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

Comparing libev/ev.pod (file contents):
Revision 1.5 by root, Mon Nov 12 08:11:57 2007 UTC vs.
Revision 1.17 by root, Mon Nov 12 08:57:03 2007 UTC

… …
26Libev supports select, poll, the linux-specific epoll and the bsd-specific 26Libev supports select, poll, the linux-specific epoll and the bsd-specific
27kqueue mechanisms for file descriptor events, relative timers, absolute 27kqueue mechanisms for file descriptor events, relative timers, absolute
28timers with customised rescheduling, signal events, process status change 28timers with customised rescheduling, signal events, process status change
29events (related to SIGCHLD), and event watchers dealing with the event 29events (related to SIGCHLD), and event watchers dealing with the event
30loop mechanism itself (idle, prepare and check watchers). It also is quite 30loop mechanism itself (idle, prepare and check watchers). It also is quite
31fast (see a L<http://libev.schmorp.de/bench.html|benchmark> comparing it 31fast (see this L<benchmark|http://libev.schmorp.de/bench.html> comparing
32to libevent). 32it to libevent for example).
33 33
34=head1 CONVENTIONS 34=head1 CONVENTIONS
35 35
36Libev is very configurable. In this manual the default configuration 36Libev is very configurable. In this manual the default configuration
37will be described, which supports multiple event loops. For more info 37will be described, which supports multiple event loops. For more info
38about various configuraiton options please have a look at the file 38about various configuration options please have a look at the file
39F<README.embed> in the libev distribution. If libev was configured without 39F<README.embed> in the libev distribution. If libev was configured without
40support for multiple event loops, then all functions taking an initial 40support for multiple event loops, then all functions taking an initial
41argument of name C<loop> (which is always of type C<struct ev_loop *>) 41argument of name C<loop> (which is always of type C<struct ev_loop *>)
42will not have this argument. 42will not have this argument.
43 43
44=head1 TIME AND OTHER GLOBAL FUNCTIONS 44=head1 TIME REPRESENTATION
45 45
46Libev represents time as a single floating point number, representing the 46Libev represents time as a single floating point number, representing the
47(fractional) number of seconds since the (POSIX) epoch (somewhere near 47(fractional) number of seconds since the (POSIX) epoch (somewhere near
48the beginning of 1970, details are complicated, don't ask). This type is 48the beginning of 1970, details are complicated, don't ask). This type is
49called C<ev_tstamp>, which is what you should use too. It usually aliases 49called C<ev_tstamp>, which is what you should use too. It usually aliases
50to the double type in C. 50to the double type in C.
51 51
52=head1 GLOBAL FUNCTIONS
53
52=over 4 54=over 4
53 55
54=item ev_tstamp ev_time () 56=item ev_tstamp ev_time ()
55 57
56Returns the current time as libev would use it. 58Returns the current time as libev would use it.
… …
63you linked against by calling the functions C<ev_version_major> and 65you linked against by calling the functions C<ev_version_major> and
64C<ev_version_minor>. If you want, you can compare against the global 66C<ev_version_minor>. If you want, you can compare against the global
65symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the 67symbols C<EV_VERSION_MAJOR> and C<EV_VERSION_MINOR>, which specify the
66version of the library your program was compiled against. 68version of the library your program was compiled against.
67 69
68Usually, its a good idea to terminate if the major versions mismatch, 70Usually, it's a good idea to terminate if the major versions mismatch,
69as this indicates an incompatible change. Minor versions are usually 71as this indicates an incompatible change. Minor versions are usually
70compatible to older versions, so a larger minor version alone is usually 72compatible to older versions, so a larger minor version alone is usually
71not a problem. 73not a problem.
72 74
73=item ev_set_allocator (void *(*cb)(void *ptr, long size)) 75=item ev_set_allocator (void *(*cb)(void *ptr, long size))
74 76
75Sets the allocation function to use (the prototype is similar to the 77Sets the allocation function to use (the prototype is similar to the
76realloc function). It is used to allocate and free memory (no surprises 78realloc C function, the semantics are identical). It is used to allocate
77here). If it returns zero when memory needs to be allocated, the library 79and free memory (no surprises here). If it returns zero when memory
78might abort or take some potentially destructive action. The default is 80needs to be allocated, the library might abort or take some potentially
79your system realloc function. 81destructive action. The default is your system realloc function.
80 82
81You could override this function in high-availability programs to, say, 83You could override this function in high-availability programs to, say,
82free some memory if it cannot allocate memory, to use a special allocator, 84free some memory if it cannot allocate memory, to use a special allocator,
83or even to sleep a while and retry until some memory is available. 85or even to sleep a while and retry until some memory is available.
84 86
… …
86 88
87Set the callback function to call on a retryable syscall error (such 89Set the callback function to call on a retryable syscall error (such
88as failed select, poll, epoll_wait). The message is a printable string 90as failed select, poll, epoll_wait). The message is a printable string
89indicating the system call or subsystem causing the problem. If this 91indicating the system call or subsystem causing the problem. If this
90callback is set, then libev will expect it to remedy the sitution, no 92callback is set, then libev will expect it to remedy the sitution, no
91matter what, when it returns. That is, libev will geenrally retry the 93matter what, when it returns. That is, libev will generally retry the
92requested operation, or, if the condition doesn't go away, do bad stuff 94requested operation, or, if the condition doesn't go away, do bad stuff
93(such as abort). 95(such as abort).
94 96
95=back 97=back
96 98
… …
99An event loop is described by a C<struct ev_loop *>. The library knows two 101An event loop is described by a C<struct ev_loop *>. The library knows two
100types of such loops, the I<default> loop, which supports signals and child 102types of such loops, the I<default> loop, which supports signals and child
101events, and dynamically created loops which do not. 103events, and dynamically created loops which do not.
102 104
103If you use threads, a common model is to run the default event loop 105If you use threads, a common model is to run the default event loop
104in your main thread (or in a separate thrad) and for each thread you 106in your main thread (or in a separate thread) and for each thread you
105create, you also create another event loop. Libev itself does no lockign 107create, you also create another event loop. Libev itself does no locking
106whatsoever, so if you mix calls to different event loops, make sure you 108whatsoever, so if you mix calls to the same event loop in different
107lock (this is usually a bad idea, though, even if done right). 109threads, make sure you lock (this is usually a bad idea, though, even if
110done correctly, because it's hideous and inefficient).
108 111
109=over 4 112=over 4
110 113
111=item struct ev_loop *ev_default_loop (unsigned int flags) 114=item struct ev_loop *ev_default_loop (unsigned int flags)
112 115
… …
117 120
118If you don't know what event loop to use, use the one returned from this 121If you don't know what event loop to use, use the one returned from this
119function. 122function.
120 123
121The flags argument can be used to specify special behaviour or specific 124The flags argument can be used to specify special behaviour or specific
122backends to use, and is usually specified as 0 (or EVFLAG_AUTO) 125backends to use, and is usually specified as 0 (or EVFLAG_AUTO).
123 126
124It supports the following flags: 127It supports the following flags:
125 128
126=over 4 129=over 4
127 130
128=item EVFLAG_AUTO 131=item C<EVFLAG_AUTO>
129 132
130The default flags value. Use this if you have no clue (its the right 133The default flags value. Use this if you have no clue (it's the right
131thing, believe me). 134thing, believe me).
132 135
133=item EVFLAG_NOENV 136=item C<EVFLAG_NOENV>
134 137
135If this flag bit is ored into the flag value then libev will I<not> look 138If this flag bit is ored into the flag value (or the program runs setuid
136at the environment variable C<LIBEV_FLAGS>. Otherwise (the default), this 139or setgid) then libev will I<not> look at the environment variable
137environment variable will override the flags completely. This is useful 140C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
141override the flags completely if it is found in the environment. This is
138to try out specific backends to tets their performance, or to work around 142useful to try out specific backends to test their performance, or to work
139bugs. 143around bugs.
140 144
141=item EVMETHOD_SELECT portable select backend 145=item C<EVMETHOD_SELECT> (portable select backend)
142 146
143=item EVMETHOD_POLL poll backend (everywhere except windows) 147=item C<EVMETHOD_POLL> (poll backend, available everywhere except on windows)
144 148
145=item EVMETHOD_EPOLL linux only 149=item C<EVMETHOD_EPOLL> (linux only)
146 150
147=item EVMETHOD_KQUEUE some bsds only 151=item C<EVMETHOD_KQUEUE> (some bsds only)
148 152
149=item EVMETHOD_DEVPOLL solaris 8 only 153=item C<EVMETHOD_DEVPOLL> (solaris 8 only)
150 154
151=item EVMETHOD_PORT solaris 10 only 155=item C<EVMETHOD_PORT> (solaris 10 only)
152 156
153If one or more of these are ored into the flags value, then only these 157If one or more of these are ored into the flags value, then only these
154backends will be tried (in the reverse order as given here). If one are 158backends will be tried (in the reverse order as given here). If one are
155specified, any backend will do. 159specified, any backend will do.
156 160
… …
165 169
166=item ev_default_destroy () 170=item ev_default_destroy ()
167 171
168Destroys the default loop again (frees all memory and kernel state 172Destroys the default loop again (frees all memory and kernel state
169etc.). This stops all registered event watchers (by not touching them in 173etc.). This stops all registered event watchers (by not touching them in
170any way whatsoever, although you cnanot rely on this :). 174any way whatsoever, although you cannot rely on this :).
171 175
172=item ev_loop_destroy (loop) 176=item ev_loop_destroy (loop)
173 177
174Like C<ev_default_destroy>, but destroys an event loop created by an 178Like C<ev_default_destroy>, but destroys an event loop created by an
175earlier call to C<ev_loop_new>. 179earlier call to C<ev_loop_new>.
… …
183 187
184You I<must> call this function after forking if and only if you want to 188You I<must> call this function after forking if and only if you want to
185use the event library in both processes. If you just fork+exec, you don't 189use the event library in both processes. If you just fork+exec, you don't
186have to call it. 190have to call it.
187 191
188The function itself is quite fast and its usually not a problem to call 192The function itself is quite fast and it's usually not a problem to call
189it just in case after a fork. To make this easy, the function will fit in 193it just in case after a fork. To make this easy, the function will fit in
190quite nicely into a call to C<pthread_atfork>: 194quite nicely into a call to C<pthread_atfork>:
191 195
192 pthread_atfork (0, 0, ev_default_fork); 196 pthread_atfork (0, 0, ev_default_fork);
193 197
… …
200=item unsigned int ev_method (loop) 204=item unsigned int ev_method (loop)
201 205
202Returns one of the C<EVMETHOD_*> flags indicating the event backend in 206Returns one of the C<EVMETHOD_*> flags indicating the event backend in
203use. 207use.
204 208
205=item ev_tstamp = ev_now (loop) 209=item ev_tstamp ev_now (loop)
206 210
207Returns the current "event loop time", which is the time the event loop 211Returns the current "event loop time", which is the time the event loop
208got events and started processing them. This timestamp does not change 212got events and started processing them. This timestamp does not change
209as long as callbacks are being processed, and this is also the base time 213as long as callbacks are being processed, and this is also the base time
210used for relative timers. You can treat it as the timestamp of the event 214used for relative timers. You can treat it as the timestamp of the event
… …
219If the flags argument is specified as 0, it will not return until either 223If the flags argument is specified as 0, it will not return until either
220no event watchers are active anymore or C<ev_unloop> was called. 224no event watchers are active anymore or C<ev_unloop> was called.
221 225
222A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle 226A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle
223those events and any outstanding ones, but will not block your process in 227those events and any outstanding ones, but will not block your process in
224case there are no events. 228case there are no events and will return after one iteration of the loop.
225 229
226A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if 230A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if
227neccessary) and will handle those and any outstanding ones. It will block 231neccessary) and will handle those and any outstanding ones. It will block
228your process until at least one new event arrives. 232your process until at least one new event arrives, and will return after
233one iteration of the loop.
229 234
230This flags value could be used to implement alternative looping 235This flags value could be used to implement alternative looping
231constructs, but the C<prepare> and C<check> watchers provide a better and 236constructs, but the C<prepare> and C<check> watchers provide a better and
232more generic mechanism. 237more generic mechanism.
233 238
234=item ev_unloop (loop, how) 239=item ev_unloop (loop, how)
235 240
236Can be used to make a call to C<ev_loop> return early. The C<how> argument 241Can be used to make a call to C<ev_loop> return early (but only after it
242has processed all outstanding events). The C<how> argument must be either
237must be either C<EVUNLOOP_ONCE>, which will make the innermost C<ev_loop> 243C<EVUNLOOP_ONCE>, which will make the innermost C<ev_loop> call return, or
238call return, or C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> 244C<EVUNLOOP_ALL>, which will make all nested C<ev_loop> calls return.
239calls return.
240 245
241=item ev_ref (loop) 246=item ev_ref (loop)
242 247
243=item ev_unref (loop) 248=item ev_unref (loop)
244 249
245Ref/unref can be used to add or remove a refcount on the event loop: Every 250Ref/unref can be used to add or remove a reference count on the event
246watcher keeps one reference. If you have a long-runing watcher you never 251loop: Every watcher keeps one reference, and as long as the reference
247unregister that should not keep ev_loop from running, ev_unref() after 252count is nonzero, C<ev_loop> will not return on its own. If you have
248starting, and ev_ref() before stopping it. Libev itself uses this for 253a watcher you never unregister that should not keep C<ev_loop> from
249example for its internal signal pipe: It is not visible to you as a user 254returning, ev_unref() after starting, and ev_ref() before stopping it. For
250and should not keep C<ev_loop> from exiting if the work is done. It is 255example, libev itself uses this for its internal signal pipe: It is not
251also an excellent way to do this for generic recurring timers or from 256visible to the libev user and should not keep C<ev_loop> from exiting if
252within third-party libraries. Just remember to unref after start and ref 257no event watchers registered by it are active. It is also an excellent
253before stop. 258way to do this for generic recurring timers or from within third-party
259libraries. Just remember to I<unref after start> and I<ref before stop>.
254 260
255=back 261=back
256 262
257=head1 ANATOMY OF A WATCHER 263=head1 ANATOMY OF A WATCHER
258 264
259A watcher is a structure that you create and register to record your 265A watcher is a structure that you create and register to record your
260interest in some event. For instance, if you want to wait for STDIN to 266interest in some event. For instance, if you want to wait for STDIN to
261become readable, you would create an ev_io watcher for that: 267become readable, you would create an C<ev_io> watcher for that:
262 268
263 static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents) 269 static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
264 { 270 {
265 ev_io_stop (w); 271 ev_io_stop (w);
266 ev_unloop (loop, EVUNLOOP_ALL); 272 ev_unloop (loop, EVUNLOOP_ALL);
… …
295 301
296As long as your watcher is active (has been started but not stopped) you 302As long as your watcher is active (has been started but not stopped) you
297must not touch the values stored in it. Most specifically you must never 303must not touch the values stored in it. Most specifically you must never
298reinitialise it or call its set method. 304reinitialise it or call its set method.
299 305
300You cna check whether an event is active by calling the C<ev_is_active 306You can check whether an event is active by calling the C<ev_is_active
301(watcher *)> macro. To see whether an event is outstanding (but the 307(watcher *)> macro. To see whether an event is outstanding (but the
302callback for it has not been called yet) you cna use the C<ev_is_pending 308callback for it has not been called yet) you can use the C<ev_is_pending
303(watcher *)> macro. 309(watcher *)> macro.
304 310
305Each and every callback receives the event loop pointer as first, the 311Each and every callback receives the event loop pointer as first, the
306registered watcher structure as second, and a bitset of received events as 312registered watcher structure as second, and a bitset of received events as
307third argument. 313third argument.
308 314
309The rceeived events usually include a single bit per event type received 315The received events usually include a single bit per event type received
310(you can receive multiple events at the same time). The possible bit masks 316(you can receive multiple events at the same time). The possible bit masks
311are: 317are:
312 318
313=over 4 319=over 4
314 320
315=item EV_READ 321=item C<EV_READ>
316 322
317=item EV_WRITE 323=item C<EV_WRITE>
318 324
319The file descriptor in the ev_io watcher has become readable and/or 325The file descriptor in the C<ev_io> watcher has become readable and/or
320writable. 326writable.
321 327
322=item EV_TIMEOUT 328=item C<EV_TIMEOUT>
323 329
324The ev_timer watcher has timed out. 330The C<ev_timer> watcher has timed out.
325 331
326=item EV_PERIODIC 332=item C<EV_PERIODIC>
327 333
328The ev_periodic watcher has timed out. 334The C<ev_periodic> watcher has timed out.
329 335
330=item EV_SIGNAL 336=item C<EV_SIGNAL>
331 337
332The signal specified in the ev_signal watcher has been received by a thread. 338The signal specified in the C<ev_signal> watcher has been received by a thread.
333 339
334=item EV_CHILD 340=item C<EV_CHILD>
335 341
336The pid specified in the ev_child watcher has received a status change. 342The pid specified in the C<ev_child> watcher has received a status change.
337 343
338=item EV_IDLE 344=item C<EV_IDLE>
339 345
340The ev_idle watcher has determined that you have nothing better to do. 346The C<ev_idle> watcher has determined that you have nothing better to do.
341 347
342=item EV_PREPARE 348=item C<EV_PREPARE>
343 349
344=item EV_CHECK 350=item C<EV_CHECK>
345 351
346All ev_prepare watchers are invoked just I<before> C<ev_loop> starts 352All C<ev_prepare> watchers are invoked just I<before> C<ev_loop> starts
347to gather new events, and all ev_check watchers are invoked just after 353to gather new events, and all C<ev_check> watchers are invoked just after
348C<ev_loop> has gathered them, but before it invokes any callbacks for any 354C<ev_loop> has gathered them, but before it invokes any callbacks for any
349received events. Callbacks of both watcher types can start and stop as 355received events. Callbacks of both watcher types can start and stop as
350many watchers as they want, and all of them will be taken into account 356many watchers as they want, and all of them will be taken into account
351(for example, a ev_prepare watcher might start an idle watcher to keep 357(for example, a C<ev_prepare> watcher might start an idle watcher to keep
352C<ev_loop> from blocking). 358C<ev_loop> from blocking).
353 359
354=item EV_ERROR 360=item C<EV_ERROR>
355 361
356An unspecified error has occured, the watcher has been stopped. This might 362An unspecified error has occured, the watcher has been stopped. This might
357happen because the watcher could not be properly started because libev 363happen because the watcher could not be properly started because libev
358ran out of memory, a file descriptor was found to be closed or any other 364ran out of memory, a file descriptor was found to be closed or any other
359problem. You best act on it by reporting the problem and somehow coping 365problem. You best act on it by reporting the problem and somehow coping
… …
368=back 374=back
369 375
370=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER 376=head2 ASSOCIATING CUSTOM DATA WITH A WATCHER
371 377
372Each watcher has, by default, a member C<void *data> that you can change 378Each watcher has, by default, a member C<void *data> that you can change
373and read at any time, libev will completely ignore it. This cna be used 379and read at any time, libev will completely ignore it. This can be used
374to associate arbitrary data with your watcher. If you need more data and 380to associate arbitrary data with your watcher. If you need more data and
375don't want to allocate memory and store a pointer to it in that data 381don't want to allocate memory and store a pointer to it in that data
376member, you can also "subclass" the watcher type and provide your own 382member, you can also "subclass" the watcher type and provide your own
377data: 383data:
378 384
… …
400=head1 WATCHER TYPES 406=head1 WATCHER TYPES
401 407
402This section describes each watcher in detail, but will not repeat 408This section describes each watcher in detail, but will not repeat
403information given in the last section. 409information given in the last section.
404 410
405=head2 struct ev_io - is my file descriptor readable or writable 411=head2 C<ev_io> - is this file descriptor readable or writable
406 412
407I/O watchers check whether a file descriptor is readable or writable 413I/O watchers check whether a file descriptor is readable or writable
408in each iteration of the event loop (This behaviour is called 414in each iteration of the event loop (This behaviour is called
409level-triggering because you keep receiving events as long as the 415level-triggering because you keep receiving events as long as the
410condition persists. Remember you cna stop the watcher if you don't want to 416condition persists. Remember you can stop the watcher if you don't want to
411act on the event and neither want to receive future events). 417act on the event and neither want to receive future events).
412 418
419In general you can register as many read and/or write event watchers oer
420fd as you want (as long as you don't confuse yourself). Setting all file
421descriptors to non-blocking mode is also usually a good idea (but not
422required if you know what you are doing).
423
424You have to be careful with dup'ed file descriptors, though. Some backends
425(the linux epoll backend is a notable example) cannot handle dup'ed file
426descriptors correctly if you register interest in two or more fds pointing
427to the same file/socket etc. description.
428
429If you must do this, then force the use of a known-to-be-good backend
430(at the time of this writing, this includes only EVMETHOD_SELECT and
431EVMETHOD_POLL).
432
413=over 4 433=over 4
414 434
415=item ev_io_init (ev_io *, callback, int fd, int events) 435=item ev_io_init (ev_io *, callback, int fd, int events)
416 436
417=item ev_io_set (ev_io *, int fd, int events) 437=item ev_io_set (ev_io *, int fd, int events)
418 438
419Configures an ev_io watcher. The fd is the file descriptor to rceeive 439Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive
420events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ | 440events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |
421EV_WRITE> to receive the given events. 441EV_WRITE> to receive the given events.
422 442
423=back 443=back
424 444
425=head2 struct ev_timer - relative and optionally recurring timeouts 445=head2 C<ev_timer> - relative and optionally recurring timeouts
426 446
427Timer watchers are simple relative timers that generate an event after a 447Timer watchers are simple relative timers that generate an event after a
428given time, and optionally repeating in regular intervals after that. 448given time, and optionally repeating in regular intervals after that.
429 449
430The timers are based on real time, that is, if you register an event that 450The timers are based on real time, that is, if you register an event that
431times out after an hour and youreset your system clock to last years 451times out after an hour and youreset your system clock to last years
432time, it will still time out after (roughly) and hour. "Roughly" because 452time, it will still time out after (roughly) and hour. "Roughly" because
433detecting time jumps is hard, and soem inaccuracies are unavoidable (the 453detecting time jumps is hard, and soem inaccuracies are unavoidable (the
434monotonic clock option helps a lot here). 454monotonic clock option helps a lot here).
435 455
456The relative timeouts are calculated relative to the C<ev_now ()>
457time. This is usually the right thing as this timestamp refers to the time
458of the event triggering whatever timeout you are modifying/starting. If
459you suspect event processing to be delayed and you *need* to base the timeout
460ion the current time, use something like this to adjust for this:
461
462 ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
463
436=over 4 464=over 4
437 465
438=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat) 466=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
439 467
440=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat) 468=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
… …
462 490
463This sounds a bit complicated, but here is a useful and typical 491This sounds a bit complicated, but here is a useful and typical
464example: Imagine you have a tcp connection and you want a so-called idle 492example: Imagine you have a tcp connection and you want a so-called idle
465timeout, that is, you want to be called when there have been, say, 60 493timeout, that is, you want to be called when there have been, say, 60
466seconds of inactivity on the socket. The easiest way to do this is to 494seconds of inactivity on the socket. The easiest way to do this is to
467configure an ev_timer with after=repeat=60 and calling ev_timer_again each 495configure an C<ev_timer> with after=repeat=60 and calling ev_timer_again each
468time you successfully read or write some data. If you go into an idle 496time you successfully read or write some data. If you go into an idle
469state where you do not expect data to travel on the socket, you can stop 497state where you do not expect data to travel on the socket, you can stop
470the timer, and again will automatically restart it if need be. 498the timer, and again will automatically restart it if need be.
471 499
472=back 500=back
473 501
474=head2 ev_periodic - to cron or not to cron it 502=head2 C<ev_periodic> - to cron or not to cron
475 503
476Periodic watchers are also timers of a kind, but they are very versatile 504Periodic watchers are also timers of a kind, but they are very versatile
477(and unfortunately a bit complex). 505(and unfortunately a bit complex).
478 506
479Unlike ev_timer's, they are not based on real time (or relative time) 507Unlike C<ev_timer>'s, they are not based on real time (or relative time)
480but on wallclock time (absolute time). You can tell a periodic watcher 508but on wallclock time (absolute time). You can tell a periodic watcher
481to trigger "at" some specific point in time. For example, if you tell a 509to trigger "at" some specific point in time. For example, if you tell a
482periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now () 510periodic watcher to trigger in 10 seconds (by specifiying e.g. c<ev_now ()
483+ 10.>) and then reset your system clock to the last year, then it will 511+ 10.>) and then reset your system clock to the last year, then it will
484take a year to trigger the event (unlike an ev_timer, which would trigger 512take a year to trigger the event (unlike an C<ev_timer>, which would trigger
485roughly 10 seconds later and of course not if you reset your system time 513roughly 10 seconds later and of course not if you reset your system time
486again). 514again).
487 515
488They can also be used to implement vastly more complex timers, such as 516They can also be used to implement vastly more complex timers, such as
489triggering an event on eahc midnight, local time. 517triggering an event on eahc midnight, local time.
… …
518 546
519 ev_periodic_set (&periodic, 0., 3600., 0); 547 ev_periodic_set (&periodic, 0., 3600., 0);
520 548
521This doesn't mean there will always be 3600 seconds in between triggers, 549This doesn't mean there will always be 3600 seconds in between triggers,
522but only that the the callback will be called when the system time shows a 550but only that the the callback will be called when the system time shows a
523full hour (UTC), or more correct, when the system time is evenly divisible 551full hour (UTC), or more correctly, when the system time is evenly divisible
524by 3600. 552by 3600.
525 553
526Another way to think about it (for the mathematically inclined) is that 554Another way to think about it (for the mathematically inclined) is that
527ev_periodic will try to run the callback in this mode at the next possible 555C<ev_periodic> will try to run the callback in this mode at the next possible
528time where C<time = at (mod interval)>, regardless of any time jumps. 556time where C<time = at (mod interval)>, regardless of any time jumps.
529 557
530=item * manual reschedule mode (reschedule_cb = callback) 558=item * manual reschedule mode (reschedule_cb = callback)
531 559
532In this mode the values for C<interval> and C<at> are both being 560In this mode the values for C<interval> and C<at> are both being
533ignored. Instead, each time the periodic watcher gets scheduled, the 561ignored. Instead, each time the periodic watcher gets scheduled, the
534reschedule callback will be called with the watcher as first, and the 562reschedule callback will be called with the watcher as first, and the
535current time as second argument. 563current time as second argument.
536 564
537NOTE: I<This callback MUST NOT stop or destroy the periodic or any other 565NOTE: I<This callback MUST NOT stop or destroy the periodic or any other
538periodic watcher, ever, or make any event loop modificstions>. If you need 566periodic watcher, ever, or make any event loop modifications>. If you need
539to stop it, return 1e30 (or so, fudge fudge) and stop it afterwards. 567to stop it, return C<now + 1e30> (or so, fudge fudge) and stop it afterwards.
540 568
569Also, I<< this callback must always return a time that is later than the
570passed C<now> value >>. Not even C<now> itself will be ok.
571
541Its prototype is c<ev_tstamp (*reschedule_cb)(struct ev_periodic *w, 572Its prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
542ev_tstamp now)>, e.g.: 573ev_tstamp now)>, e.g.:
543 574
544 static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now) 575 static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
545 { 576 {
546 return now + 60.; 577 return now + 60.;
… …
565a different time than the last time it was called (e.g. in a crond like 596a different time than the last time it was called (e.g. in a crond like
566program when the crontabs have changed). 597program when the crontabs have changed).
567 598
568=back 599=back
569 600
570=head2 ev_signal - signal me when a signal gets signalled 601=head2 C<ev_signal> - signal me when a signal gets signalled
571 602
572Signal watchers will trigger an event when the process receives a specific 603Signal watchers will trigger an event when the process receives a specific
573signal one or more times. Even though signals are very asynchronous, libev 604signal one or more times. Even though signals are very asynchronous, libev
574will try its best to deliver signals synchronously, i.e. as part of the 605will try it's best to deliver signals synchronously, i.e. as part of the
575normal event processing, like any other event. 606normal event processing, like any other event.
576 607
577You cna configure as many watchers as you like per signal. Only when the 608You can configure as many watchers as you like per signal. Only when the
578first watcher gets started will libev actually register a signal watcher 609first watcher gets started will libev actually register a signal watcher
579with the kernel (thus it coexists with your own signal handlers as long 610with the kernel (thus it coexists with your own signal handlers as long
580as you don't register any with libev). Similarly, when the last signal 611as you don't register any with libev). Similarly, when the last signal
581watcher for a signal is stopped libev will reset the signal handler to 612watcher for a signal is stopped libev will reset the signal handler to
582SIG_DFL (regardless of what it was set to before). 613SIG_DFL (regardless of what it was set to before).
… …
590Configures the watcher to trigger on the given signal number (usually one 621Configures the watcher to trigger on the given signal number (usually one
591of the C<SIGxxx> constants). 622of the C<SIGxxx> constants).
592 623
593=back 624=back
594 625
595=head2 ev_child - wait for pid status changes 626=head2 C<ev_child> - wait for pid status changes
596 627
597Child watchers trigger when your process receives a SIGCHLD in response to 628Child watchers trigger when your process receives a SIGCHLD in response to
598some child status changes (most typically when a child of yours dies). 629some child status changes (most typically when a child of yours dies).
599 630
600=over 4 631=over 4
… …
604=item ev_child_set (ev_child *, int pid) 635=item ev_child_set (ev_child *, int pid)
605 636
606Configures the watcher to wait for status changes of process C<pid> (or 637Configures the watcher to wait for status changes of process C<pid> (or
607I<any> process if C<pid> is specified as C<0>). The callback can look 638I<any> process if C<pid> is specified as C<0>). The callback can look
608at the C<rstatus> member of the C<ev_child> watcher structure to see 639at the C<rstatus> member of the C<ev_child> watcher structure to see
609the status word (use the macros from C<sys/wait.h>). The C<rpid> member 640the status word (use the macros from C<sys/wait.h> and see your systems
610contains the pid of the process causing the status change. 641C<waitpid> documentation). The C<rpid> member contains the pid of the
642process causing the status change.
611 643
612=back 644=back
613 645
614=head2 ev_idle - when you've got nothing better to do 646=head2 C<ev_idle> - when you've got nothing better to do
615 647
616Idle watchers trigger events when there are no other I/O or timer (or 648Idle watchers trigger events when there are no other events are pending
617periodic) events pending. That is, as long as your process is busy 649(prepare, check and other idle watchers do not count). That is, as long
618handling sockets or timeouts it will not be called. But when your process 650as your process is busy handling sockets or timeouts (or even signals,
619is idle all idle watchers are being called again and again - until 651imagine) it will not be triggered. But when your process is idle all idle
652watchers are being called again and again, once per event loop iteration -
620stopped, that is, or your process receives more events. 653until stopped, that is, or your process receives more events and becomes
654busy.
621 655
622The most noteworthy effect is that as long as any idle watchers are 656The most noteworthy effect is that as long as any idle watchers are
623active, the process will not block when waiting for new events. 657active, the process will not block when waiting for new events.
624 658
625Apart from keeping your process non-blocking (which is a useful 659Apart from keeping your process non-blocking (which is a useful
… …
635kind. There is a C<ev_idle_set> macro, but using it is utterly pointless, 669kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
636believe me. 670believe me.
637 671
638=back 672=back
639 673
640=head2 prepare and check - your hooks into the event loop 674=head2 C<ev_prepare> and C<ev_check> - customise your event loop
641 675
642Prepare and check watchers usually (but not always) are used in 676Prepare and check watchers are usually (but not always) used in tandem:
643tandom. Prepare watchers get invoked before the process blocks and check 677Prepare watchers get invoked before the process blocks and check watchers
644watchers afterwards. 678afterwards.
645 679
646Their main purpose is to integrate other event mechanisms into libev. This 680Their main purpose is to integrate other event mechanisms into libev. This
647could be used, for example, to track variable changes, implement your own 681could be used, for example, to track variable changes, implement your own
648watchers, integrate net-snmp or a coroutine library and lots more. 682watchers, integrate net-snmp or a coroutine library and lots more.
649 683
650This is done by examining in each prepare call which file descriptors need 684This is done by examining in each prepare call which file descriptors need
651to be watched by the other library, registering ev_io watchers for them 685to be watched by the other library, registering C<ev_io> watchers for
652and starting an ev_timer watcher for any timeouts (many libraries provide 686them and starting an C<ev_timer> watcher for any timeouts (many libraries
653just this functionality). Then, in the check watcher you check for any 687provide just this functionality). Then, in the check watcher you check for
654events that occured (by making your callbacks set soem flags for example) 688any events that occured (by checking the pending status of all watchers
655and call back into the library. 689and stopping them) and call back into the library. The I/O and timer
690callbacks will never actually be called (but must be valid neverthelles,
691because you never know, you know?).
656 692
657As another example, the perl Coro module uses these hooks to integrate 693As another example, the Perl Coro module uses these hooks to integrate
658coroutines into libev programs, by yielding to other active coroutines 694coroutines into libev programs, by yielding to other active coroutines
659during each prepare and only letting the process block if no coroutines 695during each prepare and only letting the process block if no coroutines
660are ready to run. 696are ready to run (its actually more complicated, it only runs coroutines
697with priority higher than the event loop and one lower priority once,
698using idle watchers to keep the event loop from blocking if lower-priority
699coroutines exist, thus mapping low-priority coroutines to idle/background
700tasks).
661 701
662=over 4 702=over 4
663 703
664=item ev_prepare_init (ev_prepare *, callback) 704=item ev_prepare_init (ev_prepare *, callback)
665 705
666=item ev_check_init (ev_check *, callback) 706=item ev_check_init (ev_check *, callback)
667 707
668Initialises and configures the prepare or check watcher - they have no 708Initialises and configures the prepare or check watcher - they have no
669parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set> 709parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
670macros, but using them is utterly, utterly pointless. 710macros, but using them is utterly, utterly and completely pointless.
671 711
672=back 712=back
673 713
674=head1 OTHER FUNCTIONS 714=head1 OTHER FUNCTIONS
675 715
676There are some other fucntions of possible interest. Described. Here. Now. 716There are some other functions of possible interest. Described. Here. Now.
677 717
678=over 4 718=over 4
679 719
680=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback) 720=item ev_once (loop, int fd, int events, ev_tstamp timeout, callback)
681 721
… …
683callback on whichever event happens first and automatically stop both 723callback on whichever event happens first and automatically stop both
684watchers. This is useful if you want to wait for a single event on an fd 724watchers. This is useful if you want to wait for a single event on an fd
685or timeout without havign to allocate/configure/start/stop/free one or 725or timeout without havign to allocate/configure/start/stop/free one or
686more watchers yourself. 726more watchers yourself.
687 727
688If C<fd> is less than 0, then no I/O watcher will be started and events is 728If C<fd> is less than 0, then no I/O watcher will be started and events
689ignored. Otherwise, an ev_io watcher for the given C<fd> and C<events> set 729is being ignored. Otherwise, an C<ev_io> watcher for the given C<fd> and
690will be craeted and started. 730C<events> set will be craeted and started.
691 731
692If C<timeout> is less than 0, then no timeout watcher will be 732If C<timeout> is less than 0, then no timeout watcher will be
693started. Otherwise an ev_timer watcher with after = C<timeout> (and repeat 733started. Otherwise an C<ev_timer> watcher with after = C<timeout> (and
694= 0) will be started. 734repeat = 0) will be started. While C<0> is a valid timeout, it is of
735dubious value.
695 736
696The callback has the type C<void (*cb)(int revents, void *arg)> and 737The callback has the type C<void (*cb)(int revents, void *arg)> and gets
697gets passed an events set (normally a combination of EV_ERROR, EV_READ, 738passed an events set like normal event callbacks (with a combination of
698EV_WRITE or EV_TIMEOUT) and the C<arg> value passed to C<ev_once>: 739C<EV_ERROR>, C<EV_READ>, C<EV_WRITE> or C<EV_TIMEOUT>) and the C<arg>
740value passed to C<ev_once>:
699 741
700 static void stdin_ready (int revents, void *arg) 742 static void stdin_ready (int revents, void *arg)
701 { 743 {
702 if (revents & EV_TIMEOUT) 744 if (revents & EV_TIMEOUT)
703 /* doh, nothing entered */ 745 /* doh, nothing entered */;
704 else if (revents & EV_READ) 746 else if (revents & EV_READ)
705 /* stdin might have data for us, joy! */ 747 /* stdin might have data for us, joy! */;
706 } 748 }
707 749
708 ev_once (STDIN_FILENO, EV_READm 10., stdin_ready, 0); 750 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
709 751
710=item ev_feed_event (loop, watcher, int events) 752=item ev_feed_event (loop, watcher, int events)
711 753
712Feeds the given event set into the event loop, as if the specified event 754Feeds the given event set into the event loop, as if the specified event
713has happened for the specified watcher (which must be a pointer to an 755had happened for the specified watcher (which must be a pointer to an
714initialised but not necessarily active event watcher). 756initialised but not necessarily started event watcher).
715 757
716=item ev_feed_fd_event (loop, int fd, int revents) 758=item ev_feed_fd_event (loop, int fd, int revents)
717 759
718Feed an event on the given fd, as if a file descriptor backend detected it. 760Feed an event on the given fd, as if a file descriptor backend detected
761the given events it.
719 762
720=item ev_feed_signal_event (loop, int signum) 763=item ev_feed_signal_event (loop, int signum)
721 764
722Feed an event as if the given signal occured (loop must be the default loop!). 765Feed an event as if the given signal occured (loop must be the default loop!).
723 766

Diff Legend

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