[ViewVC] Diff of: cvs/libev/ev.pod

Comparing libev/ev.pod (file contents):
Revision 1.143 by root, Sun Apr 6 14:34:52 2008 UTC vs.
Revision 1.157 by root, Tue May 20 23:49:41 2008 UTC

…		…
64		64
65	=head1 DESCRIPTION	65	=head1 DESCRIPTION
66		66
67	The newest version of this document is also available as an html-formatted	67	The newest version of this document is also available as an html-formatted
68	web page you might find easier to navigate when reading it for the first	68	web page you might find easier to navigate when reading it for the first
69	time: L<http://cvs.schmorp.de/libev/ev.html>.	69	time: L<http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod>.
70		70
71	Libev is an event loop: you register interest in certain events (such as a	71	Libev is an event loop: you register interest in certain events (such as a
72	file descriptor being readable or a timeout occurring), and it will manage	72	file descriptor being readable or a timeout occurring), and it will manage
73	these event sources and provide your program with events.	73	these event sources and provide your program with events.
74		74
…		…
196	See the description of C<ev_embed> watchers for more info.	196	See the description of C<ev_embed> watchers for more info.
197		197
198	=item ev_set_allocator (void (cb)(void *ptr, long size))	198	=item ev_set_allocator (void (cb)(void *ptr, long size))
199		199
200	Sets the allocation function to use (the prototype is similar - the	200	Sets the allocation function to use (the prototype is similar - the
201	semantics is identical - to the realloc C function). It is used to	201	semantics are identical to the C<realloc> C89/SuS/POSIX function). It is
202	allocate and free memory (no surprises here). If it returns zero when	202	used to allocate and free memory (no surprises here). If it returns zero
203	memory needs to be allocated, the library might abort or take some	203	when memory needs to be allocated (C<size != 0>), the library might abort
204	potentially destructive action. The default is your system realloc	204	or take some potentially destructive action.
205	function.	205
		206	Since some systems (at least OpenBSD and Darwin) fail to implement
		207	correct C<realloc> semantics, libev will use a wrapper around the system
		208	C<realloc> and C<free> functions by default.
206		209
207	You could override this function in high-availability programs to, say,	210	You could override this function in high-availability programs to, say,
208	free some memory if it cannot allocate memory, to use a special allocator,	211	free some memory if it cannot allocate memory, to use a special allocator,
209	or even to sleep a while and retry until some memory is available.	212	or even to sleep a while and retry until some memory is available.
210		213
211	Example: Replace the libev allocator with one that waits a bit and then	214	Example: Replace the libev allocator with one that waits a bit and then
212	retries).	215	retries (example requires a standards-compliant C<realloc>).
213		216
214	static void *	217	static void *
215	persistent_realloc (void *ptr, size_t size)	218	persistent_realloc (void *ptr, size_t size)
216	{	219	{
217	for (;;)	220	for (;;)
…		…
256		259
257	An event loop is described by a C<struct ev_loop *>. The library knows two	260	An event loop is described by a C<struct ev_loop *>. The library knows two
258	types of such loops, the I<default> loop, which supports signals and child	261	types of such loops, the I<default> loop, which supports signals and child
259	events, and dynamically created loops which do not.	262	events, and dynamically created loops which do not.
260		263
261	If you use threads, a common model is to run the default event loop
262	in your main thread (or in a separate thread) and for each thread you
263	create, you also create another event loop. Libev itself does no locking
264	whatsoever, so if you mix calls to the same event loop in different
265	threads, make sure you lock (this is usually a bad idea, though, even if
266	done correctly, because it's hideous and inefficient).
267
268	=over 4	264	=over 4
269		265
270	=item struct ev_loop *ev_default_loop (unsigned int flags)	266	=item struct ev_loop *ev_default_loop (unsigned int flags)
271		267
272	This will initialise the default event loop if it hasn't been initialised	268	This will initialise the default event loop if it hasn't been initialised
…		…
340	To get good performance out of this backend you need a high amount of	336	To get good performance out of this backend you need a high amount of
341	parallelity (most of the file descriptors should be busy). If you are	337	parallelity (most of the file descriptors should be busy). If you are
342	writing a server, you should C<accept ()> in a loop to accept as many	338	writing a server, you should C<accept ()> in a loop to accept as many
343	connections as possible during one iteration. You might also want to have	339	connections as possible during one iteration. You might also want to have
344	a look at C<ev_set_io_collect_interval ()> to increase the amount of	340	a look at C<ev_set_io_collect_interval ()> to increase the amount of
345	readyness notifications you get per iteration.	341	readiness notifications you get per iteration.
346		342
347	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)	343	=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
348		344
349	And this is your standard poll(2) backend. It's more complicated	345	And this is your standard poll(2) backend. It's more complicated
350	than select, but handles sparse fds better and has no artificial	346	than select, but handles sparse fds better and has no artificial
…		…
429	While this backend scales well, it requires one system call per active	425	While this backend scales well, it requires one system call per active
430	file descriptor per loop iteration. For small and medium numbers of file	426	file descriptor per loop iteration. For small and medium numbers of file
431	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend	427	descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
432	might perform better.	428	might perform better.
433		429
434	On the positive side, ignoring the spurious readyness notifications, this	430	On the positive side, ignoring the spurious readiness notifications, this
435	backend actually performed to specification in all tests and is fully	431	backend actually performed to specification in all tests and is fully
436	embeddable, which is a rare feat among the OS-specific backends.	432	embeddable, which is a rare feat among the OS-specific backends.
437		433
438	=item C<EVBACKEND_ALL>	434	=item C<EVBACKEND_ALL>
439		435
…		…
1036	If you must do this, then force the use of a known-to-be-good backend	1032	If you must do this, then force the use of a known-to-be-good backend
1037	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and	1033	(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
1038	C<EVBACKEND_POLL>).	1034	C<EVBACKEND_POLL>).
1039		1035
1040	Another thing you have to watch out for is that it is quite easy to	1036	Another thing you have to watch out for is that it is quite easy to
1041	receive "spurious" readyness notifications, that is your callback might	1037	receive "spurious" readiness notifications, that is your callback might
1042	be called with C<EV_READ> but a subsequent C<read>(2) will actually block	1038	be called with C<EV_READ> but a subsequent C<read>(2) will actually block
1043	because there is no data. Not only are some backends known to create a	1039	because there is no data. Not only are some backends known to create a
1044	lot of those (for example solaris ports), it is very easy to get into	1040	lot of those (for example solaris ports), it is very easy to get into
1045	this situation even with a relatively standard program structure. Thus	1041	this situation even with a relatively standard program structure. Thus
1046	it is best to always use non-blocking I/O: An extra C<read>(2) returning	1042	it is best to always use non-blocking I/O: An extra C<read>(2) returning
…		…
1155		1151
1156	Timer watchers are simple relative timers that generate an event after a	1152	Timer watchers are simple relative timers that generate an event after a
1157	given time, and optionally repeating in regular intervals after that.	1153	given time, and optionally repeating in regular intervals after that.
1158		1154
1159	The timers are based on real time, that is, if you register an event that	1155	The timers are based on real time, that is, if you register an event that
1160	times out after an hour and you reset your system clock to last years	1156	times out after an hour and you reset your system clock to january last
1161	time, it will still time out after (roughly) and hour. "Roughly" because	1157	year, it will still time out after (roughly) and hour. "Roughly" because
1162	detecting time jumps is hard, and some inaccuracies are unavoidable (the	1158	detecting time jumps is hard, and some inaccuracies are unavoidable (the
1163	monotonic clock option helps a lot here).	1159	monotonic clock option helps a lot here).
1164		1160
1165	The relative timeouts are calculated relative to the C<ev_now ()>	1161	The relative timeouts are calculated relative to the C<ev_now ()>
1166	time. This is usually the right thing as this timestamp refers to the time	1162	time. This is usually the right thing as this timestamp refers to the time
…		…
1168	you suspect event processing to be delayed and you I<need> to base the timeout	1164	you suspect event processing to be delayed and you I<need> to base the timeout
1169	on the current time, use something like this to adjust for this:	1165	on the current time, use something like this to adjust for this:
1170		1166
1171	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);	1167	ev_timer_set (&timer, after + ev_now () - ev_time (), 0.);
1172		1168
1173	The callback is guarenteed to be invoked only when its timeout has passed,	1169	The callback is guarenteed to be invoked only after its timeout has passed,
1174	but if multiple timers become ready during the same loop iteration then	1170	but if multiple timers become ready during the same loop iteration then
1175	order of execution is undefined.	1171	order of execution is undefined.
1176		1172
1177	=head3 Watcher-Specific Functions and Data Members	1173	=head3 Watcher-Specific Functions and Data Members
1178		1174
…		…
1180		1176
1181	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)	1177	=item ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)
1182		1178
1183	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)	1179	=item ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)
1184		1180
1185	Configure the timer to trigger after C<after> seconds. If C<repeat> is	1181	Configure the timer to trigger after C<after> seconds. If C<repeat>
1186	C<0.>, then it will automatically be stopped. If it is positive, then the	1182	is C<0.>, then it will automatically be stopped once the timeout is
1187	timer will automatically be configured to trigger again C<repeat> seconds	1183	reached. If it is positive, then the timer will automatically be
1188	later, again, and again, until stopped manually.	1184	configured to trigger again C<repeat> seconds later, again, and again,
		1185	until stopped manually.
1189		1186
1190	The timer itself will do a best-effort at avoiding drift, that is, if you	1187	The timer itself will do a best-effort at avoiding drift, that is, if
1191	configure a timer to trigger every 10 seconds, then it will trigger at	1188	you configure a timer to trigger every 10 seconds, then it will normally
1192	exactly 10 second intervals. If, however, your program cannot keep up with	1189	trigger at exactly 10 second intervals. If, however, your program cannot
1193	the timer (because it takes longer than those 10 seconds to do stuff) the	1190	keep up with the timer (because it takes longer than those 10 seconds to
1194	timer will not fire more than once per event loop iteration.	1191	do stuff) the timer will not fire more than once per event loop iteration.
1195		1192
1196	=item ev_timer_again (loop, ev_timer *)	1193	=item ev_timer_again (loop, ev_timer *)
1197		1194
1198	This will act as if the timer timed out and restart it again if it is	1195	This will act as if the timer timed out and restart it again if it is
1199	repeating. The exact semantics are:	1196	repeating. The exact semantics are:
…		…
1276	Periodic watchers are also timers of a kind, but they are very versatile	1273	Periodic watchers are also timers of a kind, but they are very versatile
1277	(and unfortunately a bit complex).	1274	(and unfortunately a bit complex).
1278		1275
1279	Unlike C<ev_timer>'s, they are not based on real time (or relative time)	1276	Unlike C<ev_timer>'s, they are not based on real time (or relative time)
1280	but on wallclock time (absolute time). You can tell a periodic watcher	1277	but on wallclock time (absolute time). You can tell a periodic watcher
1281	to trigger "at" some specific point in time. For example, if you tell a	1278	to trigger after some specific point in time. For example, if you tell a
1282	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()	1279	periodic watcher to trigger in 10 seconds (by specifiying e.g. C<ev_now ()
1283	+ 10.>) and then reset your system clock to the last year, then it will	1280	+ 10.>, that is, an absolute time not a delay) and then reset your system
		1281	clock to january of the previous year, then it will take more than year
1284	take a year to trigger the event (unlike an C<ev_timer>, which would trigger	1282	to trigger the event (unlike an C<ev_timer>, which would still trigger
1285	roughly 10 seconds later).	1283	roughly 10 seconds later as it uses a relative timeout).
1286		1284
1287	They can also be used to implement vastly more complex timers, such as	1285	C<ev_periodic>s can also be used to implement vastly more complex timers,
1288	triggering an event on each midnight, local time or other, complicated,	1286	such as triggering an event on each "midnight, local time", or other
1289	rules.	1287	complicated, rules.
1290		1288
1291	As with timers, the callback is guarenteed to be invoked only when the	1289	As with timers, the callback is guarenteed to be invoked only when the
1292	time (C<at>) has been passed, but if multiple periodic timers become ready	1290	time (C<at>) has passed, but if multiple periodic timers become ready
1293	during the same loop iteration then order of execution is undefined.	1291	during the same loop iteration then order of execution is undefined.
1294		1292
1295	=head3 Watcher-Specific Functions and Data Members	1293	=head3 Watcher-Specific Functions and Data Members
1296		1294
1297	=over 4	1295	=over 4
…		…
1305		1303
1306	=over 4	1304	=over 4
1307		1305
1308	=item * absolute timer (at = time, interval = reschedule_cb = 0)	1306	=item * absolute timer (at = time, interval = reschedule_cb = 0)
1309		1307
1310	In this configuration the watcher triggers an event at the wallclock time	1308	In this configuration the watcher triggers an event after the wallclock
1311	C<at> and doesn't repeat. It will not adjust when a time jump occurs,	1309	time C<at> has passed and doesn't repeat. It will not adjust when a time
1312	that is, if it is to be run at January 1st 2011 then it will run when the	1310	jump occurs, that is, if it is to be run at January 1st 2011 then it will
1313	system time reaches or surpasses this time.	1311	run when the system time reaches or surpasses this time.
1314		1312
1315	=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)	1313	=item * repeating interval timer (at = offset, interval > 0, reschedule_cb = 0)
1316		1314
1317	In this mode the watcher will always be scheduled to time out at the next	1315	In this mode the watcher will always be scheduled to time out at the next
1318	C<at + N * interval> time (for some integer N, which can also be negative)	1316	C<at + N * interval> time (for some integer N, which can also be negative)
1319	and then repeat, regardless of any time jumps.	1317	and then repeat, regardless of any time jumps.
1320		1318
1321	This can be used to create timers that do not drift with respect to system	1319	This can be used to create timers that do not drift with respect to system
1322	time:	1320	time, for example, here is a C<ev_periodic> that triggers each hour, on
		1321	the hour:
1323		1322
1324	ev_periodic_set (&periodic, 0., 3600., 0);	1323	ev_periodic_set (&periodic, 0., 3600., 0);
1325		1324
1326	This doesn't mean there will always be 3600 seconds in between triggers,	1325	This doesn't mean there will always be 3600 seconds in between triggers,
1327	but only that the the callback will be called when the system time shows a	1326	but only that the the callback will be called when the system time shows a
…		…
1332	C<ev_periodic> will try to run the callback in this mode at the next possible	1331	C<ev_periodic> will try to run the callback in this mode at the next possible
1333	time where C<time = at (mod interval)>, regardless of any time jumps.	1332	time where C<time = at (mod interval)>, regardless of any time jumps.
1334		1333
1335	For numerical stability it is preferable that the C<at> value is near	1334	For numerical stability it is preferable that the C<at> value is near
1336	C<ev_now ()> (the current time), but there is no range requirement for	1335	C<ev_now ()> (the current time), but there is no range requirement for
1337	this value.	1336	this value, and in fact is often specified as zero.
1338		1337
1339	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)	1338	=item * manual reschedule mode (at and interval ignored, reschedule_cb = callback)
1340		1339
1341	In this mode the values for C<interval> and C<at> are both being	1340	In this mode the values for C<interval> and C<at> are both being
1342	ignored. Instead, each time the periodic watcher gets scheduled, the	1341	ignored. Instead, each time the periodic watcher gets scheduled, the
1343	reschedule callback will be called with the watcher as first, and the	1342	reschedule callback will be called with the watcher as first, and the
1344	current time as second argument.	1343	current time as second argument.
1345		1344
1346	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,	1345	NOTE: I<This callback MUST NOT stop or destroy any periodic watcher,
1347	ever, or make any event loop modifications>. If you need to stop it,	1346	ever, or make ANY event loop modifications whatsoever>.
1348	return C<now + 1e30> (or so, fudge fudge) and stop it afterwards (e.g. by
1349	starting an C<ev_prepare> watcher, which is legal).
1350		1347
		1348	If you need to stop it, return C<now + 1e30> (or so, fudge fudge) and stop
		1349	it afterwards (e.g. by starting an C<ev_prepare> watcher, which is the
		1350	only event loop modification you are allowed to do).
		1351
1351	Its prototype is C<ev_tstamp (reschedule_cb)(struct ev_periodic w,	1352	The callback prototype is C<ev_tstamp (*reschedule_cb)(struct ev_periodic
1352	ev_tstamp now)>, e.g.:	1353	*w, ev_tstamp now)>, e.g.:
1353		1354
1354	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)	1355	static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
1355	{	1356	{
1356	return now + 60.;	1357	return now + 60.;
1357	}	1358	}
…		…
1359	It must return the next time to trigger, based on the passed time value	1360	It must return the next time to trigger, based on the passed time value
1360	(that is, the lowest time value larger than to the second argument). It	1361	(that is, the lowest time value larger than to the second argument). It
1361	will usually be called just before the callback will be triggered, but	1362	will usually be called just before the callback will be triggered, but
1362	might be called at other times, too.	1363	might be called at other times, too.
1363		1364
1364	NOTE: I<< This callback must always return a time that is later than the	1365	NOTE: I<< This callback must always return a time that is higher than or
1365	passed C<now> value >>. Not even C<now> itself will do, it I<must> be larger.	1366	equal to the passed C<now> value >>.
1366		1367
1367	This can be used to create very complex timers, such as a timer that	1368	This can be used to create very complex timers, such as a timer that
1368	triggers on each midnight, local time. To do this, you would calculate the	1369	triggers on "next midnight, local time". To do this, you would calculate the
1369	next midnight after C<now> and return the timestamp value for this. How	1370	next midnight after C<now> and return the timestamp value for this. How
1370	you do this is, again, up to you (but it is not trivial, which is the main	1371	you do this is, again, up to you (but it is not trivial, which is the main
1371	reason I omitted it as an example).	1372	reason I omitted it as an example).
1372		1373
1373	=back	1374	=back
…		…
1377	Simply stops and restarts the periodic watcher again. This is only useful	1378	Simply stops and restarts the periodic watcher again. This is only useful
1378	when you changed some parameters or the reschedule callback would return	1379	when you changed some parameters or the reschedule callback would return
1379	a different time than the last time it was called (e.g. in a crond like	1380	a different time than the last time it was called (e.g. in a crond like
1380	program when the crontabs have changed).	1381	program when the crontabs have changed).
1381		1382
		1383	=item ev_tstamp ev_periodic_at (ev_periodic *)
		1384
		1385	When active, returns the absolute time that the watcher is supposed to
		1386	trigger next.
		1387
1382	=item ev_tstamp offset [read-write]	1388	=item ev_tstamp offset [read-write]
1383		1389
1384	When repeating, this contains the offset value, otherwise this is the	1390	When repeating, this contains the offset value, otherwise this is the
1385	absolute point in time (the C<at> value passed to C<ev_periodic_set>).	1391	absolute point in time (the C<at> value passed to C<ev_periodic_set>).
1386		1392
…		…
1396	=item ev_tstamp (reschedule_cb)(struct ev_periodic w, ev_tstamp now) [read-write]	1402	=item ev_tstamp (reschedule_cb)(struct ev_periodic w, ev_tstamp now) [read-write]
1397		1403
1398	The current reschedule callback, or C<0>, if this functionality is	1404	The current reschedule callback, or C<0>, if this functionality is
1399	switched off. Can be changed any time, but changes only take effect when	1405	switched off. Can be changed any time, but changes only take effect when
1400	the periodic timer fires or C<ev_periodic_again> is being called.	1406	the periodic timer fires or C<ev_periodic_again> is being called.
1401
1402	=item ev_tstamp at [read-only]
1403
1404	When active, contains the absolute time that the watcher is supposed to
1405	trigger next.
1406		1407
1407	=back	1408	=back
1408		1409
1409	=head3 Examples	1410	=head3 Examples
1410		1411
…		…
1614	as even with OS-supported change notifications, this can be	1615	as even with OS-supported change notifications, this can be
1615	resource-intensive.	1616	resource-intensive.
1616		1617
1617	At the time of this writing, only the Linux inotify interface is	1618	At the time of this writing, only the Linux inotify interface is
1618	implemented (implementing kqueue support is left as an exercise for the	1619	implemented (implementing kqueue support is left as an exercise for the
		1620	reader, note, however, that the author sees no way of implementing ev_stat
1619	reader). Inotify will be used to give hints only and should not change the	1621	semantics with kqueue). Inotify will be used to give hints only and should
1620	semantics of C<ev_stat> watchers, which means that libev sometimes needs	1622	not change the semantics of C<ev_stat> watchers, which means that libev
1621	to fall back to regular polling again even with inotify, but changes are	1623	sometimes needs to fall back to regular polling again even with inotify,
1622	usually detected immediately, and if the file exists there will be no	1624	but changes are usually detected immediately, and if the file exists there
1623	polling.	1625	will be no polling.
1624		1626
1625	=head3 ABI Issues (Largefile Support)	1627	=head3 ABI Issues (Largefile Support)
1626		1628
1627	Libev by default (unless the user overrides this) uses the default	1629	Libev by default (unless the user overrides this) uses the default
1628	compilation environment, which means that on systems with optionally	1630	compilation environment, which means that on systems with optionally
…		…
1638	When C<inotify (7)> support has been compiled into libev (generally only	1640	When C<inotify (7)> support has been compiled into libev (generally only
1639	available on Linux) and present at runtime, it will be used to speed up	1641	available on Linux) and present at runtime, it will be used to speed up
1640	change detection where possible. The inotify descriptor will be created lazily	1642	change detection where possible. The inotify descriptor will be created lazily
1641	when the first C<ev_stat> watcher is being started.	1643	when the first C<ev_stat> watcher is being started.
1642		1644
1643	Inotify presense does not change the semantics of C<ev_stat> watchers	1645	Inotify presence does not change the semantics of C<ev_stat> watchers
1644	except that changes might be detected earlier, and in some cases, to avoid	1646	except that changes might be detected earlier, and in some cases, to avoid
1645	making regular C<stat> calls. Even in the presense of inotify support	1647	making regular C<stat> calls. Even in the presence of inotify support
1646	there are many cases where libev has to resort to regular C<stat> polling.	1648	there are many cases where libev has to resort to regular C<stat> polling.
1647		1649
1648	(There is no support for kqueue, as apparently it cannot be used to	1650	(There is no support for kqueue, as apparently it cannot be used to
1649	implement this functionality, due to the requirement of having a file	1651	implement this functionality, due to the requirement of having a file
1650	descriptor open on the object at all times).	1652	descriptor open on the object at all times).
…		…
1653		1655
1654	The C<stat ()> syscall only supports full-second resolution portably, and	1656	The C<stat ()> syscall only supports full-second resolution portably, and
1655	even on systems where the resolution is higher, many filesystems still	1657	even on systems where the resolution is higher, many filesystems still
1656	only support whole seconds.	1658	only support whole seconds.
1657		1659
1658	That means that, if the time is the only thing that changes, you might	1660	That means that, if the time is the only thing that changes, you can
1659	miss updates: on the first update, C<ev_stat> detects a change and calls	1661	easily miss updates: on the first update, C<ev_stat> detects a change and
1660	your callback, which does something. When there is another update within	1662	calls your callback, which does something. When there is another update
1661	the same second, C<ev_stat> will be unable to detect it.	1663	within the same second, C<ev_stat> will be unable to detect it as the stat
		1664	data does not change.
1662		1665
1663	The solution to this is to delay acting on a change for a second (or till	1666	The solution to this is to delay acting on a change for slightly more
1664	the next second boundary), using a roughly one-second delay C<ev_timer>	1667	than a second (or till slightly after the next full second boundary), using
1665	(C<ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)>). The C<.01>	1668	a roughly one-second-delay C<ev_timer> (e.g. C<ev_timer_set (w, 0., 1.02);
1666	is added to work around small timing inconsistencies of some operating	1669	ev_timer_again (loop, w)>).
1667	systems.	1670
		1671	The C<.02> offset is added to work around small timing inconsistencies
		1672	of some operating systems (where the second counter of the current time
		1673	might be be delayed. One such system is the Linux kernel, where a call to
		1674	C<gettimeofday> might return a timestamp with a full second later than
		1675	a subsequent C<time> call - if the equivalent of C<time ()> is used to
		1676	update file times then there will be a small window where the kernel uses
		1677	the previous second to update file times but libev might already execute
		1678	the timer callback).
1668		1679
1669	=head3 Watcher-Specific Functions and Data Members	1680	=head3 Watcher-Specific Functions and Data Members
1670		1681
1671	=over 4	1682	=over 4
1672		1683
…		…
1678	C<path>. The C<interval> is a hint on how quickly a change is expected to	1689	C<path>. The C<interval> is a hint on how quickly a change is expected to
1679	be detected and should normally be specified as C<0> to let libev choose	1690	be detected and should normally be specified as C<0> to let libev choose
1680	a suitable value. The memory pointed to by C<path> must point to the same	1691	a suitable value. The memory pointed to by C<path> must point to the same
1681	path for as long as the watcher is active.	1692	path for as long as the watcher is active.
1682		1693
1683	The callback will be receive C<EV_STAT> when a change was detected,	1694	The callback will receive C<EV_STAT> when a change was detected, relative
1684	relative to the attributes at the time the watcher was started (or the	1695	to the attributes at the time the watcher was started (or the last change
1685	last change was detected).	1696	was detected).
1686		1697
1687	=item ev_stat_stat (loop, ev_stat *)	1698	=item ev_stat_stat (loop, ev_stat *)
1688		1699
1689	Updates the stat buffer immediately with new values. If you change the	1700	Updates the stat buffer immediately with new values. If you change the
1690	watched path in your callback, you could call this fucntion to avoid	1701	watched path in your callback, you could call this function to avoid
1691	detecting this change (while introducing a race condition). Can also be	1702	detecting this change (while introducing a race condition if you are not
1692	useful simply to find out the new values.	1703	the only one changing the path). Can also be useful simply to find out the
		1704	new values.
1693		1705
1694	=item ev_statdata attr [read-only]	1706	=item ev_statdata attr [read-only]
1695		1707
1696	The most-recently detected attributes of the file. Although the type is of	1708	The most-recently detected attributes of the file. Although the type is
1697	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types	1709	C<ev_statdata>, this is usually the (or one of the) C<struct stat> types
1698	suitable for your system. If the C<st_nlink> member is C<0>, then there	1710	suitable for your system, but you can only rely on the POSIX-standardised
		1711	members to be present. If the C<st_nlink> member is C<0>, then there was
1699	was some error while C<stat>ing the file.	1712	some error while C<stat>ing the file.
1700		1713
1701	=item ev_statdata prev [read-only]	1714	=item ev_statdata prev [read-only]
1702		1715
1703	The previous attributes of the file. The callback gets invoked whenever	1716	The previous attributes of the file. The callback gets invoked whenever
1704	C<prev> != C<attr>.	1717	C<prev> != C<attr>, or, more precisely, one or more of these members
		1718	differ: C<st_dev>, C<st_ino>, C<st_mode>, C<st_nlink>, C<st_uid>,
		1719	C<st_gid>, C<st_rdev>, C<st_size>, C<st_atime>, C<st_mtime>, C<st_ctime>.
1705		1720
1706	=item ev_tstamp interval [read-only]	1721	=item ev_tstamp interval [read-only]
1707		1722
1708	The specified interval.	1723	The specified interval.
1709		1724
…		…
1763	}	1778	}
1764		1779
1765	...	1780	...
1766	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);	1781	ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
1767	ev_stat_start (loop, &passwd);	1782	ev_stat_start (loop, &passwd);
1768	ev_timer_init (&timer, timer_cb, 0., 1.01);	1783	ev_timer_init (&timer, timer_cb, 0., 1.02);
1769		1784
1770		1785
1771	=head2 C<ev_idle> - when you've got nothing better to do...	1786	=head2 C<ev_idle> - when you've got nothing better to do...
1772		1787
1773	Idle watchers trigger events when no other events of the same or higher	1788	Idle watchers trigger events when no other events of the same or higher
…		…
1861		1876
1862	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)	1877	It is recommended to give C<ev_check> watchers highest (C<EV_MAXPRI>)
1863	priority, to ensure that they are being run before any other watchers	1878	priority, to ensure that they are being run before any other watchers
1864	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,	1879	after the poll. Also, C<ev_check> watchers (and C<ev_prepare> watchers,
1865	too) should not activate ("feed") events into libev. While libev fully	1880	too) should not activate ("feed") events into libev. While libev fully
1866	supports this, they will be called before other C<ev_check> watchers	1881	supports this, they might get executed before other C<ev_check> watchers
1867	did their job. As C<ev_check> watchers are often used to embed other	1882	did their job. As C<ev_check> watchers are often used to embed other
1868	(non-libev) event loops those other event loops might be in an unusable	1883	(non-libev) event loops those other event loops might be in an unusable
1869	state until their C<ev_check> watcher ran (always remind yourself to	1884	state until their C<ev_check> watcher ran (always remind yourself to
1870	coexist peacefully with others).	1885	coexist peacefully with others).
1871		1886
…		…
1886	=head3 Examples	1901	=head3 Examples
1887		1902
1888	There are a number of principal ways to embed other event loops or modules	1903	There are a number of principal ways to embed other event loops or modules
1889	into libev. Here are some ideas on how to include libadns into libev	1904	into libev. Here are some ideas on how to include libadns into libev
1890	(there is a Perl module named C<EV::ADNS> that does this, which you could	1905	(there is a Perl module named C<EV::ADNS> that does this, which you could
1891	use for an actually working example. Another Perl module named C<EV::Glib>	1906	use as a working example. Another Perl module named C<EV::Glib> embeds a
1892	embeds a Glib main context into libev, and finally, C<Glib::EV> embeds EV	1907	Glib main context into libev, and finally, C<Glib::EV> embeds EV into the
1893	into the Glib event loop).	1908	Glib event loop).
1894		1909
1895	Method 1: Add IO watchers and a timeout watcher in a prepare handler,	1910	Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1896	and in a check watcher, destroy them and call into libadns. What follows	1911	and in a check watcher, destroy them and call into libadns. What follows
1897	is pseudo-code only of course. This requires you to either use a low	1912	is pseudo-code only of course. This requires you to either use a low
1898	priority for the check watcher or use C<ev_clear_pending> explicitly, as	1913	priority for the check watcher or use C<ev_clear_pending> explicitly, as
…		…
2382		2397
2383	=item * Priorities are not currently supported. Initialising priorities	2398	=item * Priorities are not currently supported. Initialising priorities
2384	will fail and all watchers will have the same priority, even though there	2399	will fail and all watchers will have the same priority, even though there
2385	is an ev_pri field.	2400	is an ev_pri field.
2386		2401
		2402	=item * In libevent, the last base created gets the signals, in libev, the
		2403	first base created (== the default loop) gets the signals.
		2404
2387	=item * Other members are not supported.	2405	=item * Other members are not supported.
2388		2406
2389	=item * The libev emulation is I<not> ABI compatible to libevent, you need	2407	=item * The libev emulation is I<not> ABI compatible to libevent, you need
2390	to use the libev header file and library.	2408	to use the libev header file and library.
2391		2409
…		…
2967	defined to be C<0>, then they are not.	2985	defined to be C<0>, then they are not.
2968		2986
2969	=item EV_MINIMAL	2987	=item EV_MINIMAL
2970		2988
2971	If you need to shave off some kilobytes of code at the expense of some	2989	If you need to shave off some kilobytes of code at the expense of some
2972	speed, define this symbol to C<1>. Currently only used for gcc to override	2990	speed, define this symbol to C<1>. Currently this is used to override some
2973	some inlining decisions, saves roughly 30% codesize of amd64.	2991	inlining decisions, saves roughly 30% codesize of amd64. It also selects a
		2992	much smaller 2-heap for timer management over the default 4-heap.
2974		2993
2975	=item EV_PID_HASHSIZE	2994	=item EV_PID_HASHSIZE
2976		2995
2977	C<ev_child> watchers use a small hash table to distribute workload by	2996	C<ev_child> watchers use a small hash table to distribute workload by
2978	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more	2997	pid. The default size is C<16> (or C<1> with C<EV_MINIMAL>), usually more
…		…
2984	C<ev_stat> watchers use a small hash table to distribute workload by	3003	C<ev_stat> watchers use a small hash table to distribute workload by
2985	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),	3004	inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
2986	usually more than enough. If you need to manage thousands of C<ev_stat>	3005	usually more than enough. If you need to manage thousands of C<ev_stat>
2987	watchers you might want to increase this value (I<must> be a power of	3006	watchers you might want to increase this value (I<must> be a power of
2988	two).	3007	two).
		3008
		3009	=item EV_USE_4HEAP
		3010
		3011	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3012	timer and periodics heap, libev uses a 4-heap when this symbol is defined
		3013	to C<1>. The 4-heap uses more complicated (longer) code but has
		3014	noticably faster performance with many (thousands) of watchers.
		3015
		3016	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3017	(disabled).
		3018
		3019	=item EV_HEAP_CACHE_AT
		3020
		3021	Heaps are not very cache-efficient. To improve the cache-efficiency of the
		3022	timer and periodics heap, libev can cache the timestamp (I<at>) within
		3023	the heap structure (selected by defining C<EV_HEAP_CACHE_AT> to C<1>),
		3024	which uses 8-12 bytes more per watcher and a few hundred bytes more code,
		3025	but avoids random read accesses on heap changes. This improves performance
		3026	noticably with with many (hundreds) of watchers.
		3027
		3028	The default is C<1> unless C<EV_MINIMAL> is set in which case it is C<0>
		3029	(disabled).
2989		3030
2990	=item EV_COMMON	3031	=item EV_COMMON
2991		3032
2992	By default, all watchers have a C<void *data> member. By redefining	3033	By default, all watchers have a C<void *data> member. By redefining
2993	this macro to a something else you can include more and other types of	3034	this macro to a something else you can include more and other types of
…		…
3067		3108
3068	#include "ev_cpp.h"	3109	#include "ev_cpp.h"
3069	#include "ev.c"	3110	#include "ev.c"
3070		3111
3071		3112
		3113	=head1 THREADS AND COROUTINES
		3114
		3115	=head2 THREADS
		3116
		3117	Libev itself is completely threadsafe, but it uses no locking. This
		3118	means that you can use as many loops as you want in parallel, as long as
		3119	only one thread ever calls into one libev function with the same loop
		3120	parameter.
		3121
		3122	Or put differently: calls with different loop parameters can be done in
		3123	parallel from multiple threads, calls with the same loop parameter must be
		3124	done serially (but can be done from different threads, as long as only one
		3125	thread ever is inside a call at any point in time, e.g. by using a mutex
		3126	per loop).
		3127
		3128	If you want to know which design is best for your problem, then I cannot
		3129	help you but by giving some generic advice:
		3130
		3131	=over 4
		3132
		3133	=item * most applications have a main thread: use the default libev loop
		3134	in that thread, or create a seperate thread running only the default loop.
		3135
		3136	This helps integrating other libraries or software modules that use libev
		3137	themselves and don't care/know about threading.
		3138
		3139	=item * one loop per thread is usually a good model.
		3140
		3141	Doing this is almost never wrong, sometimes a better-performance model
		3142	exists, but it is always a good start.
		3143
		3144	=item * other models exist, such as the leader/follower pattern, where one
		3145	loop is handed through multiple threads in a kind of round-robbin fashion.
		3146
		3147	Chosing a model is hard - look around, learn, know that usually you cna do
		3148	better than you currently do :-)
		3149
		3150	=item * often you need to talk to some other thread which blocks in the
		3151	event loop - C<ev_async> watchers can be used to wake them up from other
		3152	threads safely (or from signal contexts...).
		3153
		3154	=back
		3155
		3156	=head2 COROUTINES
		3157
		3158	Libev is much more accomodating to coroutines ("cooperative threads"):
		3159	libev fully supports nesting calls to it's functions from different
		3160	coroutines (e.g. you can call C<ev_loop> on the same loop from two
		3161	different coroutines and switch freely between both coroutines running the
		3162	loop, as long as you don't confuse yourself). The only exception is that
		3163	you must not do this from C<ev_periodic> reschedule callbacks.
		3164
		3165	Care has been invested into making sure that libev does not keep local
		3166	state inside C<ev_loop>, and other calls do not usually allow coroutine
		3167	switches.
		3168
		3169
3072	=head1 COMPLEXITIES	3170	=head1 COMPLEXITIES
3073		3171
3074	In this section the complexities of (many of) the algorithms used inside	3172	In this section the complexities of (many of) the algorithms used inside
3075	libev will be explained. For complexity discussions about backends see the	3173	libev will be explained. For complexity discussions about backends see the
3076	documentation for C<ev_default_init>.	3174	documentation for C<ev_default_init>.
…		…
3106	correct watcher to remove. The lists are usually short (you don't usually	3204	correct watcher to remove. The lists are usually short (you don't usually
3107	have many watchers waiting for the same fd or signal).	3205	have many watchers waiting for the same fd or signal).
3108		3206
3109	=item Finding the next timer in each loop iteration: O(1)	3207	=item Finding the next timer in each loop iteration: O(1)
3110		3208
3111	By virtue of using a binary heap, the next timer is always found at the	3209	By virtue of using a binary or 4-heap, the next timer is always found at a
3112	beginning of the storage array.	3210	fixed position in the storage array.
3113		3211
3114	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)	3212	=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
3115		3213
3116	A change means an I/O watcher gets started or stopped, which requires	3214	A change means an I/O watcher gets started or stopped, which requires
3117	libev to recalculate its status (and possibly tell the kernel, depending	3215	libev to recalculate its status (and possibly tell the kernel, depending
…		…
3146	model. Libev still offers limited functionality on this platform in	3244	model. Libev still offers limited functionality on this platform in
3147	the form of the C<EVBACKEND_SELECT> backend, and only supports socket	3245	the form of the C<EVBACKEND_SELECT> backend, and only supports socket
3148	descriptors. This only applies when using Win32 natively, not when using	3246	descriptors. This only applies when using Win32 natively, not when using
3149	e.g. cygwin.	3247	e.g. cygwin.
3150		3248
		3249	Lifting these limitations would basically require the full
		3250	re-implementation of the I/O system. If you are into these kinds of
		3251	things, then note that glib does exactly that for you in a very portable
		3252	way (note also that glib is the slowest event library known to man).
		3253
3151	There is no supported compilation method available on windows except	3254	There is no supported compilation method available on windows except
3152	embedding it into other applications.	3255	embedding it into other applications.
3153		3256
3154	Due to the many, low, and arbitrary limits on the win32 platform and the	3257	Due to the many, low, and arbitrary limits on the win32 platform and
3155	abysmal performance of winsockets, using a large number of sockets is not	3258	the abysmal performance of winsockets, using a large number of sockets
3156	recommended (and not reasonable). If your program needs to use more than	3259	is not recommended (and not reasonable). If your program needs to use
3157	a hundred or so sockets, then likely it needs to use a totally different	3260	more than a hundred or so sockets, then likely it needs to use a totally
3158	implementation for windows, as libev offers the POSIX model, which cannot	3261	different implementation for windows, as libev offers the POSIX readiness
3159	be implemented efficiently on windows (microsoft monopoly games).	3262	notification model, which cannot be implemented efficiently on windows
		3263	(microsoft monopoly games).
3160		3264
3161	=over 4	3265	=over 4
3162		3266
3163	=item The winsocket select function	3267	=item The winsocket select function
3164		3268
…		…
3178	Note that winsockets handling of fd sets is O(n), so you can easily get a	3282	Note that winsockets handling of fd sets is O(n), so you can easily get a
3179	complexity in the O(n²) range when using win32.	3283	complexity in the O(n²) range when using win32.
3180		3284
3181	=item Limited number of file descriptors	3285	=item Limited number of file descriptors
3182		3286
3183	Windows has numerous arbitrary (and low) limits on things. Early versions	3287	Windows has numerous arbitrary (and low) limits on things.
3184	of winsocket's select only supported waiting for a max. of C<64> handles	3288
		3289	Early versions of winsocket's select only supported waiting for a maximum
3185	(probably owning to the fact that all windows kernels can only wait for	3290	of C<64> handles (probably owning to the fact that all windows kernels
3186	C<64> things at the same time internally; microsoft recommends spawning a	3291	can only wait for C<64> things at the same time internally; microsoft
3187	chain of threads and wait for 63 handles and the previous thread in each).	3292	recommends spawning a chain of threads and wait for 63 handles and the
		3293	previous thread in each. Great).
3188		3294
3189	Newer versions support more handles, but you need to define C<FD_SETSIZE>	3295	Newer versions support more handles, but you need to define C<FD_SETSIZE>
3190	to some high number (e.g. C<2048>) before compiling the winsocket select	3296	to some high number (e.g. C<2048>) before compiling the winsocket select
3191	call (which might be in libev or elsewhere, for example, perl does its own	3297	call (which might be in libev or elsewhere, for example, perl does its own
3192	select emulation on windows).	3298	select emulation on windows).
…		…
3204	calling select (O(n²)) will likely make this unworkable.	3310	calling select (O(n²)) will likely make this unworkable.
3205		3311
3206	=back	3312	=back
3207		3313
3208		3314
		3315	=head1 PORTABILITY REQUIREMENTS
		3316
		3317	In addition to a working ISO-C implementation, libev relies on a few
		3318	additional extensions:
		3319
		3320	=over 4
		3321
		3322	=item C<sig_atomic_t volatile> must be thread-atomic as well
		3323
		3324	The type C<sig_atomic_t volatile> (or whatever is defined as
		3325	C<EV_ATOMIC_T>) must be atomic w.r.t. accesses from different
		3326	threads. This is not part of the specification for C<sig_atomic_t>, but is
		3327	believed to be sufficiently portable.
		3328
		3329	=item C<sigprocmask> must work in a threaded environment
		3330
		3331	Libev uses C<sigprocmask> to temporarily block signals. This is not
		3332	allowed in a threaded program (C<pthread_sigmask> has to be used). Typical
		3333	pthread implementations will either allow C<sigprocmask> in the "main
		3334	thread" or will block signals process-wide, both behaviours would
		3335	be compatible with libev. Interaction between C<sigprocmask> and
		3336	C<pthread_sigmask> could complicate things, however.
		3337
		3338	The most portable way to handle signals is to block signals in all threads
		3339	except the initial one, and run the default loop in the initial thread as
		3340	well.
		3341
		3342	=item C<long> must be large enough for common memory allocation sizes
		3343
		3344	To improve portability and simplify using libev, libev uses C<long>
		3345	internally instead of C<size_t> when allocating its data structures. On
		3346	non-POSIX systems (Microsoft...) this might be unexpectedly low, but
		3347	is still at least 31 bits everywhere, which is enough for hundreds of
		3348	millions of watchers.
		3349
		3350	=item C<double> must hold a time value in seconds with enough accuracy
		3351
		3352	The type C<double> is used to represent timestamps. It is required to
		3353	have at least 51 bits of mantissa (and 9 bits of exponent), which is good
		3354	enough for at least into the year 4000. This requirement is fulfilled by
		3355	implementations implementing IEEE 754 (basically all existing ones).
		3356
		3357	=back
		3358
		3359	If you know of other additional requirements drop me a note.
		3360
		3361
		3362	=head1 VALGRIND
		3363
		3364	Valgrind has a special section here because it is a popular tool that is
		3365	highly useful, but valgrind reports are very hard to interpret.
		3366
		3367	If you think you found a bug (memory leak, uninitialised data access etc.)
		3368	in libev, then check twice: If valgrind reports something like:
		3369
		3370	==2274== definitely lost: 0 bytes in 0 blocks.
		3371	==2274== possibly lost: 0 bytes in 0 blocks.
		3372	==2274== still reachable: 256 bytes in 1 blocks.
		3373
		3374	then there is no memory leak. Similarly, under some circumstances,
		3375	valgrind might report kernel bugs as if it were a bug in libev, or it
		3376	might be confused (it is a very good tool, but only a tool).
		3377
		3378	If you are unsure about something, feel free to contact the mailing list
		3379	with the full valgrind report and an explanation on why you think this is
		3380	a bug in libev. However, don't be annoyed when you get a brisk "this is
		3381	no bug" answer and take the chance of learning how to interpret valgrind
		3382	properly.
		3383
		3384	If you need, for some reason, empty reports from valgrind for your project
		3385	I suggest using suppression lists.
		3386
		3387
3209	=head1 AUTHOR	3388	=head1 AUTHOR
3210		3389
3211	Marc Lehmann <libev@schmorp.de>.	3390	Marc Lehmann <libev@schmorp.de>.
3212		3391

Diff Legend

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

Comparing libev/ev.pod (file contents): Revision 1.143 by root, Sun Apr 6 14:34:52 2008 UTC vs. Revision 1.157 by root, Tue May 20 23:49:41 2008 UTC

Diff Legend

Comparing libev/ev.pod (file contents):
Revision 1.143 by root, Sun Apr 6 14:34:52 2008 UTC vs.
Revision 1.157 by root, Tue May 20 23:49:41 2008 UTC