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

Comparing libev/ev.pod (file contents):
Revision 1.101 by ayin, Sat Dec 22 14:11:25 2007 UTC vs.
Revision 1.112 by root, Wed Dec 26 08:06:09 2007 UTC

… …
4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 #include <ev.h> 7 #include <ev.h>
8 8
9=head1 EXAMPLE PROGRAM 9=head2 EXAMPLE PROGRAM
10 10
11 #include <ev.h> 11 #include <ev.h>
12 12
13 ev_io stdin_watcher; 13 ev_io stdin_watcher;
14 ev_timer timeout_watcher; 14 ev_timer timeout_watcher;
… …
65You register interest in certain events by registering so-called I<event 65You register interest in certain events by registering so-called I<event
66watchers>, which are relatively small C structures you initialise with the 66watchers>, which are relatively small C structures you initialise with the
67details of the event, and then hand it over to libev by I<starting> the 67details of the event, and then hand it over to libev by I<starting> the
68watcher. 68watcher.
69 69
70=head1 FEATURES 70=head2 FEATURES
71 71
72Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the 72Libev supports C<select>, C<poll>, the Linux-specific C<epoll>, the
73BSD-specific C<kqueue> and the Solaris-specific event port mechanisms 73BSD-specific C<kqueue> and the Solaris-specific event port mechanisms
74for file descriptor events (C<ev_io>), the Linux C<inotify> interface 74for file descriptor events (C<ev_io>), the Linux C<inotify> interface
75(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers 75(for C<ev_stat>), relative timers (C<ev_timer>), absolute timers
… …
82 82
83It also is quite fast (see this 83It also is quite fast (see this
84L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent 84L<benchmark|http://libev.schmorp.de/bench.html> comparing it to libevent
85for example). 85for example).
86 86
87=head1 CONVENTIONS 87=head2 CONVENTIONS
88 88
89Libev is very configurable. In this manual the default configuration will 89Libev is very configurable. In this manual the default configuration will
90be described, which supports multiple event loops. For more info about 90be described, which supports multiple event loops. For more info about
91various configuration options please have a look at B<EMBED> section in 91various configuration options please have a look at B<EMBED> section in
92this manual. If libev was configured without support for multiple event 92this manual. If libev was configured without support for multiple event
93loops, then all functions taking an initial argument of name C<loop> 93loops, then all functions taking an initial argument of name C<loop>
94(which is always of type C<struct ev_loop *>) will not have this argument. 94(which is always of type C<struct ev_loop *>) will not have this argument.
95 95
96=head1 TIME REPRESENTATION 96=head2 TIME REPRESENTATION
97 97
98Libev represents time as a single floating point number, representing the 98Libev represents time as a single floating point number, representing the
99(fractional) number of seconds since the (POSIX) epoch (somewhere near 99(fractional) number of seconds since the (POSIX) epoch (somewhere near
100the beginning of 1970, details are complicated, don't ask). This type is 100the beginning of 1970, details are complicated, don't ask). This type is
101called C<ev_tstamp>, which is what you should use too. It usually aliases 101called C<ev_tstamp>, which is what you should use too. It usually aliases
… …
306=item C<EVBACKEND_SELECT> (value 1, portable select backend) 306=item C<EVBACKEND_SELECT> (value 1, portable select backend)
307 307
308This is your standard select(2) backend. Not I<completely> standard, as 308This is your standard select(2) backend. Not I<completely> standard, as
309libev tries to roll its own fd_set with no limits on the number of fds, 309libev tries to roll its own fd_set with no limits on the number of fds,
310but if that fails, expect a fairly low limit on the number of fds when 310but if that fails, expect a fairly low limit on the number of fds when
311using this backend. It doesn't scale too well (O(highest_fd)), but its usually 311using this backend. It doesn't scale too well (O(highest_fd)), but its
312the fastest backend for a low number of fds. 312usually the fastest backend for a low number of (low-numbered :) fds.
313
314To get good performance out of this backend you need a high amount of
315parallelity (most of the file descriptors should be busy). If you are
316writing a server, you should C<accept ()> in a loop to accept as many
317connections as possible during one iteration. You might also want to have
318a look at C<ev_set_io_collect_interval ()> to increase the amount of
319readyness notifications you get per iteration.
313 320
314=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows) 321=item C<EVBACKEND_POLL> (value 2, poll backend, available everywhere except on windows)
315 322
316And this is your standard poll(2) backend. It's more complicated than 323And this is your standard poll(2) backend. It's more complicated
317select, but handles sparse fds better and has no artificial limit on the 324than select, but handles sparse fds better and has no artificial
318number of fds you can use (except it will slow down considerably with a 325limit on the number of fds you can use (except it will slow down
319lot of inactive fds). It scales similarly to select, i.e. O(total_fds). 326considerably with a lot of inactive fds). It scales similarly to select,
327i.e. O(total_fds). See the entry for C<EVBACKEND_SELECT>, above, for
328performance tips.
320 329
321=item C<EVBACKEND_EPOLL> (value 4, Linux) 330=item C<EVBACKEND_EPOLL> (value 4, Linux)
322 331
323For few fds, this backend is a bit little slower than poll and select, 332For few fds, this backend is a bit little slower than poll and select,
324but it scales phenomenally better. While poll and select usually scale 333but it scales phenomenally better. While poll and select usually scale
325like O(total_fds) where n is the total number of fds (or the highest fd), 334like O(total_fds) where n is the total number of fds (or the highest fd),
326epoll scales either O(1) or O(active_fds). The epoll design has a number 335epoll scales either O(1) or O(active_fds). The epoll design has a number
327of shortcomings, such as silently dropping events in some hard-to-detect 336of shortcomings, such as silently dropping events in some hard-to-detect
328cases and rewiring a syscall per fd change, no fork support and bad 337cases and rewiring a syscall per fd change, no fork support and bad
329support for dup: 338support for dup.
330 339
331While stopping, setting and starting an I/O watcher in the same iteration 340While stopping, setting and starting an I/O watcher in the same iteration
332will result in some caching, there is still a syscall per such incident 341will result in some caching, there is still a syscall per such incident
333(because the fd could point to a different file description now), so its 342(because the fd could point to a different file description now), so its
334best to avoid that. Also, C<dup ()>'ed file descriptors might not work 343best to avoid that. Also, C<dup ()>'ed file descriptors might not work
335very well if you register events for both fds. 344very well if you register events for both fds.
336 345
337Please note that epoll sometimes generates spurious notifications, so you 346Please note that epoll sometimes generates spurious notifications, so you
338need to use non-blocking I/O or other means to avoid blocking when no data 347need to use non-blocking I/O or other means to avoid blocking when no data
339(or space) is available. 348(or space) is available.
349
350Best performance from this backend is achieved by not unregistering all
351watchers for a file descriptor until it has been closed, if possible, i.e.
352keep at least one watcher active per fd at all times.
353
354While nominally embeddeble in other event loops, this feature is broken in
355all kernel versions tested so far.
340 356
341=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones) 357=item C<EVBACKEND_KQUEUE> (value 8, most BSD clones)
342 358
343Kqueue deserves special mention, as at the time of this writing, it 359Kqueue deserves special mention, as at the time of this writing, it
344was broken on all BSDs except NetBSD (usually it doesn't work reliably 360was broken on all BSDs except NetBSD (usually it doesn't work reliably
… …
357course). While stopping, setting and starting an I/O watcher does never 373course). While stopping, setting and starting an I/O watcher does never
358cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to 374cause an extra syscall as with C<EVBACKEND_EPOLL>, it still adds up to
359two event changes per incident, support for C<fork ()> is very bad and it 375two event changes per incident, support for C<fork ()> is very bad and it
360drops fds silently in similarly hard-to-detect cases. 376drops fds silently in similarly hard-to-detect cases.
361 377
378This backend usually performs well under most conditions.
379
380While nominally embeddable in other event loops, this doesn't work
381everywhere, so you might need to test for this. And since it is broken
382almost everywhere, you should only use it when you have a lot of sockets
383(for which it usually works), by embedding it into another event loop
384(e.g. C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>) and using it only for
385sockets.
386
362=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8) 387=item C<EVBACKEND_DEVPOLL> (value 16, Solaris 8)
363 388
364This is not implemented yet (and might never be). 389This is not implemented yet (and might never be, unless you send me an
390implementation). According to reports, C</dev/poll> only supports sockets
391and is not embeddable, which would limit the usefulness of this backend
392immensely.
365 393
366=item C<EVBACKEND_PORT> (value 32, Solaris 10) 394=item C<EVBACKEND_PORT> (value 32, Solaris 10)
367 395
368This uses the Solaris 10 event port mechanism. As with everything on Solaris, 396This uses the Solaris 10 event port mechanism. As with everything on Solaris,
369it's really slow, but it still scales very well (O(active_fds)). 397it's really slow, but it still scales very well (O(active_fds)).
370 398
371Please note that solaris event ports can deliver a lot of spurious 399Please note that solaris event ports can deliver a lot of spurious
372notifications, so you need to use non-blocking I/O or other means to avoid 400notifications, so you need to use non-blocking I/O or other means to avoid
373blocking when no data (or space) is available. 401blocking when no data (or space) is available.
374 402
403While this backend scales well, it requires one system call per active
404file descriptor per loop iteration. For small and medium numbers of file
405descriptors a "slow" C<EVBACKEND_SELECT> or C<EVBACKEND_POLL> backend
406might perform better.
407
375=item C<EVBACKEND_ALL> 408=item C<EVBACKEND_ALL>
376 409
377Try all backends (even potentially broken ones that wouldn't be tried 410Try all backends (even potentially broken ones that wouldn't be tried
378with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as 411with C<EVFLAG_AUTO>). Since this is a mask, you can do stuff such as
379C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>. 412C<EVBACKEND_ALL & ~EVBACKEND_KQUEUE>.
413
414It is definitely not recommended to use this flag.
380 415
381=back 416=back
382 417
383If one or more of these are ored into the flags value, then only these 418If one or more of these are ored into the flags value, then only these
384backends will be tried (in the reverse order as given here). If none are 419backends will be tried (in the reverse order as given here). If none are
… …
948In general you can register as many read and/or write event watchers per 983In general you can register as many read and/or write event watchers per
949fd as you want (as long as you don't confuse yourself). Setting all file 984fd as you want (as long as you don't confuse yourself). Setting all file
950descriptors to non-blocking mode is also usually a good idea (but not 985descriptors to non-blocking mode is also usually a good idea (but not
951required if you know what you are doing). 986required if you know what you are doing).
952 987
953You have to be careful with dup'ed file descriptors, though. Some backends
954(the linux epoll backend is a notable example) cannot handle dup'ed file
955descriptors correctly if you register interest in two or more fds pointing
956to the same underlying file/socket/etc. description (that is, they share
957the same underlying "file open").
958
959If you must do this, then force the use of a known-to-be-good backend 988If you must do this, then force the use of a known-to-be-good backend
960(at the time of this writing, this includes only C<EVBACKEND_SELECT> and 989(at the time of this writing, this includes only C<EVBACKEND_SELECT> and
961C<EVBACKEND_POLL>). 990C<EVBACKEND_POLL>).
962 991
963Another thing you have to watch out for is that it is quite easy to 992Another thing you have to watch out for is that it is quite easy to
… …
997optimisations to libev. 1026optimisations to libev.
998 1027
999=head3 The special problem of dup'ed file descriptors 1028=head3 The special problem of dup'ed file descriptors
1000 1029
1001Some backends (e.g. epoll), cannot register events for file descriptors, 1030Some backends (e.g. epoll), cannot register events for file descriptors,
1002but only events for the underlying file descriptions. That menas when you 1031but only events for the underlying file descriptions. That means when you
1003have C<dup ()>'ed file descriptors and register events for them, only one 1032have C<dup ()>'ed file descriptors or weirder constellations, and register
1004file descriptor might actually receive events. 1033events for them, only one file descriptor might actually receive events.
1005 1034
1006There is no workaorund possible except not registering events 1035There is no workaround possible except not registering events
1007for potentially C<dup ()>'ed file descriptors or to resort to 1036for potentially C<dup ()>'ed file descriptors, or to resort to
1008C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>. 1037C<EVBACKEND_SELECT> or C<EVBACKEND_POLL>.
1009 1038
1010=head3 The special problem of fork 1039=head3 The special problem of fork
1011 1040
1012Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit 1041Some backends (epoll, kqueue) do not support C<fork ()> at all or exhibit
… …
1038=item int events [read-only] 1067=item int events [read-only]
1039 1068
1040The events being watched. 1069The events being watched.
1041 1070
1042=back 1071=back
1072
1073=head3 Examples
1043 1074
1044Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well 1075Example: Call C<stdin_readable_cb> when STDIN_FILENO has become, well
1045readable, but only once. Since it is likely line-buffered, you could 1076readable, but only once. Since it is likely line-buffered, you could
1046attempt to read a whole line in the callback. 1077attempt to read a whole line in the callback.
1047 1078
… …
1145or C<ev_timer_again> is called and determines the next timeout (if any), 1176or C<ev_timer_again> is called and determines the next timeout (if any),
1146which is also when any modifications are taken into account. 1177which is also when any modifications are taken into account.
1147 1178
1148=back 1179=back
1149 1180
1181=head3 Examples
1182
1150Example: Create a timer that fires after 60 seconds. 1183Example: Create a timer that fires after 60 seconds.
1151 1184
1152 static void 1185 static void
1153 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents) 1186 one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1154 { 1187 {
… …
1311When active, contains the absolute time that the watcher is supposed to 1344When active, contains the absolute time that the watcher is supposed to
1312trigger next. 1345trigger next.
1313 1346
1314=back 1347=back
1315 1348
1349=head3 Examples
1350
1316Example: Call a callback every hour, or, more precisely, whenever the 1351Example: Call a callback every hour, or, more precisely, whenever the
1317system clock is divisible by 3600. The callback invocation times have 1352system clock is divisible by 3600. The callback invocation times have
1318potentially a lot of jittering, but good long-term stability. 1353potentially a lot of jittering, but good long-term stability.
1319 1354
1320 static void 1355 static void
… …
1411 1446
1412The process exit/trace status caused by C<rpid> (see your systems 1447The process exit/trace status caused by C<rpid> (see your systems
1413C<waitpid> and C<sys/wait.h> documentation for details). 1448C<waitpid> and C<sys/wait.h> documentation for details).
1414 1449
1415=back 1450=back
1451
1452=head3 Examples
1416 1453
1417Example: Try to exit cleanly on SIGINT and SIGTERM. 1454Example: Try to exit cleanly on SIGINT and SIGTERM.
1418 1455
1419 static void 1456 static void
1420 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents) 1457 sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
… …
1461semantics of C<ev_stat> watchers, which means that libev sometimes needs 1498semantics of C<ev_stat> watchers, which means that libev sometimes needs
1462to fall back to regular polling again even with inotify, but changes are 1499to fall back to regular polling again even with inotify, but changes are
1463usually detected immediately, and if the file exists there will be no 1500usually detected immediately, and if the file exists there will be no
1464polling. 1501polling.
1465 1502
1503=head3 Inotify
1504
1505When C<inotify (7)> support has been compiled into libev (generally only
1506available on Linux) and present at runtime, it will be used to speed up
1507change detection where possible. The inotify descriptor will be created lazily
1508when the first C<ev_stat> watcher is being started.
1509
1510Inotify presense does not change the semantics of C<ev_stat> watchers
1511except that changes might be detected earlier, and in some cases, to avoid
1512making regular C<stat> calls. Even in the presense of inotify support
1513there are many cases where libev has to resort to regular C<stat> polling.
1514
1515(There is no support for kqueue, as apparently it cannot be used to
1516implement this functionality, due to the requirement of having a file
1517descriptor open on the object at all times).
1518
1519=head3 The special problem of stat time resolution
1520
1521The C<stat ()> syscall only supports full-second resolution portably, and
1522even on systems where the resolution is higher, many filesystems still
1523only support whole seconds.
1524
1525That means that, if the time is the only thing that changes, you might
1526miss updates: on the first update, C<ev_stat> detects a change and calls
1527your callback, which does something. When there is another update within
1528the same second, C<ev_stat> will be unable to detect it.
1529
1530The solution to this is to delay acting on a change for a second (or till
1531the next second boundary), using a roughly one-second delay C<ev_timer>
1532(C<ev_timer_set (w, 0., 1.01); ev_timer_again (loop, w)>). The C<.01>
1533is added to work around small timing inconsistencies of some operating
1534systems.
1535
1466=head3 Watcher-Specific Functions and Data Members 1536=head3 Watcher-Specific Functions and Data Members
1467 1537
1468=over 4 1538=over 4
1469 1539
1470=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval) 1540=item ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)
… …
1507=item const char *path [read-only] 1577=item const char *path [read-only]
1508 1578
1509The filesystem path that is being watched. 1579The filesystem path that is being watched.
1510 1580
1511=back 1581=back
1582
1583=head3 Examples
1512 1584
1513Example: Watch C</etc/passwd> for attribute changes. 1585Example: Watch C</etc/passwd> for attribute changes.
1514 1586
1515 static void 1587 static void
1516 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents) 1588 passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
… …
1529 } 1601 }
1530 1602
1531 ... 1603 ...
1532 ev_stat passwd; 1604 ev_stat passwd;
1533 1605
1534 ev_stat_init (&passwd, passwd_cb, "/etc/passwd"); 1606 ev_stat_init (&passwd, passwd_cb, "/etc/passwd", 0.);
1535 ev_stat_start (loop, &passwd); 1607 ev_stat_start (loop, &passwd);
1608
1609Example: Like above, but additionally use a one-second delay so we do not
1610miss updates (however, frequent updates will delay processing, too, so
1611one might do the work both on C<ev_stat> callback invocation I<and> on
1612C<ev_timer> callback invocation).
1613
1614 static ev_stat passwd;
1615 static ev_timer timer;
1616
1617 static void
1618 timer_cb (EV_P_ ev_timer *w, int revents)
1619 {
1620 ev_timer_stop (EV_A_ w);
1621
1622 /* now it's one second after the most recent passwd change */
1623 }
1624
1625 static void
1626 stat_cb (EV_P_ ev_stat *w, int revents)
1627 {
1628 /* reset the one-second timer */
1629 ev_timer_again (EV_A_ &timer);
1630 }
1631
1632 ...
1633 ev_stat_init (&passwd, stat_cb, "/etc/passwd", 0.);
1634 ev_stat_start (loop, &passwd);
1635 ev_timer_init (&timer, timer_cb, 0., 1.01);
1536 1636
1537 1637
1538=head2 C<ev_idle> - when you've got nothing better to do... 1638=head2 C<ev_idle> - when you've got nothing better to do...
1539 1639
1540Idle watchers trigger events when no other events of the same or higher 1640Idle watchers trigger events when no other events of the same or higher
… …
1565Initialises and configures the idle watcher - it has no parameters of any 1665Initialises and configures the idle watcher - it has no parameters of any
1566kind. There is a C<ev_idle_set> macro, but using it is utterly pointless, 1666kind. There is a C<ev_idle_set> macro, but using it is utterly pointless,
1567believe me. 1667believe me.
1568 1668
1569=back 1669=back
1670
1671=head3 Examples
1570 1672
1571Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the 1673Example: Dynamically allocate an C<ev_idle> watcher, start it, and in the
1572callback, free it. Also, use no error checking, as usual. 1674callback, free it. Also, use no error checking, as usual.
1573 1675
1574 static void 1676 static void
… …
1646parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set> 1748parameters of any kind. There are C<ev_prepare_set> and C<ev_check_set>
1647macros, but using them is utterly, utterly and completely pointless. 1749macros, but using them is utterly, utterly and completely pointless.
1648 1750
1649=back 1751=back
1650 1752
1753=head3 Examples
1754
1651There are a number of principal ways to embed other event loops or modules 1755There are a number of principal ways to embed other event loops or modules
1652into libev. Here are some ideas on how to include libadns into libev 1756into libev. Here are some ideas on how to include libadns into libev
1653(there is a Perl module named C<EV::ADNS> that does this, which you could 1757(there is a Perl module named C<EV::ADNS> that does this, which you could
1654use for an actually working example. Another Perl module named C<EV::Glib> 1758use for an actually working example. Another Perl module named C<EV::Glib>
1655embeds a Glib main context into libev, and finally, C<Glib::EV> embeds EV 1759embeds a Glib main context into libev, and finally, C<Glib::EV> embeds EV
… …
1823portable one. 1927portable one.
1824 1928
1825So when you want to use this feature you will always have to be prepared 1929So when you want to use this feature you will always have to be prepared
1826that you cannot get an embeddable loop. The recommended way to get around 1930that you cannot get an embeddable loop. The recommended way to get around
1827this is to have a separate variables for your embeddable loop, try to 1931this is to have a separate variables for your embeddable loop, try to
1828create it, and if that fails, use the normal loop for everything: 1932create it, and if that fails, use the normal loop for everything.
1933
1934=head3 Watcher-Specific Functions and Data Members
1935
1936=over 4
1937
1938=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)
1939
1940=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)
1941
1942Configures the watcher to embed the given loop, which must be
1943embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1944invoked automatically, otherwise it is the responsibility of the callback
1945to invoke it (it will continue to be called until the sweep has been done,
1946if you do not want thta, you need to temporarily stop the embed watcher).
1947
1948=item ev_embed_sweep (loop, ev_embed *)
1949
1950Make a single, non-blocking sweep over the embedded loop. This works
1951similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1952apropriate way for embedded loops.
1953
1954=item struct ev_loop *other [read-only]
1955
1956The embedded event loop.
1957
1958=back
1959
1960=head3 Examples
1961
1962Example: Try to get an embeddable event loop and embed it into the default
1963event loop. If that is not possible, use the default loop. The default
1964loop is stored in C<loop_hi>, while the mebeddable loop is stored in
1965C<loop_lo> (which is C<loop_hi> in the acse no embeddable loop can be
1966used).
1829 1967
1830 struct ev_loop *loop_hi = ev_default_init (0); 1968 struct ev_loop *loop_hi = ev_default_init (0);
1831 struct ev_loop *loop_lo = 0; 1969 struct ev_loop *loop_lo = 0;
1832 struct ev_embed embed; 1970 struct ev_embed embed;
1833 1971
… …
1844 ev_embed_start (loop_hi, &embed); 1982 ev_embed_start (loop_hi, &embed);
1845 } 1983 }
1846 else 1984 else
1847 loop_lo = loop_hi; 1985 loop_lo = loop_hi;
1848 1986
1849=head3 Watcher-Specific Functions and Data Members 1987Example: Check if kqueue is available but not recommended and create
1988a kqueue backend for use with sockets (which usually work with any
1989kqueue implementation). Store the kqueue/socket-only event loop in
1990C<loop_socket>. (One might optionally use C<EVFLAG_NOENV>, too).
1850 1991
1851=over 4 1992 struct ev_loop *loop = ev_default_init (0);
1993 struct ev_loop *loop_socket = 0;
1994 struct ev_embed embed;
1995
1996 if (ev_supported_backends () & ~ev_recommended_backends () & EVBACKEND_KQUEUE)
1997 if ((loop_socket = ev_loop_new (EVBACKEND_KQUEUE))
1998 {
1999 ev_embed_init (&embed, 0, loop_socket);
2000 ev_embed_start (loop, &embed);
2001 }
1852 2002
1853=item ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop) 2003 if (!loop_socket)
2004 loop_socket = loop;
1854 2005
1855=item ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop) 2006 // now use loop_socket for all sockets, and loop for everything else
1856
1857Configures the watcher to embed the given loop, which must be
1858embeddable. If the callback is C<0>, then C<ev_embed_sweep> will be
1859invoked automatically, otherwise it is the responsibility of the callback
1860to invoke it (it will continue to be called until the sweep has been done,
1861if you do not want thta, you need to temporarily stop the embed watcher).
1862
1863=item ev_embed_sweep (loop, ev_embed *)
1864
1865Make a single, non-blocking sweep over the embedded loop. This works
1866similarly to C<ev_loop (embedded_loop, EVLOOP_NONBLOCK)>, but in the most
1867apropriate way for embedded loops.
1868
1869=item struct ev_loop *other [read-only]
1870
1871The embedded event loop.
1872
1873=back
1874 2007
1875 2008
1876=head2 C<ev_fork> - the audacity to resume the event loop after a fork 2009=head2 C<ev_fork> - the audacity to resume the event loop after a fork
1877 2010
1878Fork watchers are called when a C<fork ()> was detected (usually because 2011Fork watchers are called when a C<fork ()> was detected (usually because
… …
2357wants osf handles on win32 (this is the case when the select to 2490wants osf handles on win32 (this is the case when the select to
2358be used is the winsock select). This means that it will call 2491be used is the winsock select). This means that it will call
2359C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise, 2492C<_get_osfhandle> on the fd to convert it to an OS handle. Otherwise,
2360it is assumed that all these functions actually work on fds, even 2493it is assumed that all these functions actually work on fds, even
2361on win32. Should not be defined on non-win32 platforms. 2494on win32. Should not be defined on non-win32 platforms.
2495
2496=item EV_FD_TO_WIN32_HANDLE
2497
2498If C<EV_SELECT_IS_WINSOCKET> is enabled, then libev needs a way to map
2499file descriptors to socket handles. When not defining this symbol (the
2500default), then libev will call C<_get_osfhandle>, which is usually
2501correct. In some cases, programs use their own file descriptor management,
2502in which case they can provide this function to map fds to socket handles.
2362 2503
2363=item EV_USE_POLL 2504=item EV_USE_POLL
2364 2505
2365If defined to be C<1>, libev will compile in support for the C<poll>(2) 2506If defined to be C<1>, libev will compile in support for the C<poll>(2)
2366backend. Otherwise it will be enabled on non-win32 platforms. It 2507backend. Otherwise it will be enabled on non-win32 platforms. It
… …
2403be detected at runtime. 2544be detected at runtime.
2404 2545
2405=item EV_H 2546=item EV_H
2406 2547
2407The name of the F<ev.h> header file used to include it. The default if 2548The name of the F<ev.h> header file used to include it. The default if
2408undefined is C<< <ev.h> >> in F<event.h> and C<"ev.h"> in F<ev.c>. This 2549undefined is C<"ev.h"> in F<event.h> and F<ev.c>. This can be used to
2409can be used to virtually rename the F<ev.h> header file in case of conflicts. 2550virtually rename the F<ev.h> header file in case of conflicts.
2410 2551
2411=item EV_CONFIG_H 2552=item EV_CONFIG_H
2412 2553
2413If C<EV_STANDALONE> isn't C<1>, this variable can be used to override 2554If C<EV_STANDALONE> isn't C<1>, this variable can be used to override
2414F<ev.c>'s idea of where to find the F<config.h> file, similarly to 2555F<ev.c>'s idea of where to find the F<config.h> file, similarly to
2415C<EV_H>, above. 2556C<EV_H>, above.
2416 2557
2417=item EV_EVENT_H 2558=item EV_EVENT_H
2418 2559
2419Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea 2560Similarly to C<EV_H>, this macro can be used to override F<event.c>'s idea
2420of how the F<event.h> header can be found. 2561of how the F<event.h> header can be found, the dfeault is C<"event.h">.
2421 2562
2422=item EV_PROTOTYPES 2563=item EV_PROTOTYPES
2423 2564
2424If defined to be C<0>, then F<ev.h> will not define any function 2565If defined to be C<0>, then F<ev.h> will not define any function
2425prototypes, but still define all the structs and other symbols. This is 2566prototypes, but still define all the structs and other symbols. This is
… …
2491than enough. If you need to manage thousands of children you might want to 2632than enough. If you need to manage thousands of children you might want to
2492increase this value (I<must> be a power of two). 2633increase this value (I<must> be a power of two).
2493 2634
2494=item EV_INOTIFY_HASHSIZE 2635=item EV_INOTIFY_HASHSIZE
2495 2636
2496C<ev_staz> watchers use a small hash table to distribute workload by 2637C<ev_stat> watchers use a small hash table to distribute workload by
2497inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>), 2638inotify watch id. The default size is C<16> (or C<1> with C<EV_MINIMAL>),
2498usually more than enough. If you need to manage thousands of C<ev_stat> 2639usually more than enough. If you need to manage thousands of C<ev_stat>
2499watchers you might want to increase this value (I<must> be a power of 2640watchers you might want to increase this value (I<must> be a power of
2500two). 2641two).
2501 2642
… …
2597 2738
2598=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers) 2739=item Starting and stopping timer/periodic watchers: O(log skipped_other_timers)
2599 2740
2600This means that, when you have a watcher that triggers in one hour and 2741This means that, when you have a watcher that triggers in one hour and
2601there are 100 watchers that would trigger before that then inserting will 2742there are 100 watchers that would trigger before that then inserting will
2602have to skip those 100 watchers. 2743have to skip roughly seven (C<ld 100>) of these watchers.
2603 2744
2604=item Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers) 2745=item Changing timer/periodic watchers (by autorepeat or calling again): O(log skipped_other_timers)
2605 2746
2606That means that for changing a timer costs less than removing/adding them 2747That means that changing a timer costs less than removing/adding them
2607as only the relative motion in the event queue has to be paid for. 2748as only the relative motion in the event queue has to be paid for.
2608 2749
2609=item Starting io/check/prepare/idle/signal/child watchers: O(1) 2750=item Starting io/check/prepare/idle/signal/child watchers: O(1)
2610 2751
2611These just add the watcher into an array or at the head of a list. 2752These just add the watcher into an array or at the head of a list.
2753
2612=item Stopping check/prepare/idle watchers: O(1) 2754=item Stopping check/prepare/idle watchers: O(1)
2613 2755
2614=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE)) 2756=item Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))
2615 2757
2616These watchers are stored in lists then need to be walked to find the 2758These watchers are stored in lists then need to be walked to find the
2617correct watcher to remove. The lists are usually short (you don't usually 2759correct watcher to remove. The lists are usually short (you don't usually
2618have many watchers waiting for the same fd or signal). 2760have many watchers waiting for the same fd or signal).
2619 2761
2620=item Finding the next timer per loop iteration: O(1) 2762=item Finding the next timer in each loop iteration: O(1)
2763
2764By virtue of using a binary heap, the next timer is always found at the
2765beginning of the storage array.
2621 2766
2622=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd) 2767=item Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)
2623 2768
2624A change means an I/O watcher gets started or stopped, which requires 2769A change means an I/O watcher gets started or stopped, which requires
2625libev to recalculate its status (and possibly tell the kernel). 2770libev to recalculate its status (and possibly tell the kernel, depending
2771on backend and wether C<ev_io_set> was used).
2626 2772
2627=item Activating one watcher: O(1) 2773=item Activating one watcher (putting it into the pending state): O(1)
2628 2774
2629=item Priority handling: O(number_of_priorities) 2775=item Priority handling: O(number_of_priorities)
2630 2776
2631Priorities are implemented by allocating some space for each 2777Priorities are implemented by allocating some space for each
2632priority. When doing priority-based operations, libev usually has to 2778priority. When doing priority-based operations, libev usually has to
2633linearly search all the priorities. 2779linearly search all the priorities, but starting/stopping and activating
2780watchers becomes O(1) w.r.t. prioritiy handling.
2634 2781
2635=back 2782=back
2636 2783
2637 2784
2785=head1 Win32 platform limitations and workarounds
2786
2787Win32 doesn't support any of the standards (e.g. POSIX) that libev
2788requires, and its I/O model is fundamentally incompatible with the POSIX
2789model. Libev still offers limited functionality on this platform in
2790the form of the C<EVBACKEND_SELECT> backend, and only supports socket
2791descriptors. This only applies when using Win32 natively, not when using
2792e.g. cygwin.
2793
2794There is no supported compilation method available on windows except
2795embedding it into other applications.
2796
2797Due to the many, low, and arbitrary limits on the win32 platform and the
2798abysmal performance of winsockets, using a large number of sockets is not
2799recommended (and not reasonable). If your program needs to use more than
2800a hundred or so sockets, then likely it needs to use a totally different
2801implementation for windows, as libev offers the POSIX model, which cannot
2802be implemented efficiently on windows (microsoft monopoly games).
2803
2804=over 4
2805
2806=item The winsocket select function
2807
2808The winsocket C<select> function doesn't follow POSIX in that it requires
2809socket I<handles> and not socket I<file descriptors>. This makes select
2810very inefficient, and also requires a mapping from file descriptors
2811to socket handles. See the discussion of the C<EV_SELECT_USE_FD_SET>,
2812C<EV_SELECT_IS_WINSOCKET> and C<EV_FD_TO_WIN32_HANDLE> preprocessor
2813symbols for more info.
2814
2815The configuration for a "naked" win32 using the microsoft runtime
2816libraries and raw winsocket select is:
2817
2818 #define EV_USE_SELECT 1
2819 #define EV_SELECT_IS_WINSOCKET 1 /* forces EV_SELECT_USE_FD_SET, too */
2820
2821Note that winsockets handling of fd sets is O(n), so you can easily get a
2822complexity in the O(n²) range when using win32.
2823
2824=item Limited number of file descriptors
2825
2826Windows has numerous arbitrary (and low) limits on things. Early versions
2827of winsocket's select only supported waiting for a max. of C<64> handles
2828(probably owning to the fact that all windows kernels can only wait for
2829C<64> things at the same time internally; microsoft recommends spawning a
2830chain of threads and wait for 63 handles and the previous thread in each).
2831
2832Newer versions support more handles, but you need to define C<FD_SETSIZE>
2833to some high number (e.g. C<2048>) before compiling the winsocket select
2834call (which might be in libev or elsewhere, for example, perl does its own
2835select emulation on windows).
2836
2837Another limit is the number of file descriptors in the microsoft runtime
2838libraries, which by default is C<64> (there must be a hidden I<64> fetish
2839or something like this inside microsoft). You can increase this by calling
2840C<_setmaxstdio>, which can increase this limit to C<2048> (another
2841arbitrary limit), but is broken in many versions of the microsoft runtime
2842libraries.
2843
2844This might get you to about C<512> or C<2048> sockets (depending on
2845windows version and/or the phase of the moon). To get more, you need to
2846wrap all I/O functions and provide your own fd management, but the cost of
2847calling select (O(n²)) will likely make this unworkable.
2848
2849=back
2850
2851
2638=head1 AUTHOR 2852=head1 AUTHOR
2639 2853
2640Marc Lehmann <libev@schmorp.de>. 2854Marc Lehmann <libev@schmorp.de>.
2641 2855

Diff Legend

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