priority coroutines to idle/background tasks). 1995low-priority coroutines to idle/background tasks).
1996.PP
1997It is recommended to give \f(CW\*(C`ev_check\*(C'\fR watchers highest (\f(CW\*(C`EV_MAXPRI\*(C'\fR)
1998priority, to ensure that they are being run before any other watchers
1999after the poll. Also, \f(CW\*(C`ev_check\*(C'\fR watchers (and \f(CW\*(C`ev_prepare\*(C'\fR watchers,
2000too) should not activate (\*(L"feed\*(R") events into libev. While libev fully
2001supports this, they will be called before other \f(CW\*(C`ev_check\*(C'\fR watchers
2002did their job. As \f(CW\*(C`ev_check\*(C'\fR watchers are often used to embed other
2003(non-libev) event loops those other event loops might be in an unusable
2004state until their \f(CW\*(C`ev_check\*(C'\fR watcher ran (always remind yourself to
2005coexist peacefully with others).
2006.PP
2007\fIWatcher-Specific Functions and Data Members\fR
2008.IX Subsection "Watcher-Specific Functions and Data Members"
1185.IP "ev_prepare_init (ev_prepare *, callback)" 4 2009.IP "ev_prepare_init (ev_prepare *, callback)" 4
1186.IX Item "ev_prepare_init (ev_prepare *, callback)" 2010.IX Item "ev_prepare_init (ev_prepare *, callback)"
1187.PD 0 2011.PD 0
1188.IP "ev_check_init (ev_check *, callback)" 4 2012.IP "ev_check_init (ev_check *, callback)" 4
1189.IX Item "ev_check_init (ev_check *, callback)" 2013.IX Item "ev_check_init (ev_check *, callback)"
1190.PD 2014.PD
1191Initialises and configures the prepare or check watcher \- they have no 2015Initialises and configures the prepare or check watcher \- they have no
1192parameters of any kind. There are \f(CW\*(C`ev_prepare_set\*(C'\fR and \f(CW\*(C`ev_check_set\*(C'\fR 2016parameters of any kind. There are \f(CW\*(C`ev_prepare_set\*(C'\fR and \f(CW\*(C`ev_check_set\*(C'\fR
1193macros, but using them is utterly, utterly and completely pointless. 2017macros, but using them is utterly, utterly and completely pointless.
1194.PP 2018.PP
1195Example: *TODO*. 2019\fIExamples\fR
2020.IX Subsection "Examples"
2021.PP
2022There are a number of principal ways to embed other event loops or modules
2023into libev. Here are some ideas on how to include libadns into libev
2024(there is a Perl module named \f(CW\*(C`EV::ADNS\*(C'\fR that does this, which you could
2025use for an actually working example. Another Perl module named \f(CW\*(C`EV::Glib\*(C'\fR
2026embeds a Glib main context into libev, and finally, \f(CW\*(C`Glib::EV\*(C'\fR embeds \s-1EV\s0
2027into the Glib event loop).
2028.PP
2029Method 1: Add \s-1IO\s0 watchers and a timeout watcher in a prepare handler,
2030and in a check watcher, destroy them and call into libadns. What follows
2031is pseudo-code only of course. This requires you to either use a low
2032priority for the check watcher or use \f(CW\*(C`ev_clear_pending\*(C'\fR explicitly, as
2033the callbacks for the IO/timeout watchers might not have been called yet.
2034.PP
2035.Vb 2
2036\& static ev_io iow [nfd];
2037\& static ev_timer tw;
2038\&
2039\& static void
2040\& io_cb (ev_loop *loop, ev_io *w, int revents)
2041\& {
2042\& }
2043\&
2044\& // create io watchers for each fd and a timer before blocking
2045\& static void
2046\& adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
2047\& {
2048\& int timeout = 3600000;
2049\& struct pollfd fds [nfd];
2050\& // actual code will need to loop here and realloc etc.
2051\& adns_beforepoll (ads, fds, &nfd, &timeout, timeval_from (ev_time ()));
2052\&
2053\& /* the callback is illegal, but won\*(Aqt be called as we stop during check */
2054\& ev_timer_init (&tw, 0, timeout * 1e\-3);
2055\& ev_timer_start (loop, &tw);
2056\&
2057\& // create one ev_io per pollfd
2058\& for (int i = 0; i < nfd; ++i)
2059\& {
2060\& ev_io_init (iow + i, io_cb, fds [i].fd,
2061\& ((fds [i].events & POLLIN ? EV_READ : 0)
2062\& | (fds [i].events & POLLOUT ? EV_WRITE : 0)));
2063\&
2064\& fds [i].revents = 0;
2065\& ev_io_start (loop, iow + i);
2066\& }
2067\& }
2068\&
2069\& // stop all watchers after blocking
2070\& static void
2071\& adns_check_cb (ev_loop *loop, ev_check *w, int revents)
2072\& {
2073\& ev_timer_stop (loop, &tw);
2074\&
2075\& for (int i = 0; i < nfd; ++i)
2076\& {
2077\& // set the relevant poll flags
2078\& // could also call adns_processreadable etc. here
2079\& struct pollfd *fd = fds + i;
2080\& int revents = ev_clear_pending (iow + i);
2081\& if (revents & EV_READ ) fd\->revents |= fd\->events & POLLIN;
2082\& if (revents & EV_WRITE) fd\->revents |= fd\->events & POLLOUT;
2083\&
2084\& // now stop the watcher
2085\& ev_io_stop (loop, iow + i);
2086\& }
2087\&
2088\& adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
2089\& }
2090.Ve
2091.PP
2092Method 2: This would be just like method 1, but you run \f(CW\*(C`adns_afterpoll\*(C'\fR
2093in the prepare watcher and would dispose of the check watcher.
2094.PP
2095Method 3: If the module to be embedded supports explicit event
2096notification (adns does), you can also make use of the actual watcher
2097callbacks, and only destroy/create the watchers in the prepare watcher.
2098.PP
2099.Vb 5
2100\& static void
2101\& timer_cb (EV_P_ ev_timer *w, int revents)
2102\& {
2103\& adns_state ads = (adns_state)w\->data;
2104\& update_now (EV_A);
2105\&
2106\& adns_processtimeouts (ads, &tv_now);
2107\& }
2108\&
2109\& static void
2110\& io_cb (EV_P_ ev_io *w, int revents)
2111\& {
2112\& adns_state ads = (adns_state)w\->data;
2113\& update_now (EV_A);
2114\&
2115\& if (revents & EV_READ ) adns_processreadable (ads, w\->fd, &tv_now);
2116\& if (revents & EV_WRITE) adns_processwriteable (ads, w\->fd, &tv_now);
2117\& }
2118\&
2119\& // do not ever call adns_afterpoll
2120.Ve
2121.PP
2122Method 4: Do not use a prepare or check watcher because the module you
2123want to embed is too inflexible to support it. Instead, youc na override
2124their poll function. The drawback with this solution is that the main
2125loop is now no longer controllable by \s-1EV\s0. The \f(CW\*(C`Glib::EV\*(C'\fR module does
2126this.
2127.PP
2128.Vb 4
2129\& static gint
2130\& event_poll_func (GPollFD *fds, guint nfds, gint timeout)
2131\& {
2132\& int got_events = 0;
2133\&
2134\& for (n = 0; n < nfds; ++n)
2135\& // create/start io watcher that sets the relevant bits in fds[n] and increment got_events
2136\&
2137\& if (timeout >= 0)
2138\& // create/start timer
2139\&
2140\& // poll
2141\& ev_loop (EV_A_ 0);
2142\&
2143\& // stop timer again
2144\& if (timeout >= 0)
2145\& ev_timer_stop (EV_A_ &to);
2146\&
2147\& // stop io watchers again \- their callbacks should have set
2148\& for (n = 0; n < nfds; ++n)
2149\& ev_io_stop (EV_A_ iow [n]);
2150\&
2151\& return got_events;
2152\& }
2153.Ve
2154.ie n .Sh """ev_embed"" \- when one backend isn't enough..."
2155.el .Sh "\f(CWev_embed\fP \- when one backend isn't enough..."
2156.IX Subsection "ev_embed - when one backend isn't enough..."
2157This is a rather advanced watcher type that lets you embed one event loop
2158into another (currently only \f(CW\*(C`ev_io\*(C'\fR events are supported in the embedded
2159loop, other types of watchers might be handled in a delayed or incorrect
2160fashion and must not be used).
2161.PP
2162There are primarily two reasons you would want that: work around bugs and
2163prioritise I/O.
2164.PP
2165As an example for a bug workaround, the kqueue backend might only support
2166sockets on some platform, so it is unusable as generic backend, but you
2167still want to make use of it because you have many sockets and it scales
2168so nicely. In this case, you would create a kqueue-based loop and embed it
2169into your default loop (which might use e.g. poll). Overall operation will
2170be a bit slower because first libev has to poll and then call kevent, but
2171at least you can use both at what they are best.
2172.PP
2173As for prioritising I/O: rarely you have the case where some fds have
2174to be watched and handled very quickly (with low latency), and even
2175priorities and idle watchers might have too much overhead. In this case
2176you would put all the high priority stuff in one loop and all the rest in
2177a second one, and embed the second one in the first.
2178.PP
2179As long as the watcher is active, the callback will be invoked every time
2180there might be events pending in the embedded loop. The callback must then
2181call \f(CW\*(C`ev_embed_sweep (mainloop, watcher)\*(C'\fR to make a single sweep and invoke
2182their callbacks (you could also start an idle watcher to give the embedded
2183loop strictly lower priority for example). You can also set the callback
2184to \f(CW0\fR, in which case the embed watcher will automatically execute the
2185embedded loop sweep.
2186.PP
2187As long as the watcher is started it will automatically handle events. The
2188callback will be invoked whenever some events have been handled. You can
2189set the callback to \f(CW0\fR to avoid having to specify one if you are not
2190interested in that.
2191.PP
2192Also, there have not currently been made special provisions for forking:
2193when you fork, you not only have to call \f(CW\*(C`ev_loop_fork\*(C'\fR on both loops,
2194but you will also have to stop and restart any \f(CW\*(C`ev_embed\*(C'\fR watchers
2195yourself.
2196.PP
2197Unfortunately, not all backends are embeddable, only the ones returned by
2198\&\f(CW\*(C`ev_embeddable_backends\*(C'\fR are, which, unfortunately, does not include any
2199portable one.
2200.PP
2201So when you want to use this feature you will always have to be prepared
2202that you cannot get an embeddable loop. The recommended way to get around
2203this is to have a separate variables for your embeddable loop, try to
2204create it, and if that fails, use the normal loop for everything.
2205.PP
2206\fIWatcher-Specific Functions and Data Members\fR
2207.IX Subsection "Watcher-Specific Functions and Data Members"
2208.IP "ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)" 4
2209.IX Item "ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)"
2210.PD 0
2211.IP "ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)" 4
2212.IX Item "ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)"
2213.PD
2214Configures the watcher to embed the given loop, which must be
2215embeddable. If the callback is \f(CW0\fR, then \f(CW\*(C`ev_embed_sweep\*(C'\fR will be
2216invoked automatically, otherwise it is the responsibility of the callback
2217to invoke it (it will continue to be called until the sweep has been done,
2218if you do not want thta, you need to temporarily stop the embed watcher).
2219.IP "ev_embed_sweep (loop, ev_embed *)" 4
2220.IX Item "ev_embed_sweep (loop, ev_embed *)"
2221Make a single, non-blocking sweep over the embedded loop. This works
2222similarly to \f(CW\*(C`ev_loop (embedded_loop, EVLOOP_NONBLOCK)\*(C'\fR, but in the most
2223apropriate way for embedded loops.
2224.IP "struct ev_loop *other [read\-only]" 4
2225.IX Item "struct ev_loop *other [read-only]"
2226The embedded event loop.
2227.PP
2228\fIExamples\fR
2229.IX Subsection "Examples"
2230.PP
2231Example: Try to get an embeddable event loop and embed it into the default
2232event loop. If that is not possible, use the default loop. The default
2233loop is stored in \f(CW\*(C`loop_hi\*(C'\fR, while the mebeddable loop is stored in
2234\&\f(CW\*(C`loop_lo\*(C'\fR (which is \f(CW\*(C`loop_hi\*(C'\fR in the acse no embeddable loop can be
2235used).
2236.PP
2237.Vb 3
2238\& struct ev_loop *loop_hi = ev_default_init (0);
2239\& struct ev_loop *loop_lo = 0;
2240\& struct ev_embed embed;
2241\&
2242\& // see if there is a chance of getting one that works
2243\& // (remember that a flags value of 0 means autodetection)
2244\& loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
2245\& ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
2246\& : 0;
2247\&
2248\& // if we got one, then embed it, otherwise default to loop_hi
2249\& if (loop_lo)
2250\& {
2251\& ev_embed_init (&embed, 0, loop_lo);
2252\& ev_embed_start (loop_hi, &embed);
2253\& }
2254\& else
2255\& loop_lo = loop_hi;
2256.Ve
2257.PP
2258Example: Check if kqueue is available but not recommended and create
2259a kqueue backend for use with sockets (which usually work with any
2260kqueue implementation). Store the kqueue/socket\-only event loop in
2261\&\f(CW\*(C`loop_socket\*(C'\fR. (One might optionally use \f(CW\*(C`EVFLAG_NOENV\*(C'\fR, too).
2262.PP
2263.Vb 3
2264\& struct ev_loop *loop = ev_default_init (0);
2265\& struct ev_loop *loop_socket = 0;
2266\& struct ev_embed embed;
2267\&
2268\& if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
2269\& if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
2270\& {
2271\& ev_embed_init (&embed, 0, loop_socket);
2272\& ev_embed_start (loop, &embed);
2273\& }
2274\&
2275\& if (!loop_socket)
2276\& loop_socket = loop;
2277\&
2278\& // now use loop_socket for all sockets, and loop for everything else
2279.Ve
2280.ie n .Sh """ev_fork"" \- the audacity to resume the event loop after a fork"
2281.el .Sh "\f(CWev_fork\fP \- the audacity to resume the event loop after a fork"
2282.IX Subsection "ev_fork - the audacity to resume the event loop after a fork"
2283Fork watchers are called when a \f(CW\*(C`fork ()\*(C'\fR was detected (usually because
2284whoever is a good citizen cared to tell libev about it by calling
2285\&\f(CW\*(C`ev_default_fork\*(C'\fR or \f(CW\*(C`ev_loop_fork\*(C'\fR). The invocation is done before the
2286event loop blocks next and before \f(CW\*(C`ev_check\*(C'\fR watchers are being called,
2287and only in the child after the fork. If whoever good citizen calling
2288\&\f(CW\*(C`ev_default_fork\*(C'\fR cheats and calls it in the wrong process, the fork
2289handlers will be invoked, too, of course.
2290.PP
2291\fIWatcher-Specific Functions and Data Members\fR
2292.IX Subsection "Watcher-Specific Functions and Data Members"
2293.IP "ev_fork_init (ev_signal *, callback)" 4
2294.IX Item "ev_fork_init (ev_signal *, callback)"
2295Initialises and configures the fork watcher \- it has no parameters of any
2296kind. There is a \f(CW\*(C`ev_fork_set\*(C'\fR macro, but using it is utterly pointless,
2297believe me.
2298.ie n .Sh """ev_async"" \- how to wake up another event loop"
2299.el .Sh "\f(CWev_async\fP \- how to wake up another event loop"
2300.IX Subsection "ev_async - how to wake up another event loop"
2301In general, you cannot use an \f(CW\*(C`ev_loop\*(C'\fR from multiple threads or other
2302asynchronous sources such as signal handlers (as opposed to multiple event
2303loops \- those are of course safe to use in different threads).
2304.PP
2305Sometimes, however, you need to wake up another event loop you do not
2306control, for example because it belongs to another thread. This is what
2307\&\f(CW\*(C`ev_async\*(C'\fR watchers do: as long as the \f(CW\*(C`ev_async\*(C'\fR watcher is active, you
2308can signal it by calling \f(CW\*(C`ev_async_send\*(C'\fR, which is thread\- and signal
2309safe.
2310.PP
2311This functionality is very similar to \f(CW\*(C`ev_signal\*(C'\fR watchers, as signals,
2312too, are asynchronous in nature, and signals, too, will be compressed
2313(i.e. the number of callback invocations may be less than the number of
2314\&\f(CW\*(C`ev_async_sent\*(C'\fR calls).
2315.PP
2316Unlike \f(CW\*(C`ev_signal\*(C'\fR watchers, \f(CW\*(C`ev_async\*(C'\fR works with any event loop, not
2317just the default loop.
2318.PP
2319\fIQueueing\fR
2320.IX Subsection "Queueing"
2321.PP
2322\&\f(CW\*(C`ev_async\*(C'\fR does not support queueing of data in any way. The reason
2323is that the author does not know of a simple (or any) algorithm for a
2324multiple-writer-single-reader queue that works in all cases and doesn't
2325need elaborate support such as pthreads.
2326.PP
2327That means that if you want to queue data, you have to provide your own
2328queue. But at least I can tell you would implement locking around your
2329queue:
2330.IP "queueing from a signal handler context" 4
2331.IX Item "queueing from a signal handler context"
2332To implement race-free queueing, you simply add to the queue in the signal
2333handler but you block the signal handler in the watcher callback. Here is an example that does that for
2334some fictitiuous \s-1SIGUSR1\s0 handler:
2335.Sp
2336.Vb 1
2337\& static ev_async mysig;
2338\&
2339\& static void
2340\& sigusr1_handler (void)
2341\& {
2342\& sometype data;
2343\&
2344\& // no locking etc.
2345\& queue_put (data);
2346\& ev_async_send (EV_DEFAULT_ &mysig);
2347\& }
2348\&
2349\& static void
2350\& mysig_cb (EV_P_ ev_async *w, int revents)
2351\& {
2352\& sometype data;
2353\& sigset_t block, prev;
2354\&
2355\& sigemptyset (&block);
2356\& sigaddset (&block, SIGUSR1);
2357\& sigprocmask (SIG_BLOCK, &block, &prev);
2358\&
2359\& while (queue_get (&data))
2360\& process (data);
2361\&
2362\& if (sigismember (&prev, SIGUSR1)
2363\& sigprocmask (SIG_UNBLOCK, &block, 0);
2364\& }
2365.Ve
2366.Sp
2367(Note: pthreads in theory requires you to use \f(CW\*(C`pthread_setmask\*(C'\fR
2368instead of \f(CW\*(C`sigprocmask\*(C'\fR when you use threads, but libev doesn't do it
2369either...).
2370.IP "queueing from a thread context" 4
2371.IX Item "queueing from a thread context"
2372The strategy for threads is different, as you cannot (easily) block
2373threads but you can easily preempt them, so to queue safely you need to
2374employ a traditional mutex lock, such as in this pthread example:
2375.Sp
2376.Vb 2
2377\& static ev_async mysig;
2378\& static pthread_mutex_t mymutex = PTHREAD_MUTEX_INITIALIZER;
2379\&
2380\& static void
2381\& otherthread (void)
2382\& {
2383\& // only need to lock the actual queueing operation
2384\& pthread_mutex_lock (&mymutex);
2385\& queue_put (data);
2386\& pthread_mutex_unlock (&mymutex);
2387\&
2388\& ev_async_send (EV_DEFAULT_ &mysig);
2389\& }
2390\&
2391\& static void
2392\& mysig_cb (EV_P_ ev_async *w, int revents)
2393\& {
2394\& pthread_mutex_lock (&mymutex);
2395\&
2396\& while (queue_get (&data))
2397\& process (data);
2398\&
2399\& pthread_mutex_unlock (&mymutex);
2400\& }
2401.Ve
2402.PP
2403\fIWatcher-Specific Functions and Data Members\fR
2404.IX Subsection "Watcher-Specific Functions and Data Members"
2405.IP "ev_async_init (ev_async *, callback)" 4
2406.IX Item "ev_async_init (ev_async *, callback)"
2407Initialises and configures the async watcher \- it has no parameters of any
2408kind. There is a \f(CW\*(C`ev_asynd_set\*(C'\fR macro, but using it is utterly pointless,
2409believe me.
2410.IP "ev_async_send (loop, ev_async *)" 4
2411.IX Item "ev_async_send (loop, ev_async *)"
2412Sends/signals/activates the given \f(CW\*(C`ev_async\*(C'\fR watcher, that is, feeds
2413an \f(CW\*(C`EV_ASYNC\*(C'\fR event on the watcher into the event loop. Unlike
2414\&\f(CW\*(C`ev_feed_event\*(C'\fR, this call is safe to do in other threads, signal or
2415similar contexts (see the dicusssion of \f(CW\*(C`EV_ATOMIC_T\*(C'\fR in the embedding
2416section below on what exactly this means).
2417.Sp
2418This call incurs the overhead of a syscall only once per loop iteration,
2419so while the overhead might be noticable, it doesn't apply to repeated
2420calls to \f(CW\*(C`ev_async_send\*(C'\fR.
2421.IP "bool = ev_async_pending (ev_async *)" 4
2422.IX Item "bool = ev_async_pending (ev_async *)"
2423Returns a non-zero value when \f(CW\*(C`ev_async_send\*(C'\fR has been called on the
2424watcher but the event has not yet been processed (or even noted) by the
2425event loop.
2426.Sp
2427\&\f(CW\*(C`ev_async_send\*(C'\fR sets a flag in the watcher and wakes up the loop. When
2428the loop iterates next and checks for the watcher to have become active,
2429it will reset the flag again. \f(CW\*(C`ev_async_pending\*(C'\fR can be used to very
2430quickly check wether invoking the loop might be a good idea.
2431.Sp
2432Not that this does \fInot\fR check wether the watcher itself is pending, only
2433wether it has been requested to make this watcher pending.
1196.SH "OTHER FUNCTIONS" 2434.SH "OTHER FUNCTIONS"
1197.IX Header "OTHER FUNCTIONS" 2435.IX Header "OTHER FUNCTIONS"
1198There are some other functions of possible interest. Described. Here. Now. 2436There are some other functions of possible interest. Described. Here. Now.
1199.IP "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 4 2437.IP "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 4
1200.IX Item "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)" 2438.IX Item "ev_once (loop, int fd, int events, ev_tstamp timeout, callback)"
… …
1224\& if (revents & EV_TIMEOUT) 2462\& if (revents & EV_TIMEOUT)
1225\& /* doh, nothing entered */; 2463\& /* doh, nothing entered */;
1226\& else if (revents & EV_READ) 2464\& else if (revents & EV_READ)
1227\& /* stdin might have data for us, joy! */; 2465\& /* stdin might have data for us, joy! */;
1228\& } 2466\& }
1229.Ve 2467\&
1230.Sp
1231.Vb 1
1232\& ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0); 2468\& ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1233.Ve 2469.Ve
1234.IP "ev_feed_event (loop, watcher, int events)" 4 2470.IP "ev_feed_event (ev_loop *, watcher *, int revents)" 4
1235.IX Item "ev_feed_event (loop, watcher, int events)" 2471.IX Item "ev_feed_event (ev_loop *, watcher *, int revents)"
1236Feeds the given event set into the event loop, as if the specified event 2472Feeds the given event set into the event loop, as if the specified event
1237had happened for the specified watcher (which must be a pointer to an 2473had happened for the specified watcher (which must be a pointer to an
1238initialised but not necessarily started event watcher). 2474initialised but not necessarily started event watcher).
1239.IP "ev_feed_fd_event (loop, int fd, int revents)" 4 2475.IP "ev_feed_fd_event (ev_loop *, int fd, int revents)" 4
1240.IX Item "ev_feed_fd_event (loop, int fd, int revents)" 2476.IX Item "ev_feed_fd_event (ev_loop *, int fd, int revents)"
1241Feed an event on the given fd, as if a file descriptor backend detected 2477Feed an event on the given fd, as if a file descriptor backend detected
1242the given events it. 2478the given events it.
1243.IP "ev_feed_signal_event (loop, int signum)" 4 2479.IP "ev_feed_signal_event (ev_loop *loop, int signum)" 4
1244.IX Item "ev_feed_signal_event (loop, int signum)" 2480.IX Item "ev_feed_signal_event (ev_loop *loop, int signum)"
1245Feed an event as if the given signal occured (loop must be the default loop!). 2481Feed an event as if the given signal occured (\f(CW\*(C`loop\*(C'\fR must be the default
2482loop!).
1246.SH "LIBEVENT EMULATION" 2483.SH "LIBEVENT EMULATION"
1247.IX Header "LIBEVENT EMULATION" 2484.IX Header "LIBEVENT EMULATION"
1248Libev offers a compatibility emulation layer for libevent. It cannot 2485Libev offers a compatibility emulation layer for libevent. It cannot
1249emulate the internals of libevent, so here are some usage hints: 2486emulate the internals of libevent, so here are some usage hints:
2487.IP "\(bu" 4
1250.IP "* Use it by including <event.h>, as usual." 4 2488Use it by including <event.h>, as usual.
1251.IX Item "Use it by including <event.h>, as usual." 2489.IP "\(bu" 4
1252.PD 0 2490The following members are fully supported: ev_base, ev_callback,
1253.IP "* The following members are fully supported: ev_base, ev_callback, ev_arg, ev_fd, ev_res, ev_events." 4 2491ev_arg, ev_fd, ev_res, ev_events.
1254.IX Item "The following members are fully supported: ev_base, ev_callback, ev_arg, ev_fd, ev_res, ev_events." 2492.IP "\(bu" 4
1255.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 2493Avoid using ev_flags and the EVLIST_*\-macros, while it is
1256.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)." 2494maintained by libev, it does not work exactly the same way as in libevent (consider
1257.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 2495it a private \s-1API\s0).
1258.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." 2496.IP "\(bu" 4
2497Priorities are not currently supported. Initialising priorities
2498will fail and all watchers will have the same priority, even though there
2499is an ev_pri field.
2500.IP "\(bu" 4
1259.IP "* Other members are not supported." 4 2501Other members are not supported.
1260.IX Item "Other members are not supported." 2502.IP "\(bu" 4
1261.IP "* The libev emulation is \fInot\fR \s-1ABI\s0 compatible to libevent, you need to use the libev header file and library." 4 2503The libev emulation is \fInot\fR \s-1ABI\s0 compatible to libevent, you need
1262.IX Item "The libev emulation is not ABI compatible to libevent, you need to use the libev header file and library." 2504to use the libev header file and library.
1263.PD
1264.SH "\*(C+ SUPPORT" 2505.SH "\*(C+ SUPPORT"
1265.IX Header " SUPPORT" 2506.IX Header " SUPPORT"
1266\&\s-1TBD\s0. 2507Libev comes with some simplistic wrapper classes for \*(C+ that mainly allow
2508you to use some convinience methods to start/stop watchers and also change
2509the callback model to a model using method callbacks on objects.
2510.PP
2511To use it,
2512.PP
2513.Vb 1
2514\& #include <ev++.h>
2515.Ve
2516.PP
2517This automatically includes \fIev.h\fR and puts all of its definitions (many
2518of them macros) into the global namespace. All \*(C+ specific things are
2519put into the \f(CW\*(C`ev\*(C'\fR namespace. It should support all the same embedding
2520options as \fIev.h\fR, most notably \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR.
2521.PP
2522Care has been taken to keep the overhead low. The only data member the \*(C+
2523classes add (compared to plain C\-style watchers) is the event loop pointer
2524that the watcher is associated with (or no additional members at all if
2525you disable \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR when embedding libev).
2526.PP
2527Currently, functions, and static and non-static member functions can be
2528used as callbacks. Other types should be easy to add as long as they only
2529need one additional pointer for context. If you need support for other
2530types of functors please contact the author (preferably after implementing
2531it).
2532.PP
2533Here is a list of things available in the \f(CW\*(C`ev\*(C'\fR namespace:
2534.ie n .IP """ev::READ""\fR, \f(CW""ev::WRITE"" etc." 4
2535.el .IP "\f(CWev::READ\fR, \f(CWev::WRITE\fR etc." 4
2536.IX Item "ev::READ, ev::WRITE etc."
2537These are just enum values with the same values as the \f(CW\*(C`EV_READ\*(C'\fR etc.
2538macros from \fIev.h\fR.
2539.ie n .IP """ev::tstamp""\fR, \f(CW""ev::now""" 4
2540.el .IP "\f(CWev::tstamp\fR, \f(CWev::now\fR" 4
2541.IX Item "ev::tstamp, ev::now"
2542Aliases to the same types/functions as with the \f(CW\*(C`ev_\*(C'\fR prefix.
2543.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
2544.el .IP "\f(CWev::io\fR, \f(CWev::timer\fR, \f(CWev::periodic\fR, \f(CWev::idle\fR, \f(CWev::sig\fR etc." 4
2545.IX Item "ev::io, ev::timer, ev::periodic, ev::idle, ev::sig etc."
2546For each \f(CW\*(C`ev_TYPE\*(C'\fR watcher in \fIev.h\fR there is a corresponding class of
2547the same name in the \f(CW\*(C`ev\*(C'\fR namespace, with the exception of \f(CW\*(C`ev_signal\*(C'\fR
2548which is called \f(CW\*(C`ev::sig\*(C'\fR to avoid clashes with the \f(CW\*(C`signal\*(C'\fR macro
2549defines by many implementations.
2550.Sp
2551All of those classes have these methods:
2552.RS 4
2553.IP "ev::TYPE::TYPE ()" 4
2554.IX Item "ev::TYPE::TYPE ()"
2555.PD 0
2556.IP "ev::TYPE::TYPE (struct ev_loop *)" 4
2557.IX Item "ev::TYPE::TYPE (struct ev_loop *)"
2558.IP "ev::TYPE::~TYPE" 4
2559.IX Item "ev::TYPE::~TYPE"
2560.PD
2561The constructor (optionally) takes an event loop to associate the watcher
2562with. If it is omitted, it will use \f(CW\*(C`EV_DEFAULT\*(C'\fR.
2563.Sp
2564The constructor calls \f(CW\*(C`ev_init\*(C'\fR for you, which means you have to call the
2565\&\f(CW\*(C`set\*(C'\fR method before starting it.
2566.Sp
2567It will not set a callback, however: You have to call the templated \f(CW\*(C`set\*(C'\fR
2568method to set a callback before you can start the watcher.
2569.Sp
2570(The reason why you have to use a method is a limitation in \*(C+ which does
2571not allow explicit template arguments for constructors).
2572.Sp
2573The destructor automatically stops the watcher if it is active.
2574.IP "w\->set<class, &class::method> (object *)" 4
2575.IX Item "w->set<class, &class::method> (object *)"
2576This method sets the callback method to call. The method has to have a
2577signature of \f(CW\*(C`void (*)(ev_TYPE &, int)\*(C'\fR, it receives the watcher as
2578first argument and the \f(CW\*(C`revents\*(C'\fR as second. The object must be given as
2579parameter and is stored in the \f(CW\*(C`data\*(C'\fR member of the watcher.
2580.Sp
2581This method synthesizes efficient thunking code to call your method from
2582the C callback that libev requires. If your compiler can inline your
2583callback (i.e. it is visible to it at the place of the \f(CW\*(C`set\*(C'\fR call and
2584your compiler is good :), then the method will be fully inlined into the
2585thunking function, making it as fast as a direct C callback.
2586.Sp
2587Example: simple class declaration and watcher initialisation
2588.Sp
2589.Vb 4
2590\& struct myclass
2591\& {
2592\& void io_cb (ev::io &w, int revents) { }
2593\& }
2594\&
2595\& myclass obj;
2596\& ev::io iow;
2597\& iow.set <myclass, &myclass::io_cb> (&obj);
2598.Ve
2599.IP "w\->set<function> (void *data = 0)" 4
2600.IX Item "w->set<function> (void *data = 0)"
2601Also sets a callback, but uses a static method or plain function as
2602callback. The optional \f(CW\*(C`data\*(C'\fR argument will be stored in the watcher's
2603\&\f(CW\*(C`data\*(C'\fR member and is free for you to use.
2604.Sp
2605The prototype of the \f(CW\*(C`function\*(C'\fR must be \f(CW\*(C`void (*)(ev::TYPE &w, int)\*(C'\fR.
2606.Sp
2607See the method\-\f(CW\*(C`set\*(C'\fR above for more details.
2608.Sp
2609Example:
2610.Sp
2611.Vb 2
2612\& static void io_cb (ev::io &w, int revents) { }
2613\& iow.set <io_cb> ();
2614.Ve
2615.IP "w\->set (struct ev_loop *)" 4
2616.IX Item "w->set (struct ev_loop *)"
2617Associates a different \f(CW\*(C`struct ev_loop\*(C'\fR with this watcher. You can only
2618do this when the watcher is inactive (and not pending either).
2619.IP "w\->set ([args])" 4
2620.IX Item "w->set ([args])"
2621Basically the same as \f(CW\*(C`ev_TYPE_set\*(C'\fR, with the same args. Must be
2622called at least once. Unlike the C counterpart, an active watcher gets
2623automatically stopped and restarted when reconfiguring it with this
2624method.
2625.IP "w\->start ()" 4
2626.IX Item "w->start ()"
2627Starts the watcher. Note that there is no \f(CW\*(C`loop\*(C'\fR argument, as the
2628constructor already stores the event loop.
2629.IP "w\->stop ()" 4
2630.IX Item "w->stop ()"
2631Stops the watcher if it is active. Again, no \f(CW\*(C`loop\*(C'\fR argument.
2632.ie n .IP "w\->again () (""ev::timer""\fR, \f(CW""ev::periodic"" only)" 4
2633.el .IP "w\->again () (\f(CWev::timer\fR, \f(CWev::periodic\fR only)" 4
2634.IX Item "w->again () (ev::timer, ev::periodic only)"
2635For \f(CW\*(C`ev::timer\*(C'\fR and \f(CW\*(C`ev::periodic\*(C'\fR, this invokes the corresponding
2636\&\f(CW\*(C`ev_TYPE_again\*(C'\fR function.
2637.ie n .IP "w\->sweep () (""ev::embed"" only)" 4
2638.el .IP "w\->sweep () (\f(CWev::embed\fR only)" 4
2639.IX Item "w->sweep () (ev::embed only)"
2640Invokes \f(CW\*(C`ev_embed_sweep\*(C'\fR.
2641.ie n .IP "w\->update () (""ev::stat"" only)" 4
2642.el .IP "w\->update () (\f(CWev::stat\fR only)" 4
2643.IX Item "w->update () (ev::stat only)"
2644Invokes \f(CW\*(C`ev_stat_stat\*(C'\fR.
2645.RE
2646.RS 4
2647.RE
2648.PP
2649Example: Define a class with an \s-1IO\s0 and idle watcher, start one of them in
2650the constructor.
2651.PP
2652.Vb 4
2653\& class myclass
2654\& {
2655\& ev::io io; void io_cb (ev::io &w, int revents);
2656\& ev:idle idle void idle_cb (ev::idle &w, int revents);
2657\&
2658\& myclass (int fd)
2659\& {
2660\& io .set <myclass, &myclass::io_cb > (this);
2661\& idle.set <myclass, &myclass::idle_cb> (this);
2662\&
2663\& io.start (fd, ev::READ);
2664\& }
2665\& };
2666.Ve
2667.SH "OTHER LANGUAGE BINDINGS"
2668.IX Header "OTHER LANGUAGE BINDINGS"
2669Libev does not offer other language bindings itself, but bindings for a
2670numbe rof languages exist in the form of third-party packages. If you know
2671any interesting language binding in addition to the ones listed here, drop
2672me a note.
2673.IP "Perl" 4
2674.IX Item "Perl"
2675The \s-1EV\s0 module implements the full libev \s-1API\s0 and is actually used to test
2676libev. \s-1EV\s0 is developed together with libev. Apart from the \s-1EV\s0 core module,
2677there are additional modules that implement libev-compatible interfaces
2678to \f(CW\*(C`libadns\*(C'\fR (\f(CW\*(C`EV::ADNS\*(C'\fR), \f(CW\*(C`Net::SNMP\*(C'\fR (\f(CW\*(C`Net::SNMP::EV\*(C'\fR) and the
2679\&\f(CW\*(C`libglib\*(C'\fR event core (\f(CW\*(C`Glib::EV\*(C'\fR and \f(CW\*(C`EV::Glib\*(C'\fR).
2680.Sp
2681It can be found and installed via \s-1CPAN\s0, its homepage is found at
2682<http://software.schmorp.de/pkg/EV>.
2683.IP "Ruby" 4
2684.IX Item "Ruby"
2685Tony Arcieri has written a ruby extension that offers access to a subset
2686of the libev \s-1API\s0 and adds filehandle abstractions, asynchronous \s-1DNS\s0 and
2687more on top of it. It can be found via gem servers. Its homepage is at
2688<http://rev.rubyforge.org/>.
2689.IP "D" 4
2690.IX Item "D"
2691Leandro Lucarella has written a D language binding (\fIev.d\fR) for libev, to
2692be found at <http://git.llucax.com.ar/?p=software/ev.d.git;a=summary>.
2693.SH "MACRO MAGIC"
2694.IX Header "MACRO MAGIC"
2695Libev can be compiled with a variety of options, the most fundamantal
2696of which is \f(CW\*(C`EV_MULTIPLICITY\*(C'\fR. This option determines whether (most)
2697functions and callbacks have an initial \f(CW\*(C`struct ev_loop *\*(C'\fR argument.
2698.PP
2699To make it easier to write programs that cope with either variant, the
2700following macros are defined:
2701.ie n .IP """EV_A""\fR, \f(CW""EV_A_""" 4
2702.el .IP "\f(CWEV_A\fR, \f(CWEV_A_\fR" 4
2703.IX Item "EV_A, EV_A_"
2704This provides the loop \fIargument\fR for functions, if one is required (\*(L"ev
2705loop argument\*(R"). The \f(CW\*(C`EV_A\*(C'\fR form is used when this is the sole argument,
2706\&\f(CW\*(C`EV_A_\*(C'\fR is used when other arguments are following. Example:
2707.Sp
2708.Vb 3
2709\& ev_unref (EV_A);
2710\& ev_timer_add (EV_A_ watcher);
2711\& ev_loop (EV_A_ 0);
2712.Ve
2713.Sp
2714It assumes the variable \f(CW\*(C`loop\*(C'\fR of type \f(CW\*(C`struct ev_loop *\*(C'\fR is in scope,
2715which is often provided by the following macro.
2716.ie n .IP """EV_P""\fR, \f(CW""EV_P_""" 4
2717.el .IP "\f(CWEV_P\fR, \f(CWEV_P_\fR" 4
2718.IX Item "EV_P, EV_P_"
2719This provides the loop \fIparameter\fR for functions, if one is required (\*(L"ev
2720loop parameter\*(R"). The \f(CW\*(C`EV_P\*(C'\fR form is used when this is the sole parameter,
2721\&\f(CW\*(C`EV_P_\*(C'\fR is used when other parameters are following. Example:
2722.Sp
2723.Vb 2
2724\& // this is how ev_unref is being declared
2725\& static void ev_unref (EV_P);
2726\&
2727\& // this is how you can declare your typical callback
2728\& static void cb (EV_P_ ev_timer *w, int revents)
2729.Ve
2730.Sp
2731It declares a parameter \f(CW\*(C`loop\*(C'\fR of type \f(CW\*(C`struct ev_loop *\*(C'\fR, quite
2732suitable for use with \f(CW\*(C`EV_A\*(C'\fR.
2733.ie n .IP """EV_DEFAULT""\fR, \f(CW""EV_DEFAULT_""" 4
2734.el .IP "\f(CWEV_DEFAULT\fR, \f(CWEV_DEFAULT_\fR" 4
2735.IX Item "EV_DEFAULT, EV_DEFAULT_"
2736Similar to the other two macros, this gives you the value of the default
2737loop, if multiple loops are supported (\*(L"ev loop default\*(R").
2738.PP
2739Example: Declare and initialise a check watcher, utilising the above
2740macros so it will work regardless of whether multiple loops are supported
2741or not.
2742.PP
2743.Vb 5
2744\& static void
2745\& check_cb (EV_P_ ev_timer *w, int revents)
2746\& {
2747\& ev_check_stop (EV_A_ w);
2748\& }
2749\&
2750\& ev_check check;
2751\& ev_check_init (&check, check_cb);
2752\& ev_check_start (EV_DEFAULT_ &check);
2753\& ev_loop (EV_DEFAULT_ 0);
2754.Ve
2755.SH "EMBEDDING"
2756.IX Header "EMBEDDING"
2757Libev can (and often is) directly embedded into host
2758applications. Examples of applications that embed it include the Deliantra
2759Game Server, the \s-1EV\s0 perl module, the \s-1GNU\s0 Virtual Private Ethernet (gvpe)
2760and rxvt-unicode.
2761.PP
2762The goal is to enable you to just copy the necessary files into your
2763source directory without having to change even a single line in them, so
2764you can easily upgrade by simply copying (or having a checked-out copy of
2765libev somewhere in your source tree).
2766.Sh "\s-1FILESETS\s0"
2767.IX Subsection "FILESETS"
2768Depending on what features you need you need to include one or more sets of files
2769in your app.
2770.PP
2771\fI\s-1CORE\s0 \s-1EVENT\s0 \s-1LOOP\s0\fR
2772.IX Subsection "CORE EVENT LOOP"
2773.PP
2774To include only the libev core (all the \f(CW\*(C`ev_*\*(C'\fR functions), with manual
2775configuration (no autoconf):
2776.PP
2777.Vb 2
2778\& #define EV_STANDALONE 1
2779\& #include "ev.c"
2780.Ve
2781.PP
2782This will automatically include \fIev.h\fR, too, and should be done in a
2783single C source file only to provide the function implementations. To use
2784it, do the same for \fIev.h\fR in all files wishing to use this \s-1API\s0 (best
2785done by writing a wrapper around \fIev.h\fR that you can include instead and
2786where you can put other configuration options):
2787.PP
2788.Vb 2
2789\& #define EV_STANDALONE 1
2790\& #include "ev.h"
2791.Ve
2792.PP
2793Both header files and implementation files can be compiled with a \*(C+
2794compiler (at least, thats a stated goal, and breakage will be treated
2795as a bug).
2796.PP
2797You need the following files in your source tree, or in a directory
2798in your include path (e.g. in libev/ when using \-Ilibev):
2799.PP
2800.Vb 4
2801\& ev.h
2802\& ev.c
2803\& ev_vars.h
2804\& ev_wrap.h
2805\&
2806\& ev_win32.c required on win32 platforms only
2807\&
2808\& ev_select.c only when select backend is enabled (which is enabled by default)
2809\& ev_poll.c only when poll backend is enabled (disabled by default)
2810\& ev_epoll.c only when the epoll backend is enabled (disabled by default)
2811\& ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2812\& ev_port.c only when the solaris port backend is enabled (disabled by default)
2813.Ve
2814.PP
2815\&\fIev.c\fR includes the backend files directly when enabled, so you only need
2816to compile this single file.
2817.PP
2818\fI\s-1LIBEVENT\s0 \s-1COMPATIBILITY\s0 \s-1API\s0\fR
2819.IX Subsection "LIBEVENT COMPATIBILITY API"
2820.PP
2821To include the libevent compatibility \s-1API\s0, also include:
2822.PP
2823.Vb 1
2824\& #include "event.c"
2825.Ve
2826.PP
2827in the file including \fIev.c\fR, and:
2828.PP
2829.Vb 1
2830\& #include "event.h"
2831.Ve
2832.PP
2833in the files that want to use the libevent \s-1API\s0. This also includes \fIev.h\fR.
2834.PP
2835You need the following additional files for this:
2836.PP
2837.Vb 2
2838\& event.h
2839\& event.c
2840.Ve
2841.PP
2842\fI\s-1AUTOCONF\s0 \s-1SUPPORT\s0\fR
2843.IX Subsection "AUTOCONF SUPPORT"
2844.PP
2845Instead of using \f(CW\*(C`EV_STANDALONE=1\*(C'\fR and providing your config in
2846whatever way you want, you can also \f(CW\*(C`m4_include([libev.m4])\*(C'\fR in your
2847\&\fIconfigure.ac\fR and leave \f(CW\*(C`EV_STANDALONE\*(C'\fR undefined. \fIev.c\fR will then
2848include \fIconfig.h\fR and configure itself accordingly.
2849.PP
2850For this of course you need the m4 file:
2851.PP
2852.Vb 1
2853\& libev.m4
2854.Ve
2855.Sh "\s-1PREPROCESSOR\s0 \s-1SYMBOLS/MACROS\s0"
2856.IX Subsection "PREPROCESSOR SYMBOLS/MACROS"
2857Libev can be configured via a variety of preprocessor symbols you have to define
2858before including any of its files. The default is not to build for multiplicity
2859and only include the select backend.
2860.IP "\s-1EV_STANDALONE\s0" 4
2861.IX Item "EV_STANDALONE"
2862Must always be \f(CW1\fR if you do not use autoconf configuration, which
2863keeps libev from including \fIconfig.h\fR, and it also defines dummy
2864implementations for some libevent functions (such as logging, which is not
2865supported). It will also not define any of the structs usually found in
2866\&\fIevent.h\fR that are not directly supported by the libev core alone.
2867.IP "\s-1EV_USE_MONOTONIC\s0" 4
2868.IX Item "EV_USE_MONOTONIC"
2869If defined to be \f(CW1\fR, libev will try to detect the availability of the
2870monotonic clock option at both compiletime and runtime. Otherwise no use
2871of the monotonic clock option will be attempted. If you enable this, you
2872usually have to link against librt or something similar. Enabling it when
2873the functionality isn't available is safe, though, although you have
2874to make sure you link against any libraries where the \f(CW\*(C`clock_gettime\*(C'\fR
2875function is hiding in (often \fI\-lrt\fR).
2876.IP "\s-1EV_USE_REALTIME\s0" 4
2877.IX Item "EV_USE_REALTIME"
2878If defined to be \f(CW1\fR, libev will try to detect the availability of the
2879realtime clock option at compiletime (and assume its availability at
2880runtime if successful). Otherwise no use of the realtime clock option will
2881be attempted. This effectively replaces \f(CW\*(C`gettimeofday\*(C'\fR by \f(CW\*(C`clock_get
2882(CLOCK_REALTIME, ...)\*(C'\fR and will not normally affect correctness. See the
2883note about libraries in the description of \f(CW\*(C`EV_USE_MONOTONIC\*(C'\fR, though.
2884.IP "\s-1EV_USE_NANOSLEEP\s0" 4
2885.IX Item "EV_USE_NANOSLEEP"
2886If defined to be \f(CW1\fR, libev will assume that \f(CW\*(C`nanosleep ()\*(C'\fR is available
2887and will use it for delays. Otherwise it will use \f(CW\*(C`select ()\*(C'\fR.
2888.IP "\s-1EV_USE_SELECT\s0" 4
2889.IX Item "EV_USE_SELECT"
2890If undefined or defined to be \f(CW1\fR, libev will compile in support for the
2891\&\f(CW\*(C`select\*(C'\fR(2) backend. No attempt at autodetection will be done: if no
2892other method takes over, select will be it. Otherwise the select backend
2893will not be compiled in.
2894.IP "\s-1EV_SELECT_USE_FD_SET\s0" 4
2895.IX Item "EV_SELECT_USE_FD_SET"
2896If defined to \f(CW1\fR, then the select backend will use the system \f(CW\*(C`fd_set\*(C'\fR
2897structure. This is useful if libev doesn't compile due to a missing
2898\&\f(CW\*(C`NFDBITS\*(C'\fR or \f(CW\*(C`fd_mask\*(C'\fR definition or it misguesses the bitset layout on
2899exotic systems. This usually limits the range of file descriptors to some
2900low limit such as 1024 or might have other limitations (winsocket only
2901allows 64 sockets). The \f(CW\*(C`FD_SETSIZE\*(C'\fR macro, set before compilation, might
2902influence the size of the \f(CW\*(C`fd_set\*(C'\fR used.
2903.IP "\s-1EV_SELECT_IS_WINSOCKET\s0" 4
2904.IX Item "EV_SELECT_IS_WINSOCKET"
2905When defined to \f(CW1\fR, the select backend will assume that
2906select/socket/connect etc. don't understand file descriptors but
2907wants osf handles on win32 (this is the case when the select to
2908be used is the winsock select). This means that it will call
2909\&\f(CW\*(C`_get_osfhandle\*(C'\fR on the fd to convert it to an \s-1OS\s0 handle. Otherwise,
2910it is assumed that all these functions actually work on fds, even
2911on win32. Should not be defined on non\-win32 platforms.
2912.IP "\s-1EV_FD_TO_WIN32_HANDLE\s0" 4
2913.IX Item "EV_FD_TO_WIN32_HANDLE"
2914If \f(CW\*(C`EV_SELECT_IS_WINSOCKET\*(C'\fR is enabled, then libev needs a way to map
2915file descriptors to socket handles. When not defining this symbol (the
2916default), then libev will call \f(CW\*(C`_get_osfhandle\*(C'\fR, which is usually
2917correct. In some cases, programs use their own file descriptor management,
2918in which case they can provide this function to map fds to socket handles.
2919.IP "\s-1EV_USE_POLL\s0" 4
2920.IX Item "EV_USE_POLL"
2921If defined to be \f(CW1\fR, libev will compile in support for the \f(CW\*(C`poll\*(C'\fR(2)
2922backend. Otherwise it will be enabled on non\-win32 platforms. It
2923takes precedence over select.
2924.IP "\s-1EV_USE_EPOLL\s0" 4
2925.IX Item "EV_USE_EPOLL"
2926If defined to be \f(CW1\fR, libev will compile in support for the Linux
2927\&\f(CW\*(C`epoll\*(C'\fR(7) backend. Its availability will be detected at runtime,
2928otherwise another method will be used as fallback. This is the
2929preferred backend for GNU/Linux systems.
2930.IP "\s-1EV_USE_KQUEUE\s0" 4
2931.IX Item "EV_USE_KQUEUE"
2932If defined to be \f(CW1\fR, libev will compile in support for the \s-1BSD\s0 style
2933\&\f(CW\*(C`kqueue\*(C'\fR(2) backend. Its actual availability will be detected at runtime,
2934otherwise another method will be used as fallback. This is the preferred
2935backend for \s-1BSD\s0 and BSD-like systems, although on most BSDs kqueue only
2936supports some types of fds correctly (the only platform we found that
2937supports ptys for example was NetBSD), so kqueue might be compiled in, but
2938not be used unless explicitly requested. The best way to use it is to find
2939out whether kqueue supports your type of fd properly and use an embedded
2940kqueue loop.
2941.IP "\s-1EV_USE_PORT\s0" 4
2942.IX Item "EV_USE_PORT"
2943If defined to be \f(CW1\fR, libev will compile in support for the Solaris
294410 port style backend. Its availability will be detected at runtime,
2945otherwise another method will be used as fallback. This is the preferred
2946backend for Solaris 10 systems.
2947.IP "\s-1EV_USE_DEVPOLL\s0" 4
2948.IX Item "EV_USE_DEVPOLL"
2949reserved for future expansion, works like the \s-1USE\s0 symbols above.
2950.IP "\s-1EV_USE_INOTIFY\s0" 4
2951.IX Item "EV_USE_INOTIFY"
2952If defined to be \f(CW1\fR, libev will compile in support for the Linux inotify
2953interface to speed up \f(CW\*(C`ev_stat\*(C'\fR watchers. Its actual availability will
2954be detected at runtime.
2955.IP "\s-1EV_ATOMIC_T\s0" 4
2956.IX Item "EV_ATOMIC_T"
2957Libev requires an integer type (suitable for storing \f(CW0\fR or \f(CW1\fR) whose
2958access is atomic with respect to other threads or signal contexts. No such
2959type is easily found in the C language, so you can provide your own type
2960that you know is safe for your purposes. It is used both for signal handler \*(L"locking\*(R"
2961as well as for signal and thread safety in \f(CW\*(C`ev_async\*(C'\fR watchers.
2962.Sp
2963In the absense of this define, libev will use \f(CW\*(C`sig_atomic_t volatile\*(C'\fR
2964(from \fIsignal.h\fR), which is usually good enough on most platforms.
2965.IP "\s-1EV_H\s0" 4
2966.IX Item "EV_H"
2967The name of the \fIev.h\fR header file used to include it. The default if
2968undefined is \f(CW"ev.h"\fR in \fIevent.h\fR, \fIev.c\fR and \fIev++.h\fR. This can be
2969used to virtually rename the \fIev.h\fR header file in case of conflicts.
2970.IP "\s-1EV_CONFIG_H\s0" 4
2971.IX Item "EV_CONFIG_H"
2972If \f(CW\*(C`EV_STANDALONE\*(C'\fR isn't \f(CW1\fR, this variable can be used to override
2973\&\fIev.c\fR's idea of where to find the \fIconfig.h\fR file, similarly to
2974\&\f(CW\*(C`EV_H\*(C'\fR, above.
2975.IP "\s-1EV_EVENT_H\s0" 4
2976.IX Item "EV_EVENT_H"
2977Similarly to \f(CW\*(C`EV_H\*(C'\fR, this macro can be used to override \fIevent.c\fR's idea
2978of how the \fIevent.h\fR header can be found, the default is \f(CW"event.h"\fR.
2979.IP "\s-1EV_PROTOTYPES\s0" 4
2980.IX Item "EV_PROTOTYPES"
2981If defined to be \f(CW0\fR, then \fIev.h\fR will not define any function
2982prototypes, but still define all the structs and other symbols. This is
2983occasionally useful if you want to provide your own wrapper functions
2984around libev functions.
2985.IP "\s-1EV_MULTIPLICITY\s0" 4
2986.IX Item "EV_MULTIPLICITY"
2987If undefined or defined to \f(CW1\fR, then all event-loop-specific functions
2988will have the \f(CW\*(C`struct ev_loop *\*(C'\fR as first argument, and you can create
2989additional independent event loops. Otherwise there will be no support
2990for multiple event loops and there is no first event loop pointer
2991argument. Instead, all functions act on the single default loop.
2992.IP "\s-1EV_MINPRI\s0" 4
2993.IX Item "EV_MINPRI"
2994.PD 0
2995.IP "\s-1EV_MAXPRI\s0" 4
2996.IX Item "EV_MAXPRI"
2997.PD
2998The range of allowed priorities. \f(CW\*(C`EV_MINPRI\*(C'\fR must be smaller or equal to
2999\&\f(CW\*(C`EV_MAXPRI\*(C'\fR, but otherwise there are no non-obvious limitations. You can
3000provide for more priorities by overriding those symbols (usually defined
3001to be \f(CW\*(C`\-2\*(C'\fR and \f(CW2\fR, respectively).
3002.Sp
3003When doing priority-based operations, libev usually has to linearly search
3004all the priorities, so having many of them (hundreds) uses a lot of space
3005and time, so using the defaults of five priorities (\-2 .. +2) is usually
3006fine.
3007.Sp
3008If your embedding app does not need any priorities, defining these both to
3009\&\f(CW0\fR will save some memory and cpu.
3010.IP "\s-1EV_PERIODIC_ENABLE\s0" 4
3011.IX Item "EV_PERIODIC_ENABLE"
3012If undefined or defined to be \f(CW1\fR, then periodic timers are supported. If
3013defined to be \f(CW0\fR, then they are not. Disabling them saves a few kB of
3014code.
3015.IP "\s-1EV_IDLE_ENABLE\s0" 4
3016.IX Item "EV_IDLE_ENABLE"
3017If undefined or defined to be \f(CW1\fR, then idle watchers are supported. If
3018defined to be \f(CW0\fR, then they are not. Disabling them saves a few kB of
3019code.
3020.IP "\s-1EV_EMBED_ENABLE\s0" 4
3021.IX Item "EV_EMBED_ENABLE"
3022If undefined or defined to be \f(CW1\fR, then embed watchers are supported. If
3023defined to be \f(CW0\fR, then they are not.
3024.IP "\s-1EV_STAT_ENABLE\s0" 4
3025.IX Item "EV_STAT_ENABLE"
3026If undefined or defined to be \f(CW1\fR, then stat watchers are supported. If
3027defined to be \f(CW0\fR, then they are not.
3028.IP "\s-1EV_FORK_ENABLE\s0" 4
3029.IX Item "EV_FORK_ENABLE"
3030If undefined or defined to be \f(CW1\fR, then fork watchers are supported. If
3031defined to be \f(CW0\fR, then they are not.
3032.IP "\s-1EV_ASYNC_ENABLE\s0" 4
3033.IX Item "EV_ASYNC_ENABLE"
3034If undefined or defined to be \f(CW1\fR, then async watchers are supported. If
3035defined to be \f(CW0\fR, then they are not.
3036.IP "\s-1EV_MINIMAL\s0" 4
3037.IX Item "EV_MINIMAL"
3038If you need to shave off some kilobytes of code at the expense of some
3039speed, define this symbol to \f(CW1\fR. Currently only used for gcc to override
3040some inlining decisions, saves roughly 30% codesize of amd64.
3041.IP "\s-1EV_PID_HASHSIZE\s0" 4
3042.IX Item "EV_PID_HASHSIZE"
3043\&\f(CW\*(C`ev_child\*(C'\fR watchers use a small hash table to distribute workload by
3044pid. The default size is \f(CW16\fR (or \f(CW1\fR with \f(CW\*(C`EV_MINIMAL\*(C'\fR), usually more
3045than enough. If you need to manage thousands of children you might want to
3046increase this value (\fImust\fR be a power of two).
3047.IP "\s-1EV_INOTIFY_HASHSIZE\s0" 4
3048.IX Item "EV_INOTIFY_HASHSIZE"
3049\&\f(CW\*(C`ev_stat\*(C'\fR watchers use a small hash table to distribute workload by
3050inotify watch id. The default size is \f(CW16\fR (or \f(CW1\fR with \f(CW\*(C`EV_MINIMAL\*(C'\fR),
3051usually more than enough. If you need to manage thousands of \f(CW\*(C`ev_stat\*(C'\fR
3052watchers you might want to increase this value (\fImust\fR be a power of
3053two).
3054.IP "\s-1EV_COMMON\s0" 4
3055.IX Item "EV_COMMON"
3056By default, all watchers have a \f(CW\*(C`void *data\*(C'\fR member. By redefining
3057this macro to a something else you can include more and other types of
3058members. You have to define it each time you include one of the files,
3059though, and it must be identical each time.
3060.Sp
3061For example, the perl \s-1EV\s0 module uses something like this:
3062.Sp
3063.Vb 3
3064\& #define EV_COMMON \e
3065\& SV *self; /* contains this struct */ \e
3066\& SV *cb_sv, *fh /* note no trailing ";" */
3067.Ve
3068.IP "\s-1EV_CB_DECLARE\s0 (type)" 4
3069.IX Item "EV_CB_DECLARE (type)"
3070.PD 0
3071.IP "\s-1EV_CB_INVOKE\s0 (watcher, revents)" 4
3072.IX Item "EV_CB_INVOKE (watcher, revents)"
3073.IP "ev_set_cb (ev, cb)" 4
3074.IX Item "ev_set_cb (ev, cb)"
3075.PD
3076Can be used to change the callback member declaration in each watcher,
3077and the way callbacks are invoked and set. Must expand to a struct member
3078definition and a statement, respectively. See the \fIev.h\fR header file for
3079their default definitions. One possible use for overriding these is to
3080avoid the \f(CW\*(C`struct ev_loop *\*(C'\fR as first argument in all cases, or to use
3081method calls instead of plain function calls in \*(C+.
3082.Sh "\s-1EXPORTED\s0 \s-1API\s0 \s-1SYMBOLS\s0"
3083.IX Subsection "EXPORTED API SYMBOLS"
3084If you need to re-export the \s-1API\s0 (e.g. via a dll) and you need a list of
3085exported symbols, you can use the provided \fISymbol.*\fR files which list
3086all public symbols, one per line:
3087.PP
3088.Vb 2
3089\& Symbols.ev for libev proper
3090\& Symbols.event for the libevent emulation
3091.Ve
3092.PP
3093This can also be used to rename all public symbols to avoid clashes with
3094multiple versions of libev linked together (which is obviously bad in
3095itself, but sometimes it is inconvinient to avoid this).
3096.PP
3097A sed command like this will create wrapper \f(CW\*(C`#define\*(C'\fR's that you need to
3098include before including \fIev.h\fR:
3099.PP
3100.Vb 1
3101\& <Symbols.ev sed \-e "s/.*/#define & myprefix_&/" >wrap.h
3102.Ve
3103.PP
3104This would create a file \fIwrap.h\fR which essentially looks like this:
3105.PP
3106.Vb 4
3107\& #define ev_backend myprefix_ev_backend
3108\& #define ev_check_start myprefix_ev_check_start
3109\& #define ev_check_stop myprefix_ev_check_stop
3110\& ...
3111.Ve
3112.Sh "\s-1EXAMPLES\s0"
3113.IX Subsection "EXAMPLES"
3114For a real-world example of a program the includes libev
3115verbatim, you can have a look at the \s-1EV\s0 perl module
3116(<http://software.schmorp.de/pkg/EV.html>). It has the libev files in
3117the \fIlibev/\fR subdirectory and includes them in the \fI\s-1EV/EVAPI\s0.h\fR (public
3118interface) and \fI\s-1EV\s0.xs\fR (implementation) files. Only the \fI\s-1EV\s0.xs\fR file
3119will be compiled. It is pretty complex because it provides its own header
3120file.
3121.PP
3122The usage in rxvt-unicode is simpler. It has a \fIev_cpp.h\fR header file
3123that everybody includes and which overrides some configure choices:
3124.PP
3125.Vb 9
3126\& #define EV_MINIMAL 1
3127\& #define EV_USE_POLL 0
3128\& #define EV_MULTIPLICITY 0
3129\& #define EV_PERIODIC_ENABLE 0
3130\& #define EV_STAT_ENABLE 0
3131\& #define EV_FORK_ENABLE 0
3132\& #define EV_CONFIG_H <config.h>
3133\& #define EV_MINPRI 0
3134\& #define EV_MAXPRI 0
3135\&
3136\& #include "ev++.h"
3137.Ve
3138.PP
3139And a \fIev_cpp.C\fR implementation file that contains libev proper and is compiled:
3140.PP
3141.Vb 2
3142\& #include "ev_cpp.h"
3143\& #include "ev.c"
3144.Ve
3145.SH "COMPLEXITIES"
3146.IX Header "COMPLEXITIES"
3147In this section the complexities of (many of) the algorithms used inside
3148libev will be explained. For complexity discussions about backends see the
3149documentation for \f(CW\*(C`ev_default_init\*(C'\fR.
3150.PP
3151All of the following are about amortised time: If an array needs to be
3152extended, libev needs to realloc and move the whole array, but this
3153happens asymptotically never with higher number of elements, so O(1) might
3154mean it might do a lengthy realloc operation in rare cases, but on average
3155it is much faster and asymptotically approaches constant time.
3156.IP "Starting and stopping timer/periodic watchers: O(log skipped_other_timers)" 4
3157.IX Item "Starting and stopping timer/periodic watchers: O(log skipped_other_timers)"
3158This means that, when you have a watcher that triggers in one hour and
3159there are 100 watchers that would trigger before that then inserting will
3160have to skip roughly seven (\f(CW\*(C`ld 100\*(C'\fR) of these watchers.
3161.IP "Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)" 4
3162.IX Item "Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)"
3163That means that changing a timer costs less than removing/adding them
3164as only the relative motion in the event queue has to be paid for.
3165.IP "Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)" 4
3166.IX Item "Starting io/check/prepare/idle/signal/child/fork/async watchers: O(1)"
3167These just add the watcher into an array or at the head of a list.
3168.IP "Stopping check/prepare/idle/fork/async watchers: O(1)" 4
3169.IX Item "Stopping check/prepare/idle/fork/async watchers: O(1)"
3170.PD 0
3171.IP "Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % \s-1EV_PID_HASHSIZE\s0))" 4
3172.IX Item "Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))"
3173.PD
3174These watchers are stored in lists then need to be walked to find the
3175correct watcher to remove. The lists are usually short (you don't usually
3176have many watchers waiting for the same fd or signal).
3177.IP "Finding the next timer in each loop iteration: O(1)" 4
3178.IX Item "Finding the next timer in each loop iteration: O(1)"
3179By virtue of using a binary heap, the next timer is always found at the
3180beginning of the storage array.
3181.IP "Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)" 4
3182.IX Item "Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)"
3183A change means an I/O watcher gets started or stopped, which requires
3184libev to recalculate its status (and possibly tell the kernel, depending
3185on backend and wether \f(CW\*(C`ev_io_set\*(C'\fR was used).
3186.IP "Activating one watcher (putting it into the pending state): O(1)" 4
3187.IX Item "Activating one watcher (putting it into the pending state): O(1)"
3188.PD 0
3189.IP "Priority handling: O(number_of_priorities)" 4
3190.IX Item "Priority handling: O(number_of_priorities)"
3191.PD
3192Priorities are implemented by allocating some space for each
3193priority. When doing priority-based operations, libev usually has to
3194linearly search all the priorities, but starting/stopping and activating
3195watchers becomes O(1) w.r.t. priority handling.
3196.IP "Sending an ev_async: O(1)" 4
3197.IX Item "Sending an ev_async: O(1)"
3198.PD 0
3199.IP "Processing ev_async_send: O(number_of_async_watchers)" 4
3200.IX Item "Processing ev_async_send: O(number_of_async_watchers)"
3201.IP "Processing signals: O(max_signal_number)" 4
3202.IX Item "Processing signals: O(max_signal_number)"
3203.PD
3204Sending involves a syscall \fIiff\fR there were no other \f(CW\*(C`ev_async_send\*(C'\fR
3205calls in the current loop iteration. Checking for async and signal events
3206involves iterating over all running async watchers or all signal numbers.
3207.SH "Win32 platform limitations and workarounds"
3208.IX Header "Win32 platform limitations and workarounds"
3209Win32 doesn't support any of the standards (e.g. \s-1POSIX\s0) that libev
3210requires, and its I/O model is fundamentally incompatible with the \s-1POSIX\s0
3211model. Libev still offers limited functionality on this platform in
3212the form of the \f(CW\*(C`EVBACKEND_SELECT\*(C'\fR backend, and only supports socket
3213descriptors. This only applies when using Win32 natively, not when using
3214e.g. cygwin.
3215.PP
3216There is no supported compilation method available on windows except
3217embedding it into other applications.
3218.PP
3219Due to the many, low, and arbitrary limits on the win32 platform and the
3220abysmal performance of winsockets, using a large number of sockets is not
3221recommended (and not reasonable). If your program needs to use more than
3222a hundred or so sockets, then likely it needs to use a totally different
3223implementation for windows, as libev offers the \s-1POSIX\s0 model, which cannot
3224be implemented efficiently on windows (microsoft monopoly games).
3225.IP "The winsocket select function" 4
3226.IX Item "The winsocket select function"
3227The winsocket \f(CW\*(C`select\*(C'\fR function doesn't follow \s-1POSIX\s0 in that it requires
3228socket \fIhandles\fR and not socket \fIfile descriptors\fR. This makes select
3229very inefficient, and also requires a mapping from file descriptors
3230to socket handles. See the discussion of the \f(CW\*(C`EV_SELECT_USE_FD_SET\*(C'\fR,
3231\&\f(CW\*(C`EV_SELECT_IS_WINSOCKET\*(C'\fR and \f(CW\*(C`EV_FD_TO_WIN32_HANDLE\*(C'\fR preprocessor
3232symbols for more info.
3233.Sp
3234The configuration for a \*(L"naked\*(R" win32 using the microsoft runtime
3235libraries and raw winsocket select is:
3236.Sp
3237.Vb 2
3238\& #define EV_USE_SELECT 1
3239\& #define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
3240.Ve
3241.Sp
3242Note that winsockets handling of fd sets is O(n), so you can easily get a
3243complexity in the O(nA\*^X) range when using win32.
3244.IP "Limited number of file descriptors" 4
3245.IX Item "Limited number of file descriptors"
3246Windows has numerous arbitrary (and low) limits on things. Early versions
3247of winsocket's select only supported waiting for a max. of \f(CW64\fR handles
3248(probably owning to the fact that all windows kernels can only wait for
3249\&\f(CW64\fR things at the same time internally; microsoft recommends spawning a
3250chain of threads and wait for 63 handles and the previous thread in each).
3251.Sp
3252Newer versions support more handles, but you need to define \f(CW\*(C`FD_SETSIZE\*(C'\fR
3253to some high number (e.g. \f(CW2048\fR) before compiling the winsocket select
3254call (which might be in libev or elsewhere, for example, perl does its own
3255select emulation on windows).
3256.Sp
3257Another limit is the number of file descriptors in the microsoft runtime
3258libraries, which by default is \f(CW64\fR (there must be a hidden \fI64\fR fetish
3259or something like this inside microsoft). You can increase this by calling
3260\&\f(CW\*(C`_setmaxstdio\*(C'\fR, which can increase this limit to \f(CW2048\fR (another
3261arbitrary limit), but is broken in many versions of the microsoft runtime
3262libraries.
3263.Sp
3264This might get you to about \f(CW512\fR or \f(CW2048\fR sockets (depending on
3265windows version and/or the phase of the moon). To get more, you need to
3266wrap all I/O functions and provide your own fd management, but the cost of
3267calling select (O(nA\*^X)) will likely make this unworkable.
1267.SH "AUTHOR" 3268.SH "AUTHOR"
1268.IX Header "AUTHOR" 3269.IX Header "AUTHOR"
1269Marc Lehmann <libev@schmorp.de>. 3270Marc Lehmann <libev@schmorp.de>.
3271.SH "POD ERRORS"
3272.IX Header "POD ERRORS"
3273Hey! \fBThe above document had some coding errors, which are explained below:\fR
3274.IP "Around line 2996:" 4
3275.IX Item "Around line 2996:"
3276You forgot a '=back' before '=head2'

Diff Legend

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