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

Comparing libev/ev.pod (file contents):
Revision 1.29 by root, Thu Nov 22 12:28:28 2007 UTC vs.
Revision 1.35 by root, Fri Nov 23 19:35:09 2007 UTC

… …
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 C<double> type in C, and when you need to do any calculations on
51it, you should treat it as such.
52
51 53
52=head1 GLOBAL FUNCTIONS 54=head1 GLOBAL FUNCTIONS
53 55
54These functions can be called anytime, even before initialising the 56These functions can be called anytime, even before initialising the
55library in any way. 57library in any way.
… …
75Usually, it's a good idea to terminate if the major versions mismatch, 77Usually, it's a good idea to terminate if the major versions mismatch,
76as this indicates an incompatible change. Minor versions are usually 78as this indicates an incompatible change. Minor versions are usually
77compatible to older versions, so a larger minor version alone is usually 79compatible to older versions, so a larger minor version alone is usually
78not a problem. 80not a problem.
79 81
82Example: make sure we haven't accidentally been linked against the wrong
83version:
84
85 assert (("libev version mismatch",
86 ev_version_major () == EV_VERSION_MAJOR
87 && ev_version_minor () >= EV_VERSION_MINOR));
88
89=item unsigned int ev_supported_backends ()
90
91Return the set of all backends (i.e. their corresponding C<EV_BACKEND_*>
92value) compiled into this binary of libev (independent of their
93availability on the system you are running on). See C<ev_default_loop> for
94a description of the set values.
95
96Example: make sure we have the epoll method, because yeah this is cool and
97a must have and can we have a torrent of it please!!!11
98
99 assert (("sorry, no epoll, no sex",
100 ev_supported_backends () & EVBACKEND_EPOLL));
101
102=item unsigned int ev_recommended_backends ()
103
104Return the set of all backends compiled into this binary of libev and also
105recommended for this platform. This set is often smaller than the one
106returned by C<ev_supported_backends>, as for example kqueue is broken on
107most BSDs and will not be autodetected unless you explicitly request it
108(assuming you know what you are doing). This is the set of backends that
109libev will probe for if you specify no backends explicitly.
110
111=item unsigned int ev_embeddable_backends ()
112
113Returns the set of backends that are embeddable in other event loops. This
114is the theoretical, all-platform, value. To find which backends
115might be supported on the current system, you would need to look at
116C<ev_embeddable_backends () & ev_supported_backends ()>, likewise for
117recommended ones.
118
119See the description of C<ev_embed> watchers for more info.
120
80=item ev_set_allocator (void *(*cb)(void *ptr, long size)) 121=item ev_set_allocator (void *(*cb)(void *ptr, long size))
81 122
82Sets the allocation function to use (the prototype is similar to the 123Sets the allocation function to use (the prototype is similar to the
83realloc C function, the semantics are identical). It is used to allocate 124realloc C function, the semantics are identical). It is used to allocate
84and free memory (no surprises here). If it returns zero when memory 125and free memory (no surprises here). If it returns zero when memory
… …
86destructive action. The default is your system realloc function. 127destructive action. The default is your system realloc function.
87 128
88You could override this function in high-availability programs to, say, 129You could override this function in high-availability programs to, say,
89free some memory if it cannot allocate memory, to use a special allocator, 130free some memory if it cannot allocate memory, to use a special allocator,
90or even to sleep a while and retry until some memory is available. 131or even to sleep a while and retry until some memory is available.
132
133Example: replace the libev allocator with one that waits a bit and then
134retries: better than mine).
135
136 static void *
137 persistent_realloc (void *ptr, long size)
138 {
139 for (;;)
140 {
141 void *newptr = realloc (ptr, size);
142
143 if (newptr)
144 return newptr;
145
146 sleep (60);
147 }
148 }
149
150 ...
151 ev_set_allocator (persistent_realloc);
91 152
92=item ev_set_syserr_cb (void (*cb)(const char *msg)); 153=item ev_set_syserr_cb (void (*cb)(const char *msg));
93 154
94Set the callback function to call on a retryable syscall error (such 155Set the callback function to call on a retryable syscall error (such
95as failed select, poll, epoll_wait). The message is a printable string 156as failed select, poll, epoll_wait). The message is a printable string
… …
97callback is set, then libev will expect it to remedy the sitution, no 158callback is set, then libev will expect it to remedy the sitution, no
98matter what, when it returns. That is, libev will generally retry the 159matter what, when it returns. That is, libev will generally retry the
99requested operation, or, if the condition doesn't go away, do bad stuff 160requested operation, or, if the condition doesn't go away, do bad stuff
100(such as abort). 161(such as abort).
101 162
163Example: do the same thing as libev does internally:
164
165 static void
166 fatal_error (const char *msg)
167 {
168 perror (msg);
169 abort ();
170 }
171
172 ...
173 ev_set_syserr_cb (fatal_error);
174
102=back 175=back
103 176
104=head1 FUNCTIONS CONTROLLING THE EVENT LOOP 177=head1 FUNCTIONS CONTROLLING THE EVENT LOOP
105 178
106An event loop is described by a C<struct ev_loop *>. The library knows two 179An event loop is described by a C<struct ev_loop *>. The library knows two
… …
119=item struct ev_loop *ev_default_loop (unsigned int flags) 192=item struct ev_loop *ev_default_loop (unsigned int flags)
120 193
121This will initialise the default event loop if it hasn't been initialised 194This will initialise the default event loop if it hasn't been initialised
122yet and return it. If the default loop could not be initialised, returns 195yet and return it. If the default loop could not be initialised, returns
123false. If it already was initialised it simply returns it (and ignores the 196false. If it already was initialised it simply returns it (and ignores the
124flags). 197flags. If that is troubling you, check C<ev_backend ()> afterwards).
125 198
126If you don't know what event loop to use, use the one returned from this 199If you don't know what event loop to use, use the one returned from this
127function. 200function.
128 201
129The flags argument can be used to specify special behaviour or specific 202The flags argument can be used to specify special behaviour or specific
130backends to use, and is usually specified as 0 (or EVFLAG_AUTO). 203backends to use, and is usually specified as C<0> (or C<EVFLAG_AUTO>).
131 204
132It supports the following flags: 205The following flags are supported:
133 206
134=over 4 207=over 4
135 208
136=item C<EVFLAG_AUTO> 209=item C<EVFLAG_AUTO>
137 210
… …
145C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will 218C<LIBEV_FLAGS>. Otherwise (the default), this environment variable will
146override the flags completely if it is found in the environment. This is 219override the flags completely if it is found in the environment. This is
147useful to try out specific backends to test their performance, or to work 220useful to try out specific backends to test their performance, or to work
148around bugs. 221around bugs.
149 222
150=item C<EVMETHOD_SELECT> (value 1, portable select backend) 223=item C<EVBACKEND_SELECT> (value 1, portable select backend)
151 224
152This is your standard select(2) backend. Not I<completely> standard, as 225This is your standard select(2) backend. Not I<completely> standard, as
153libev tries to roll its own fd_set with no limits on the number of fds, 226libev tries to roll its own fd_set with no limits on the number of fds,
154but if that fails, expect a fairly low limit on the number of fds when 227but if that fails, expect a fairly low limit on the number of fds when
155using this backend. It doesn't scale too well (O(highest_fd)), but its usually 228using this backend. It doesn't scale too well (O(highest_fd)), but its usually
156the fastest backend for a low number of fds. 229the fastest backend for a low number of fds.
157 230
158=item C<EVMETHOD_POLL> (value 2, poll backend, available everywhere except on windows) 231=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
159 232
160And this is your standard poll(2) backend. It's more complicated than 233And this is your standard poll(2) backend. It's more complicated than
161select, but handles sparse fds better and has no artificial limit on the 234select, but handles sparse fds better and has no artificial limit on the
162number of fds you can use (except it will slow down considerably with a 235number of fds you can use (except it will slow down considerably with a
163lot of inactive fds). It scales similarly to select, i.e. O(total_fds). 236lot of inactive fds). It scales similarly to select, i.e. O(total_fds).
164 237
165=item C<EVMETHOD_EPOLL> (value 4, Linux) 238=item C<EVBACKEND_EPOLL> (value 4, Linux)
166 239
167For few fds, this backend is a bit little slower than poll and select, 240For few fds, this backend is a bit little slower than poll and select,
168but it scales phenomenally better. While poll and select usually scale like 241but it scales phenomenally better. While poll and select usually scale like
169O(total_fds) where n is the total number of fds (or the highest fd), epoll scales 242O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
170either O(1) or O(active_fds). 243either O(1) or O(active_fds).
… …
173result in some caching, there is still a syscall per such incident 246result in some caching, there is still a syscall per such incident
174(because the fd could point to a different file description now), so its 247(because the fd could point to a different file description now), so its
175best to avoid that. Also, dup()ed file descriptors might not work very 248best to avoid that. Also, dup()ed file descriptors might not work very
176well if you register events for both fds. 249well if you register events for both fds.
177 250
251Please note that epoll sometimes generates spurious notifications, so you
252need to use non-blocking I/O or other means to avoid blocking when no data
253(or space) is available.
254
178=item C<EVMETHOD_KQUEUE> (value 8, most BSD clones) 255=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
179 256
180Kqueue deserves special mention, as at the time of this writing, it 257Kqueue deserves special mention, as at the time of this writing, it
181was broken on all BSDs except NetBSD (usually it doesn't work with 258was broken on all BSDs except NetBSD (usually it doesn't work with
182anything but sockets and pipes, except on Darwin, where of course its 259anything but sockets and pipes, except on Darwin, where of course its
183completely useless). For this reason its not being "autodetected" unless 260completely useless). For this reason its not being "autodetected"
184you explicitly specify the flags (i.e. you don't use EVFLAG_AUTO). 261unless you explicitly specify it explicitly in the flags (i.e. using
262C<EVBACKEND_KQUEUE>).
185 263
186It scales in the same way as the epoll backend, but the interface to the 264It scales in the same way as the epoll backend, but the interface to the
187kernel is more efficient (which says nothing about its actual speed, of 265kernel is more efficient (which says nothing about its actual speed, of
188course). While starting and stopping an I/O watcher does not cause an 266course). While starting and stopping an I/O watcher does not cause an
189extra syscall as with epoll, it still adds up to four event changes per 267extra syscall as with epoll, it still adds up to four event changes per
190incident, so its best to avoid that. 268incident, so its best to avoid that.
191 269
192=item C<EVMETHOD_DEVPOLL> (value 16, Solaris 8) 270=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
193 271
194This is not implemented yet (and might never be). 272This is not implemented yet (and might never be).
195 273
196=item C<EVMETHOD_PORT> (value 32, Solaris 10) 274=item C<EVBACKEND_PORT> (value 32, Solaris 10)
197 275
198This uses the Solaris 10 port mechanism. As with everything on Solaris, 276This uses the Solaris 10 port mechanism. As with everything on Solaris,
199it's really slow, but it still scales very well (O(active_fds)). 277it's really slow, but it still scales very well (O(active_fds)).
200 278
279Please note that solaris ports can result in a lot of spurious
280notifications, so you need to use non-blocking I/O or other means to avoid
281blocking when no data (or space) is available.
282
201=item C<EVMETHOD_ALL> 283=item C<EVBACKEND_ALL>
202 284
203Try all backends (even potentially broken ones that wouldn't be tried 285Try all backends (even potentially broken ones that wouldn't be tried
204with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as 286with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
205C<EVMETHOD_ALL & ~EVMETHOD_KQUEUE>. 287C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
206 288
207=back 289=back
208 290
209If one or more of these are ored into the flags value, then only these 291If one or more of these are ored into the flags value, then only these
210backends will be tried (in the reverse order as given here). If none are 292backends will be tried (in the reverse order as given here). If none are
211specified, most compiled-in backend will be tried, usually in reverse 293specified, most compiled-in backend will be tried, usually in reverse
212order of their flag values :) 294order of their flag values :)
213 295
296The most typical usage is like this:
297
298 if (!ev_default_loop (0))
299 fatal ("could not initialise libev, bad $LIBEV_FLAGS in environment?");
300
301Restrict libev to the select and poll backends, and do not allow
302environment settings to be taken into account:
303
304 ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
305
306Use whatever libev has to offer, but make sure that kqueue is used if
307available (warning, breaks stuff, best use only with your own private
308event loop and only if you know the OS supports your types of fds):
309
310 ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
311
214=item struct ev_loop *ev_loop_new (unsigned int flags) 312=item struct ev_loop *ev_loop_new (unsigned int flags)
215 313
216Similar to C<ev_default_loop>, but always creates a new event loop that is 314Similar to C<ev_default_loop>, but always creates a new event loop that is
217always distinct from the default loop. Unlike the default loop, it cannot 315always distinct from the default loop. Unlike the default loop, it cannot
218handle signal and child watchers, and attempts to do so will be greeted by 316handle signal and child watchers, and attempts to do so will be greeted by
219undefined behaviour (or a failed assertion if assertions are enabled). 317undefined behaviour (or a failed assertion if assertions are enabled).
220 318
319Example: try to create a event loop that uses epoll and nothing else.
320
321 struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
322 if (!epoller)
323 fatal ("no epoll found here, maybe it hides under your chair");
324
221=item ev_default_destroy () 325=item ev_default_destroy ()
222 326
223Destroys the default loop again (frees all memory and kernel state 327Destroys the default loop again (frees all memory and kernel state
224etc.). This stops all registered event watchers (by not touching them in 328etc.). This stops all registered event watchers (by not touching them in
225any way whatsoever, although you cannot rely on this :). 329any way whatsoever, although you cannot rely on this :).
… …
234This function reinitialises the kernel state for backends that have 338This function reinitialises the kernel state for backends that have
235one. Despite the name, you can call it anytime, but it makes most sense 339one. Despite the name, you can call it anytime, but it makes most sense
236after forking, in either the parent or child process (or both, but that 340after forking, in either the parent or child process (or both, but that
237again makes little sense). 341again makes little sense).
238 342
239You I<must> call this function after forking if and only if you want to 343You I<must> call this function in the child process after forking if and
240use the event library in both processes. If you just fork+exec, you don't 344only if you want to use the event library in both processes. If you just
241have to call it. 345fork+exec, you don't have to call it.
242 346
243The function itself is quite fast and it's usually not a problem to call 347The function itself is quite fast and it's usually not a problem to call
244it just in case after a fork. To make this easy, the function will fit in 348it just in case after a fork. To make this easy, the function will fit in
245quite nicely into a call to C<pthread_atfork>: 349quite nicely into a call to C<pthread_atfork>:
246 350
247 pthread_atfork (0, 0, ev_default_fork); 351 pthread_atfork (0, 0, ev_default_fork);
248 352
353At the moment, C<EVBACKEND_SELECT> and C<EVBACKEND_POLL> are safe to use
354without calling this function, so if you force one of those backends you
355do not need to care.
356
249=item ev_loop_fork (loop) 357=item ev_loop_fork (loop)
250 358
251Like C<ev_default_fork>, but acts on an event loop created by 359Like C<ev_default_fork>, but acts on an event loop created by
252C<ev_loop_new>. Yes, you have to call this on every allocated event loop 360C<ev_loop_new>. Yes, you have to call this on every allocated event loop
253after fork, and how you do this is entirely your own problem. 361after fork, and how you do this is entirely your own problem.
254 362
255=item unsigned int ev_method (loop) 363=item unsigned int ev_backend (loop)
256 364
257Returns one of the C<EVMETHOD_*> flags indicating the event backend in 365Returns one of the C<EVBACKEND_*> flags indicating the event backend in
258use. 366use.
259 367
260=item ev_tstamp ev_now (loop) 368=item ev_tstamp ev_now (loop)
261 369
262Returns the current "event loop time", which is the time the event loop 370Returns the current "event loop time", which is the time the event loop
263got events and started processing them. This timestamp does not change 371received events and started processing them. This timestamp does not
264as long as callbacks are being processed, and this is also the base time 372change as long as callbacks are being processed, and this is also the base
265used for relative timers. You can treat it as the timestamp of the event 373time used for relative timers. You can treat it as the timestamp of the
266occuring (or more correctly, the mainloop finding out about it). 374event occuring (or more correctly, libev finding out about it).
267 375
268=item ev_loop (loop, int flags) 376=item ev_loop (loop, int flags)
269 377
270Finally, this is it, the event handler. This function usually is called 378Finally, this is it, the event handler. This function usually is called
271after you initialised all your watchers and you want to start handling 379after you initialised all your watchers and you want to start handling
272events. 380events.
273 381
274If the flags argument is specified as 0, it will not return until either 382If the flags argument is specified as C<0>, it will not return until
275no event watchers are active anymore or C<ev_unloop> was called. 383either no event watchers are active anymore or C<ev_unloop> was called.
384
385Please note that an explicit C<ev_unloop> is usually better than
386relying on all watchers to be stopped when deciding when a program has
387finished (especially in interactive programs), but having a program that
388automatically loops as long as it has to and no longer by virtue of
389relying on its watchers stopping correctly is a thing of beauty.
276 390
277A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle 391A flags value of C<EVLOOP_NONBLOCK> will look for new events, will handle
278those events and any outstanding ones, but will not block your process in 392those events and any outstanding ones, but will not block your process in
279case there are no events and will return after one iteration of the loop. 393case there are no events and will return after one iteration of the loop.
280 394
281A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if 395A flags value of C<EVLOOP_ONESHOT> will look for new events (waiting if
282neccessary) and will handle those and any outstanding ones. It will block 396neccessary) and will handle those and any outstanding ones. It will block
283your process until at least one new event arrives, and will return after 397your process until at least one new event arrives, and will return after
284one iteration of the loop. 398one iteration of the loop. This is useful if you are waiting for some
399external event in conjunction with something not expressible using other
400libev watchers. However, a pair of C<ev_prepare>/C<ev_check> watchers is
401usually a better approach for this kind of thing.
285 402
286This flags value could be used to implement alternative looping
287constructs, but the C<prepare> and C<check> watchers provide a better and
288more generic mechanism.
289
290Here are the gory details of what ev_loop does: 403Here are the gory details of what C<ev_loop> does:
291 404
292 1. If there are no active watchers (reference count is zero), return. 405 * If there are no active watchers (reference count is zero), return.
293 2. Queue and immediately call all prepare watchers. 406 - Queue prepare watchers and then call all outstanding watchers.
294 3. If we have been forked, recreate the kernel state. 407 - If we have been forked, recreate the kernel state.
295 4. Update the kernel state with all outstanding changes. 408 - Update the kernel state with all outstanding changes.
296 5. Update the "event loop time". 409 - Update the "event loop time".
297 6. Calculate for how long to block. 410 - Calculate for how long to block.
298 7. Block the process, waiting for events. 411 - Block the process, waiting for any events.
412 - Queue all outstanding I/O (fd) events.
299 8. Update the "event loop time" and do time jump handling. 413 - Update the "event loop time" and do time jump handling.
300 9. Queue all outstanding timers. 414 - Queue all outstanding timers.
301 10. Queue all outstanding periodics. 415 - Queue all outstanding periodics.
302 11. If no events are pending now, queue all idle watchers. 416 - If no events are pending now, queue all idle watchers.
303 12. Queue all check watchers. 417 - Queue all check watchers.
304 13. Call all queued watchers in reverse order (i.e. check watchers first). 418 - Call all queued watchers in reverse order (i.e. check watchers first).
419 Signals and child watchers are implemented as I/O watchers, and will
420 be handled here by queueing them when their watcher gets executed.
305 14. If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK 421 - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
306 was used, return, otherwise continue with step #1. 422 were used, return, otherwise continue with step *.
423
424Example: queue some jobs and then loop until no events are outsanding
425anymore.
426
427 ... queue jobs here, make sure they register event watchers as long
428 ... as they still have work to do (even an idle watcher will do..)
429 ev_loop (my_loop, 0);
430 ... jobs done. yeah!
307 431
308=item ev_unloop (loop, how) 432=item ev_unloop (loop, how)
309 433
310Can be used to make a call to C<ev_loop> return early (but only after it 434Can be used to make a call to C<ev_loop> return early (but only after it
311has processed all outstanding events). The C<how> argument must be either 435has processed all outstanding events). The C<how> argument must be either
… …
325visible to the libev user and should not keep C<ev_loop> from exiting if 449visible to the libev user and should not keep C<ev_loop> from exiting if
326no event watchers registered by it are active. It is also an excellent 450no event watchers registered by it are active. It is also an excellent
327way to do this for generic recurring timers or from within third-party 451way to do this for generic recurring timers or from within third-party
328libraries. Just remember to I<unref after start> and I<ref before stop>. 452libraries. Just remember to I<unref after start> and I<ref before stop>.
329 453
454Example: create a signal watcher, but keep it from keeping C<ev_loop>
455running when nothing else is active.
456
457 struct dv_signal exitsig;
458 ev_signal_init (&exitsig, sig_cb, SIGINT);
459 ev_signal_start (myloop, &exitsig);
460 evf_unref (myloop);
461
462Example: for some weird reason, unregister the above signal handler again.
463
464 ev_ref (myloop);
465 ev_signal_stop (myloop, &exitsig);
466
330=back 467=back
331 468
332=head1 ANATOMY OF A WATCHER 469=head1 ANATOMY OF A WATCHER
333 470
334A watcher is a structure that you create and register to record your 471A watcher is a structure that you create and register to record your
… …
368*) >>), and you can stop watching for events at any time by calling the 505*) >>), and you can stop watching for events at any time by calling the
369corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>. 506corresponding stop function (C<< ev_<type>_stop (loop, watcher *) >>.
370 507
371As long as your watcher is active (has been started but not stopped) you 508As long as your watcher is active (has been started but not stopped) you
372must not touch the values stored in it. Most specifically you must never 509must not touch the values stored in it. Most specifically you must never
373reinitialise it or call its set method. 510reinitialise it or call its set macro.
374 511
375You can check whether an event is active by calling the C<ev_is_active 512You can check whether an event is active by calling the C<ev_is_active
376(watcher *)> macro. To see whether an event is outstanding (but the 513(watcher *)> macro. To see whether an event is outstanding (but the
377callback for it has not been called yet) you can use the C<ev_is_pending 514callback for it has not been called yet) you can use the C<ev_is_pending
378(watcher *)> macro. 515(watcher *)> macro.
… …
475=head1 WATCHER TYPES 612=head1 WATCHER TYPES
476 613
477This section describes each watcher in detail, but will not repeat 614This section describes each watcher in detail, but will not repeat
478information given in the last section. 615information given in the last section.
479 616
617
480=head2 C<ev_io> - is this file descriptor readable or writable 618=head2 C<ev_io> - is this file descriptor readable or writable
481 619
482I/O watchers check whether a file descriptor is readable or writable 620I/O watchers check whether a file descriptor is readable or writable
483in each iteration of the event loop (This behaviour is called 621in each iteration of the event loop (This behaviour is called
484level-triggering because you keep receiving events as long as the 622level-triggering because you keep receiving events as long as the
… …
495descriptors correctly if you register interest in two or more fds pointing 633descriptors correctly if you register interest in two or more fds pointing
496to the same underlying file/socket etc. description (that is, they share 634to the same underlying file/socket etc. description (that is, they share
497the same underlying "file open"). 635the same underlying "file open").
498 636
499If you must do this, then force the use of a known-to-be-good backend 637If you must do this, then force the use of a known-to-be-good backend
500(at the time of this writing, this includes only EVMETHOD_SELECT and 638(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
501EVMETHOD_POLL). 639C<EVBACKEND_POLL>).
502 640
503=over 4 641=over 4
504 642
505=item ev_io_init (ev_io *, callback, int fd, int events) 643=item ev_io_init (ev_io *, callback, int fd, int events)
506 644
… …
508 646
509Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive 647Configures an C<ev_io> watcher. The fd is the file descriptor to rceeive
510events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ | 648events for and events is either C<EV_READ>, C<EV_WRITE> or C<EV_READ |
511EV_WRITE> to receive the given events. 649EV_WRITE> to receive the given events.
512 650
651Please note that most of the more scalable backend mechanisms (for example
652epoll and solaris ports) can result in spurious readyness notifications
653for file descriptors, so you practically need to use non-blocking I/O (and
654treat callback invocation as hint only), or retest separately with a safe
655interface before doing I/O (XLib can do this), or force the use of either
656C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>, which don't suffer from this
657problem. Also note that it is quite easy to have your callback invoked
658when the readyness condition is no longer valid even when employing
659typical ways of handling events, so its a good idea to use non-blocking
660I/O unconditionally.
661
513=back 662=back
663
664Example: call C<stdin_readable_cb> when STDIN_FILENO has become, well
665readable, but only once. Since it is likely line-buffered, you could
666attempt to read a whole line in the callback:
667
668 static void
669 stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
670 {
671 ev_io_stop (loop, w);
672 .. read from stdin here (or from w->fd) and haqndle any I/O errors
673 }
674
675 ...
676 struct ev_loop *loop = ev_default_init (0);
677 struct ev_io stdin_readable;
678 ev_io_init (&stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
679 ev_io_start (loop, &stdin_readable);
680 ev_loop (loop, 0);
681
514 682
515=head2 C<ev_timer> - relative and optionally recurring timeouts 683=head2 C<ev_timer> - relative and optionally recurring timeouts
516 684
517Timer watchers are simple relative timers that generate an event after a 685Timer watchers are simple relative timers that generate an event after a
518given time, and optionally repeating in regular intervals after that. 686given time, and optionally repeating in regular intervals after that.
… …
571state where you do not expect data to travel on the socket, you can stop 739state where you do not expect data to travel on the socket, you can stop
572the timer, and again will automatically restart it if need be. 740the timer, and again will automatically restart it if need be.
573 741
574=back 742=back
575 743
744Example: create a timer that fires after 60 seconds.
745
746 static void
747 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
748 {
749 .. one minute over, w is actually stopped right here
750 }
751
752 struct ev_timer mytimer;
753 ev_timer_init (&mytimer, one_minute_cb, 60., 0.);
754 ev_timer_start (loop, &mytimer);
755
756Example: create a timeout timer that times out after 10 seconds of
757inactivity.
758
759 static void
760 timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
761 {
762 .. ten seconds without any activity
763 }
764
765 struct ev_timer mytimer;
766 ev_timer_init (&mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
767 ev_timer_again (&mytimer); /* start timer */
768 ev_loop (loop, 0);
769
770 // and in some piece of code that gets executed on any "activity":
771 // reset the timeout to start ticking again at 10 seconds
772 ev_timer_again (&mytimer);
773
774
576=head2 C<ev_periodic> - to cron or not to cron 775=head2 C<ev_periodic> - to cron or not to cron
577 776
578Periodic watchers are also timers of a kind, but they are very versatile 777Periodic watchers are also timers of a kind, but they are very versatile
579(and unfortunately a bit complex). 778(and unfortunately a bit complex).
580 779
… …
675a different time than the last time it was called (e.g. in a crond like 874a different time than the last time it was called (e.g. in a crond like
676program when the crontabs have changed). 875program when the crontabs have changed).
677 876
678=back 877=back
679 878
879Example: call a callback every hour, or, more precisely, whenever the
880system clock is divisible by 3600. The callback invocation times have
881potentially a lot of jittering, but good long-term stability.
882
883 static void
884 clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
885 {
886 ... its now a full hour (UTC, or TAI or whatever your clock follows)
887 }
888
889 struct ev_periodic hourly_tick;
890 ev_periodic_init (&hourly_tick, clock_cb, 0., 3600., 0);
891 ev_periodic_start (loop, &hourly_tick);
892
893Example: the same as above, but use a reschedule callback to do it:
894
895 #include <math.h>
896
897 static ev_tstamp
898 my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
899 {
900 return fmod (now, 3600.) + 3600.;
901 }
902
903 ev_periodic_init (&hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
904
905Example: call a callback every hour, starting now:
906
907 struct ev_periodic hourly_tick;
908 ev_periodic_init (&hourly_tick, clock_cb,
909 fmod (ev_now (loop), 3600.), 3600., 0);
910 ev_periodic_start (loop, &hourly_tick);
911
912
680=head2 C<ev_signal> - signal me when a signal gets signalled 913=head2 C<ev_signal> - signal me when a signal gets signalled
681 914
682Signal watchers will trigger an event when the process receives a specific 915Signal watchers will trigger an event when the process receives a specific
683signal one or more times. Even though signals are very asynchronous, libev 916signal one or more times. Even though signals are very asynchronous, libev
684will try it's best to deliver signals synchronously, i.e. as part of the 917will try it's best to deliver signals synchronously, i.e. as part of the
… …
700Configures the watcher to trigger on the given signal number (usually one 933Configures the watcher to trigger on the given signal number (usually one
701of the C<SIGxxx> constants). 934of the C<SIGxxx> constants).
702 935
703=back 936=back
704 937
938
705=head2 C<ev_child> - wait for pid status changes 939=head2 C<ev_child> - wait for pid status changes
706 940
707Child watchers trigger when your process receives a SIGCHLD in response to 941Child watchers trigger when your process receives a SIGCHLD in response to
708some child status changes (most typically when a child of yours dies). 942some child status changes (most typically when a child of yours dies).
709 943
… …
719the status word (use the macros from C<sys/wait.h> and see your systems 953the status word (use the macros from C<sys/wait.h> and see your systems
720C<waitpid> documentation). The C<rpid> member contains the pid of the 954C<waitpid> documentation). The C<rpid> member contains the pid of the
721process causing the status change. 955process causing the status change.
722 956
723=back 957=back
958
959Example: try to exit cleanly on SIGINT and SIGTERM.
960
961 static void
962 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
963 {
964 ev_unloop (loop, EVUNLOOP_ALL);
965 }
966
967 struct ev_signal signal_watcher;
968 ev_signal_init (&signal_watcher, sigint_cb, SIGINT);
969 ev_signal_start (loop, &sigint_cb);
970
724 971
725=head2 C<ev_idle> - when you've got nothing better to do 972=head2 C<ev_idle> - when you've got nothing better to do
726 973
727Idle watchers trigger events when there are no other events are pending 974Idle watchers trigger events when there are no other events are pending
728(prepare, check and other idle watchers do not count). That is, as long 975(prepare, check and other idle watchers do not count). That is, as long
… …
748kind. There is a C<ev_idle_set> macro, but using it is utterly pointless, 995kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
749believe me. 996believe me.
750 997
751=back 998=back
752 999
1000Example: dynamically allocate an C<ev_idle>, start it, and in the
1001callback, free it. Alos, use no error checking, as usual.
1002
1003 static void
1004 idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1005 {
1006 free (w);
1007 // now do something you wanted to do when the program has
1008 // no longer asnything immediate to do.
1009 }
1010
1011 struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1012 ev_idle_init (idle_watcher, idle_cb);
1013 ev_idle_start (loop, idle_cb);
1014
1015
753=head2 C<ev_prepare> and C<ev_check> - customise your event loop 1016=head2 C<ev_prepare> and C<ev_check> - customise your event loop
754 1017
755Prepare and check watchers are usually (but not always) used in tandem: 1018Prepare and check watchers are usually (but not always) used in tandem:
756prepare watchers get invoked before the process blocks and check watchers 1019prepare watchers get invoked before the process blocks and check watchers
757afterwards. 1020afterwards.
758 1021
759Their main purpose is to integrate other event mechanisms into libev. This 1022Their main purpose is to integrate other event mechanisms into libev and
760could be used, for example, to track variable changes, implement your own 1023their use is somewhat advanced. This could be used, for example, to track
761watchers, integrate net-snmp or a coroutine library and lots more. 1024variable changes, implement your own watchers, integrate net-snmp or a
1025coroutine library and lots more.
762 1026
763This is done by examining in each prepare call which file descriptors need 1027This is done by examining in each prepare call which file descriptors need
764to be watched by the other library, registering C<ev_io> watchers for 1028to be watched by the other library, registering C<ev_io> watchers for
765them and starting an C<ev_timer> watcher for any timeouts (many libraries 1029them and starting an C<ev_timer> watcher for any timeouts (many libraries
766provide just this functionality). Then, in the check watcher you check for 1030provide just this functionality). Then, in the check watcher you check for
… …
788parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set> 1052parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
789macros, but using them is utterly, utterly and completely pointless. 1053macros, but using them is utterly, utterly and completely pointless.
790 1054
791=back 1055=back
792 1056
1057Example: *TODO*.
1058
1059
1060=head2 C<ev_embed> - when one backend isn't enough
1061
1062This is a rather advanced watcher type that lets you embed one event loop
1063into another.
1064
1065There are primarily two reasons you would want that: work around bugs and
1066prioritise I/O.
1067
1068As an example for a bug workaround, the kqueue backend might only support
1069sockets on some platform, so it is unusable as generic backend, but you
1070still want to make use of it because you have many sockets and it scales
1071so nicely. In this case, you would create a kqueue-based loop and embed it
1072into your default loop (which might use e.g. poll). Overall operation will
1073be a bit slower because first libev has to poll and then call kevent, but
1074at least you can use both at what they are best.
1075
1076As for prioritising I/O: rarely you have the case where some fds have
1077to be watched and handled very quickly (with low latency), and even
1078priorities and idle watchers might have too much overhead. In this case
1079you would put all the high priority stuff in one loop and all the rest in
1080a second one, and embed the second one in the first.
1081
1082As long as the watcher is started it will automatically handle events. The
1083callback will be invoked whenever some events have been handled. You can
1084set the callback to C<0> to avoid having to specify one if you are not
1085interested in that.
1086
1087Also, there have not currently been made special provisions for forking:
1088when you fork, you not only have to call C<ev_loop_fork> on both loops,
1089but you will also have to stop and restart any C<ev_embed> watchers
1090yourself.
1091
1092Unfortunately, not all backends are embeddable, only the ones returned by
1093C<ev_embeddable_backends> are, which, unfortunately, does not include any
1094portable one.
1095
1096So when you want to use this feature you will always have to be prepared
1097that you cannot get an embeddable loop. The recommended way to get around
1098this is to have a separate variables for your embeddable loop, try to
1099create it, and if that fails, use the normal loop for everything:
1100
1101 struct ev_loop *loop_hi = ev_default_init (0);
1102 struct ev_loop *loop_lo = 0;
1103 struct ev_embed embed;
1104
1105 // see if there is a chance of getting one that works
1106 // (remember that a flags value of 0 means autodetection)
1107 loop_lo = ev_embeddable_backends () & ev_recommended_backends ()
1108 ? ev_loop_new (ev_embeddable_backends () & ev_recommended_backends ())
1109 : 0;
1110
1111 // if we got one, then embed it, otherwise default to loop_hi
1112 if (loop_lo)
1113 {
1114 ev_embed_init (&embed, 0, loop_lo);
1115 ev_embed_start (loop_hi, &embed);
1116 }
1117 else
1118 loop_lo = loop_hi;
1119
1120=over 4
1121
1122=item ev_embed_init (ev_embed *, callback, struct ev_loop *loop)
1123
1124=item ev_embed_set (ev_embed *, callback, struct ev_loop *loop)
1125
1126Configures the watcher to embed the given loop, which must be embeddable.
1127
1128=back
1129
1130
793=head1 OTHER FUNCTIONS 1131=head1 OTHER FUNCTIONS
794 1132
795There are some other functions of possible interest. Described. Here. Now. 1133There are some other functions of possible interest. Described. Here. Now.
796 1134
797=over 4 1135=over 4
… …
843 1181
844Feed an event as if the given signal occured (loop must be the default loop!). 1182Feed an event as if the given signal occured (loop must be the default loop!).
845 1183
846=back 1184=back
847 1185
1186
848=head1 LIBEVENT EMULATION 1187=head1 LIBEVENT EMULATION
849 1188
850Libev offers a compatibility emulation layer for libevent. It cannot 1189Libev offers a compatibility emulation layer for libevent. It cannot
851emulate the internals of libevent, so here are some usage hints: 1190emulate the internals of libevent, so here are some usage hints:
852 1191

Diff Legend

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