Mirror of :pserver:anonymous@cvs.schmorp.de/schmorpforge libev http://software.schmorp.de/pkg/libev.html
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

1696 lines
76 KiB

14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
14 years ago
  1. <?xml version="1.0" encoding="UTF-8"?>
  2. <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
  3. <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
  4. <head>
  5. <title>libev</title>
  6. <meta name="description" content="Pod documentation for libev" />
  7. <meta name="inputfile" content="&lt;standard input&gt;" />
  8. <meta name="outputfile" content="&lt;standard output&gt;" />
  9. <meta name="created" content="Sat Nov 24 11:19:13 2007" />
  10. <meta name="generator" content="Pod::Xhtml 1.57" />
  11. <link rel="stylesheet" href="http://res.tst.eu/pod.css"/></head>
  12. <body>
  13. <div class="pod">
  14. <!-- INDEX START -->
  15. <h3 id="TOP">Index</h3>
  16. <ul><li><a href="#NAME">NAME</a></li>
  17. <li><a href="#SYNOPSIS">SYNOPSIS</a></li>
  18. <li><a href="#DESCRIPTION">DESCRIPTION</a></li>
  19. <li><a href="#FEATURES">FEATURES</a></li>
  20. <li><a href="#CONVENTIONS">CONVENTIONS</a></li>
  21. <li><a href="#TIME_REPRESENTATION">TIME REPRESENTATION</a></li>
  22. <li><a href="#GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</a></li>
  23. <li><a href="#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</a></li>
  24. <li><a href="#ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</a>
  25. <ul><li><a href="#SUMMARY_OF_GENERIC_WATCHER_FUNCTIONS">SUMMARY OF GENERIC WATCHER FUNCTIONS</a></li>
  26. <li><a href="#ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</a></li>
  27. </ul>
  28. </li>
  29. <li><a href="#WATCHER_TYPES">WATCHER TYPES</a>
  30. <ul><li><a href="#code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable</a></li>
  31. <li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally recurring timeouts</a></li>
  32. <li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron</a></li>
  33. <li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled</a></li>
  34. <li><a href="#code_ev_child_code_wait_for_pid_stat"><code>ev_child</code> - wait for pid status changes</a></li>
  35. <li><a href="#code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do</a></li>
  36. <li><a href="#code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop</a></li>
  37. <li><a href="#code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough</a></li>
  38. </ul>
  39. </li>
  40. <li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
  41. <li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
  42. <li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
  43. <li><a href="#EMBEDDING">EMBEDDING</a>
  44. <ul><li><a href="#FILESETS">FILESETS</a>
  45. <ul><li><a href="#CORE_EVENT_LOOP">CORE EVENT LOOP</a></li>
  46. <li><a href="#LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</a></li>
  47. <li><a href="#AUTOCONF_SUPPORT">AUTOCONF SUPPORT</a></li>
  48. </ul>
  49. </li>
  50. <li><a href="#PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</a></li>
  51. <li><a href="#EXAMPLES">EXAMPLES</a></li>
  52. </ul>
  53. </li>
  54. <li><a href="#AUTHOR">AUTHOR</a>
  55. </li>
  56. </ul><hr />
  57. <!-- INDEX END -->
  58. <h1 id="NAME">NAME</h1><p><a href="#TOP" class="toplink">Top</a></p>
  59. <div id="NAME_CONTENT">
  60. <p>libev - a high performance full-featured event loop written in C</p>
  61. </div>
  62. <h1 id="SYNOPSIS">SYNOPSIS</h1><p><a href="#TOP" class="toplink">Top</a></p>
  63. <div id="SYNOPSIS_CONTENT">
  64. <pre> #include &lt;ev.h&gt;
  65. </pre>
  66. </div>
  67. <h1 id="DESCRIPTION">DESCRIPTION</h1><p><a href="#TOP" class="toplink">Top</a></p>
  68. <div id="DESCRIPTION_CONTENT">
  69. <p>Libev is an event loop: you register interest in certain events (such as a
  70. file descriptor being readable or a timeout occuring), and it will manage
  71. these event sources and provide your program with events.</p>
  72. <p>To do this, it must take more or less complete control over your process
  73. (or thread) by executing the <i>event loop</i> handler, and will then
  74. communicate events via a callback mechanism.</p>
  75. <p>You register interest in certain events by registering so-called <i>event
  76. watchers</i>, which are relatively small C structures you initialise with the
  77. details of the event, and then hand it over to libev by <i>starting</i> the
  78. watcher.</p>
  79. </div>
  80. <h1 id="FEATURES">FEATURES</h1><p><a href="#TOP" class="toplink">Top</a></p>
  81. <div id="FEATURES_CONTENT">
  82. <p>Libev supports select, poll, the linux-specific epoll and the bsd-specific
  83. kqueue mechanisms for file descriptor events, relative timers, absolute
  84. timers with customised rescheduling, signal events, process status change
  85. events (related to SIGCHLD), and event watchers dealing with the event
  86. loop mechanism itself (idle, prepare and check watchers). It also is quite
  87. fast (see this <a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing
  88. it to libevent for example).</p>
  89. </div>
  90. <h1 id="CONVENTIONS">CONVENTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
  91. <div id="CONVENTIONS_CONTENT">
  92. <p>Libev is very configurable. In this manual the default configuration
  93. will be described, which supports multiple event loops. For more info
  94. about various configuration options please have a look at the file
  95. <cite>README.embed</cite> in the libev distribution. If libev was configured without
  96. support for multiple event loops, then all functions taking an initial
  97. argument of name <code>loop</code> (which is always of type <code>struct ev_loop *</code>)
  98. will not have this argument.</p>
  99. </div>
  100. <h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1><p><a href="#TOP" class="toplink">Top</a></p>
  101. <div id="TIME_REPRESENTATION_CONTENT">
  102. <p>Libev represents time as a single floating point number, representing the
  103. (fractional) number of seconds since the (POSIX) epoch (somewhere near
  104. the beginning of 1970, details are complicated, don't ask). This type is
  105. called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
  106. to the <code>double</code> type in C, and when you need to do any calculations on
  107. it, you should treat it as such.</p>
  108. </div>
  109. <h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
  110. <div id="GLOBAL_FUNCTIONS_CONTENT">
  111. <p>These functions can be called anytime, even before initialising the
  112. library in any way.</p>
  113. <dl>
  114. <dt>ev_tstamp ev_time ()</dt>
  115. <dd>
  116. <p>Returns the current time as libev would use it. Please note that the
  117. <code>ev_now</code> function is usually faster and also often returns the timestamp
  118. you actually want to know.</p>
  119. </dd>
  120. <dt>int ev_version_major ()</dt>
  121. <dt>int ev_version_minor ()</dt>
  122. <dd>
  123. <p>You can find out the major and minor version numbers of the library
  124. you linked against by calling the functions <code>ev_version_major</code> and
  125. <code>ev_version_minor</code>. If you want, you can compare against the global
  126. symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
  127. version of the library your program was compiled against.</p>
  128. <p>Usually, it's a good idea to terminate if the major versions mismatch,
  129. as this indicates an incompatible change. Minor versions are usually
  130. compatible to older versions, so a larger minor version alone is usually
  131. not a problem.</p>
  132. <p>Example: make sure we haven't accidentally been linked against the wrong
  133. version:</p>
  134. <pre> assert ((&quot;libev version mismatch&quot;,
  135. ev_version_major () == EV_VERSION_MAJOR
  136. &amp;&amp; ev_version_minor () &gt;= EV_VERSION_MINOR));
  137. </pre>
  138. </dd>
  139. <dt>unsigned int ev_supported_backends ()</dt>
  140. <dd>
  141. <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
  142. value) compiled into this binary of libev (independent of their
  143. availability on the system you are running on). See <code>ev_default_loop</code> for
  144. a description of the set values.</p>
  145. <p>Example: make sure we have the epoll method, because yeah this is cool and
  146. a must have and can we have a torrent of it please!!!11</p>
  147. <pre> assert ((&quot;sorry, no epoll, no sex&quot;,
  148. ev_supported_backends () &amp; EVBACKEND_EPOLL));
  149. </pre>
  150. </dd>
  151. <dt>unsigned int ev_recommended_backends ()</dt>
  152. <dd>
  153. <p>Return the set of all backends compiled into this binary of libev and also
  154. recommended for this platform. This set is often smaller than the one
  155. returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
  156. most BSDs and will not be autodetected unless you explicitly request it
  157. (assuming you know what you are doing). This is the set of backends that
  158. libev will probe for if you specify no backends explicitly.</p>
  159. </dd>
  160. <dt>unsigned int ev_embeddable_backends ()</dt>
  161. <dd>
  162. <p>Returns the set of backends that are embeddable in other event loops. This
  163. is the theoretical, all-platform, value. To find which backends
  164. might be supported on the current system, you would need to look at
  165. <code>ev_embeddable_backends () &amp; ev_supported_backends ()</code>, likewise for
  166. recommended ones.</p>
  167. <p>See the description of <code>ev_embed</code> watchers for more info.</p>
  168. </dd>
  169. <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
  170. <dd>
  171. <p>Sets the allocation function to use (the prototype is similar to the
  172. realloc C function, the semantics are identical). It is used to allocate
  173. and free memory (no surprises here). If it returns zero when memory
  174. needs to be allocated, the library might abort or take some potentially
  175. destructive action. The default is your system realloc function.</p>
  176. <p>You could override this function in high-availability programs to, say,
  177. free some memory if it cannot allocate memory, to use a special allocator,
  178. or even to sleep a while and retry until some memory is available.</p>
  179. <p>Example: replace the libev allocator with one that waits a bit and then
  180. retries: better than mine).</p>
  181. <pre> static void *
  182. persistent_realloc (void *ptr, long size)
  183. {
  184. for (;;)
  185. {
  186. void *newptr = realloc (ptr, size);
  187. if (newptr)
  188. return newptr;
  189. sleep (60);
  190. }
  191. }
  192. ...
  193. ev_set_allocator (persistent_realloc);
  194. </pre>
  195. </dd>
  196. <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
  197. <dd>
  198. <p>Set the callback function to call on a retryable syscall error (such
  199. as failed select, poll, epoll_wait). The message is a printable string
  200. indicating the system call or subsystem causing the problem. If this
  201. callback is set, then libev will expect it to remedy the sitution, no
  202. matter what, when it returns. That is, libev will generally retry the
  203. requested operation, or, if the condition doesn't go away, do bad stuff
  204. (such as abort).</p>
  205. <p>Example: do the same thing as libev does internally:</p>
  206. <pre> static void
  207. fatal_error (const char *msg)
  208. {
  209. perror (msg);
  210. abort ();
  211. }
  212. ...
  213. ev_set_syserr_cb (fatal_error);
  214. </pre>
  215. </dd>
  216. </dl>
  217. </div>
  218. <h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1><p><a href="#TOP" class="toplink">Top</a></p>
  219. <div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
  220. <p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
  221. types of such loops, the <i>default</i> loop, which supports signals and child
  222. events, and dynamically created loops which do not.</p>
  223. <p>If you use threads, a common model is to run the default event loop
  224. in your main thread (or in a separate thread) and for each thread you
  225. create, you also create another event loop. Libev itself does no locking
  226. whatsoever, so if you mix calls to the same event loop in different
  227. threads, make sure you lock (this is usually a bad idea, though, even if
  228. done correctly, because it's hideous and inefficient).</p>
  229. <dl>
  230. <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
  231. <dd>
  232. <p>This will initialise the default event loop if it hasn't been initialised
  233. yet and return it. If the default loop could not be initialised, returns
  234. false. If it already was initialised it simply returns it (and ignores the
  235. flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
  236. <p>If you don't know what event loop to use, use the one returned from this
  237. function.</p>
  238. <p>The flags argument can be used to specify special behaviour or specific
  239. backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
  240. <p>The following flags are supported:</p>
  241. <p>
  242. <dl>
  243. <dt><code>EVFLAG_AUTO</code></dt>
  244. <dd>
  245. <p>The default flags value. Use this if you have no clue (it's the right
  246. thing, believe me).</p>
  247. </dd>
  248. <dt><code>EVFLAG_NOENV</code></dt>
  249. <dd>
  250. <p>If this flag bit is ored into the flag value (or the program runs setuid
  251. or setgid) then libev will <i>not</i> look at the environment variable
  252. <code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
  253. override the flags completely if it is found in the environment. This is
  254. useful to try out specific backends to test their performance, or to work
  255. around bugs.</p>
  256. </dd>
  257. <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
  258. <dd>
  259. <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
  260. libev tries to roll its own fd_set with no limits on the number of fds,
  261. but if that fails, expect a fairly low limit on the number of fds when
  262. using this backend. It doesn't scale too well (O(highest_fd)), but its usually
  263. the fastest backend for a low number of fds.</p>
  264. </dd>
  265. <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
  266. <dd>
  267. <p>And this is your standard poll(2) backend. It's more complicated than
  268. select, but handles sparse fds better and has no artificial limit on the
  269. number of fds you can use (except it will slow down considerably with a
  270. lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
  271. </dd>
  272. <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
  273. <dd>
  274. <p>For few fds, this backend is a bit little slower than poll and select,
  275. but it scales phenomenally better. While poll and select usually scale like
  276. O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
  277. either O(1) or O(active_fds).</p>
  278. <p>While stopping and starting an I/O watcher in the same iteration will
  279. result in some caching, there is still a syscall per such incident
  280. (because the fd could point to a different file description now), so its
  281. best to avoid that. Also, dup()ed file descriptors might not work very
  282. well if you register events for both fds.</p>
  283. <p>Please note that epoll sometimes generates spurious notifications, so you
  284. need to use non-blocking I/O or other means to avoid blocking when no data
  285. (or space) is available.</p>
  286. </dd>
  287. <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
  288. <dd>
  289. <p>Kqueue deserves special mention, as at the time of this writing, it
  290. was broken on all BSDs except NetBSD (usually it doesn't work with
  291. anything but sockets and pipes, except on Darwin, where of course its
  292. completely useless). For this reason its not being &quot;autodetected&quot;
  293. unless you explicitly specify it explicitly in the flags (i.e. using
  294. <code>EVBACKEND_KQUEUE</code>).</p>
  295. <p>It scales in the same way as the epoll backend, but the interface to the
  296. kernel is more efficient (which says nothing about its actual speed, of
  297. course). While starting and stopping an I/O watcher does not cause an
  298. extra syscall as with epoll, it still adds up to four event changes per
  299. incident, so its best to avoid that.</p>
  300. </dd>
  301. <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
  302. <dd>
  303. <p>This is not implemented yet (and might never be).</p>
  304. </dd>
  305. <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
  306. <dd>
  307. <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
  308. it's really slow, but it still scales very well (O(active_fds)).</p>
  309. <p>Please note that solaris ports can result in a lot of spurious
  310. notifications, so you need to use non-blocking I/O or other means to avoid
  311. blocking when no data (or space) is available.</p>
  312. </dd>
  313. <dt><code>EVBACKEND_ALL</code></dt>
  314. <dd>
  315. <p>Try all backends (even potentially broken ones that wouldn't be tried
  316. with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
  317. <code>EVBACKEND_ALL &amp; ~EVBACKEND_KQUEUE</code>.</p>
  318. </dd>
  319. </dl>
  320. </p>
  321. <p>If one or more of these are ored into the flags value, then only these
  322. backends will be tried (in the reverse order as given here). If none are
  323. specified, most compiled-in backend will be tried, usually in reverse
  324. order of their flag values :)</p>
  325. <p>The most typical usage is like this:</p>
  326. <pre> if (!ev_default_loop (0))
  327. fatal (&quot;could not initialise libev, bad $LIBEV_FLAGS in environment?&quot;);
  328. </pre>
  329. <p>Restrict libev to the select and poll backends, and do not allow
  330. environment settings to be taken into account:</p>
  331. <pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
  332. </pre>
  333. <p>Use whatever libev has to offer, but make sure that kqueue is used if
  334. available (warning, breaks stuff, best use only with your own private
  335. event loop and only if you know the OS supports your types of fds):</p>
  336. <pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
  337. </pre>
  338. </dd>
  339. <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
  340. <dd>
  341. <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
  342. always distinct from the default loop. Unlike the default loop, it cannot
  343. handle signal and child watchers, and attempts to do so will be greeted by
  344. undefined behaviour (or a failed assertion if assertions are enabled).</p>
  345. <p>Example: try to create a event loop that uses epoll and nothing else.</p>
  346. <pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
  347. if (!epoller)
  348. fatal (&quot;no epoll found here, maybe it hides under your chair&quot;);
  349. </pre>
  350. </dd>
  351. <dt>ev_default_destroy ()</dt>
  352. <dd>
  353. <p>Destroys the default loop again (frees all memory and kernel state
  354. etc.). None of the active event watchers will be stopped in the normal
  355. sense, so e.g. <code>ev_is_active</code> might still return true. It is your
  356. responsibility to either stop all watchers cleanly yoursef <i>before</i>
  357. calling this function, or cope with the fact afterwards (which is usually
  358. the easiest thing, youc na just ignore the watchers and/or <code>free ()</code> them
  359. for example).</p>
  360. </dd>
  361. <dt>ev_loop_destroy (loop)</dt>
  362. <dd>
  363. <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
  364. earlier call to <code>ev_loop_new</code>.</p>
  365. </dd>
  366. <dt>ev_default_fork ()</dt>
  367. <dd>
  368. <p>This function reinitialises the kernel state for backends that have
  369. one. Despite the name, you can call it anytime, but it makes most sense
  370. after forking, in either the parent or child process (or both, but that
  371. again makes little sense).</p>
  372. <p>You <i>must</i> call this function in the child process after forking if and
  373. only if you want to use the event library in both processes. If you just
  374. fork+exec, you don't have to call it.</p>
  375. <p>The function itself is quite fast and it's usually not a problem to call
  376. it just in case after a fork. To make this easy, the function will fit in
  377. quite nicely into a call to <code>pthread_atfork</code>:</p>
  378. <pre> pthread_atfork (0, 0, ev_default_fork);
  379. </pre>
  380. <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
  381. without calling this function, so if you force one of those backends you
  382. do not need to care.</p>
  383. </dd>
  384. <dt>ev_loop_fork (loop)</dt>
  385. <dd>
  386. <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
  387. <code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
  388. after fork, and how you do this is entirely your own problem.</p>
  389. </dd>
  390. <dt>unsigned int ev_backend (loop)</dt>
  391. <dd>
  392. <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
  393. use.</p>
  394. </dd>
  395. <dt>ev_tstamp ev_now (loop)</dt>
  396. <dd>
  397. <p>Returns the current &quot;event loop time&quot;, which is the time the event loop
  398. received events and started processing them. This timestamp does not
  399. change as long as callbacks are being processed, and this is also the base
  400. time used for relative timers. You can treat it as the timestamp of the
  401. event occuring (or more correctly, libev finding out about it).</p>
  402. </dd>
  403. <dt>ev_loop (loop, int flags)</dt>
  404. <dd>
  405. <p>Finally, this is it, the event handler. This function usually is called
  406. after you initialised all your watchers and you want to start handling
  407. events.</p>
  408. <p>If the flags argument is specified as <code>0</code>, it will not return until
  409. either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
  410. <p>Please note that an explicit <code>ev_unloop</code> is usually better than
  411. relying on all watchers to be stopped when deciding when a program has
  412. finished (especially in interactive programs), but having a program that
  413. automatically loops as long as it has to and no longer by virtue of
  414. relying on its watchers stopping correctly is a thing of beauty.</p>
  415. <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
  416. those events and any outstanding ones, but will not block your process in
  417. case there are no events and will return after one iteration of the loop.</p>
  418. <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
  419. neccessary) and will handle those and any outstanding ones. It will block
  420. your process until at least one new event arrives, and will return after
  421. one iteration of the loop. This is useful if you are waiting for some
  422. external event in conjunction with something not expressible using other
  423. libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
  424. usually a better approach for this kind of thing.</p>
  425. <p>Here are the gory details of what <code>ev_loop</code> does:</p>
  426. <pre> * If there are no active watchers (reference count is zero), return.
  427. - Queue prepare watchers and then call all outstanding watchers.
  428. - If we have been forked, recreate the kernel state.
  429. - Update the kernel state with all outstanding changes.
  430. - Update the &quot;event loop time&quot;.
  431. - Calculate for how long to block.
  432. - Block the process, waiting for any events.
  433. - Queue all outstanding I/O (fd) events.
  434. - Update the &quot;event loop time&quot; and do time jump handling.
  435. - Queue all outstanding timers.
  436. - Queue all outstanding periodics.
  437. - If no events are pending now, queue all idle watchers.
  438. - Queue all check watchers.
  439. - Call all queued watchers in reverse order (i.e. check watchers first).
  440. Signals and child watchers are implemented as I/O watchers, and will
  441. be handled here by queueing them when their watcher gets executed.
  442. - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
  443. were used, return, otherwise continue with step *.
  444. </pre>
  445. <p>Example: queue some jobs and then loop until no events are outsanding
  446. anymore.</p>
  447. <pre> ... queue jobs here, make sure they register event watchers as long
  448. ... as they still have work to do (even an idle watcher will do..)
  449. ev_loop (my_loop, 0);
  450. ... jobs done. yeah!
  451. </pre>
  452. </dd>
  453. <dt>ev_unloop (loop, how)</dt>
  454. <dd>
  455. <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
  456. has processed all outstanding events). The <code>how</code> argument must be either
  457. <code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
  458. <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
  459. </dd>
  460. <dt>ev_ref (loop)</dt>
  461. <dt>ev_unref (loop)</dt>
  462. <dd>
  463. <p>Ref/unref can be used to add or remove a reference count on the event
  464. loop: Every watcher keeps one reference, and as long as the reference
  465. count is nonzero, <code>ev_loop</code> will not return on its own. If you have
  466. a watcher you never unregister that should not keep <code>ev_loop</code> from
  467. returning, ev_unref() after starting, and ev_ref() before stopping it. For
  468. example, libev itself uses this for its internal signal pipe: It is not
  469. visible to the libev user and should not keep <code>ev_loop</code> from exiting if
  470. no event watchers registered by it are active. It is also an excellent
  471. way to do this for generic recurring timers or from within third-party
  472. libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
  473. <p>Example: create a signal watcher, but keep it from keeping <code>ev_loop</code>
  474. running when nothing else is active.</p>
  475. <pre> struct dv_signal exitsig;
  476. ev_signal_init (&amp;exitsig, sig_cb, SIGINT);
  477. ev_signal_start (myloop, &amp;exitsig);
  478. evf_unref (myloop);
  479. </pre>
  480. <p>Example: for some weird reason, unregister the above signal handler again.</p>
  481. <pre> ev_ref (myloop);
  482. ev_signal_stop (myloop, &amp;exitsig);
  483. </pre>
  484. </dd>
  485. </dl>
  486. </div>
  487. <h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1><p><a href="#TOP" class="toplink">Top</a></p>
  488. <div id="ANATOMY_OF_A_WATCHER_CONTENT">
  489. <p>A watcher is a structure that you create and register to record your
  490. interest in some event. For instance, if you want to wait for STDIN to
  491. become readable, you would create an <code>ev_io</code> watcher for that:</p>
  492. <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
  493. {
  494. ev_io_stop (w);
  495. ev_unloop (loop, EVUNLOOP_ALL);
  496. }
  497. struct ev_loop *loop = ev_default_loop (0);
  498. struct ev_io stdin_watcher;
  499. ev_init (&amp;stdin_watcher, my_cb);
  500. ev_io_set (&amp;stdin_watcher, STDIN_FILENO, EV_READ);
  501. ev_io_start (loop, &amp;stdin_watcher);
  502. ev_loop (loop, 0);
  503. </pre>
  504. <p>As you can see, you are responsible for allocating the memory for your
  505. watcher structures (and it is usually a bad idea to do this on the stack,
  506. although this can sometimes be quite valid).</p>
  507. <p>Each watcher structure must be initialised by a call to <code>ev_init
  508. (watcher *, callback)</code>, which expects a callback to be provided. This
  509. callback gets invoked each time the event occurs (or, in the case of io
  510. watchers, each time the event loop detects that the file descriptor given
  511. is readable and/or writable).</p>
  512. <p>Each watcher type has its own <code>ev_&lt;type&gt;_set (watcher *, ...)</code> macro
  513. with arguments specific to this watcher type. There is also a macro
  514. to combine initialisation and setting in one call: <code>ev_&lt;type&gt;_init
  515. (watcher *, callback, ...)</code>.</p>
  516. <p>To make the watcher actually watch out for events, you have to start it
  517. with a watcher-specific start function (<code>ev_&lt;type&gt;_start (loop, watcher
  518. *)</code>), and you can stop watching for events at any time by calling the
  519. corresponding stop function (<code>ev_&lt;type&gt;_stop (loop, watcher *)</code>.</p>
  520. <p>As long as your watcher is active (has been started but not stopped) you
  521. must not touch the values stored in it. Most specifically you must never
  522. reinitialise it or call its <code>set</code> macro.</p>
  523. <p>Each and every callback receives the event loop pointer as first, the
  524. registered watcher structure as second, and a bitset of received events as
  525. third argument.</p>
  526. <p>The received events usually include a single bit per event type received
  527. (you can receive multiple events at the same time). The possible bit masks
  528. are:</p>
  529. <dl>
  530. <dt><code>EV_READ</code></dt>
  531. <dt><code>EV_WRITE</code></dt>
  532. <dd>
  533. <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
  534. writable.</p>
  535. </dd>
  536. <dt><code>EV_TIMEOUT</code></dt>
  537. <dd>
  538. <p>The <code>ev_timer</code> watcher has timed out.</p>
  539. </dd>
  540. <dt><code>EV_PERIODIC</code></dt>
  541. <dd>
  542. <p>The <code>ev_periodic</code> watcher has timed out.</p>
  543. </dd>
  544. <dt><code>EV_SIGNAL</code></dt>
  545. <dd>
  546. <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
  547. </dd>
  548. <dt><code>EV_CHILD</code></dt>
  549. <dd>
  550. <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
  551. </dd>
  552. <dt><code>EV_IDLE</code></dt>
  553. <dd>
  554. <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
  555. </dd>
  556. <dt><code>EV_PREPARE</code></dt>
  557. <dt><code>EV_CHECK</code></dt>
  558. <dd>
  559. <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
  560. to gather new events, and all <code>ev_check</code> watchers are invoked just after
  561. <code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
  562. received events. Callbacks of both watcher types can start and stop as
  563. many watchers as they want, and all of them will be taken into account
  564. (for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
  565. <code>ev_loop</code> from blocking).</p>
  566. </dd>
  567. <dt><code>EV_ERROR</code></dt>
  568. <dd>
  569. <p>An unspecified error has occured, the watcher has been stopped. This might
  570. happen because the watcher could not be properly started because libev
  571. ran out of memory, a file descriptor was found to be closed or any other
  572. problem. You best act on it by reporting the problem and somehow coping
  573. with the watcher being stopped.</p>
  574. <p>Libev will usually signal a few &quot;dummy&quot; events together with an error,
  575. for example it might indicate that a fd is readable or writable, and if
  576. your callbacks is well-written it can just attempt the operation and cope
  577. with the error from read() or write(). This will not work in multithreaded
  578. programs, though, so beware.</p>
  579. </dd>
  580. </dl>
  581. </div>
  582. <h2 id="SUMMARY_OF_GENERIC_WATCHER_FUNCTIONS">SUMMARY OF GENERIC WATCHER FUNCTIONS</h2>
  583. <div id="SUMMARY_OF_GENERIC_WATCHER_FUNCTIONS-2">
  584. <p>In the following description, <code>TYPE</code> stands for the watcher type,
  585. e.g. <code>timer</code> for <code>ev_timer</code> watchers and <code>io</code> for <code>ev_io</code> watchers.</p>
  586. <dl>
  587. <dt><code>ev_init</code> (ev_TYPE *watcher, callback)</dt>
  588. <dd>
  589. <p>This macro initialises the generic portion of a watcher. The contents
  590. of the watcher object can be arbitrary (so <code>malloc</code> will do). Only
  591. the generic parts of the watcher are initialised, you <i>need</i> to call
  592. the type-specific <code>ev_TYPE_set</code> macro afterwards to initialise the
  593. type-specific parts. For each type there is also a <code>ev_TYPE_init</code> macro
  594. which rolls both calls into one.</p>
  595. <p>You can reinitialise a watcher at any time as long as it has been stopped
  596. (or never started) and there are no pending events outstanding.</p>
  597. <p>The callbakc is always of type <code>void (*)(ev_loop *loop, ev_TYPE *watcher,
  598. int revents)</code>.</p>
  599. </dd>
  600. <dt><code>ev_TYPE_set</code> (ev_TYPE *, [args])</dt>
  601. <dd>
  602. <p>This macro initialises the type-specific parts of a watcher. You need to
  603. call <code>ev_init</code> at least once before you call this macro, but you can
  604. call <code>ev_TYPE_set</code> any number of times. You must not, however, call this
  605. macro on a watcher that is active (it can be pending, however, which is a
  606. difference to the <code>ev_init</code> macro).</p>
  607. <p>Although some watcher types do not have type-specific arguments
  608. (e.g. <code>ev_prepare</code>) you still need to call its <code>set</code> macro.</p>
  609. </dd>
  610. <dt><code>ev_TYPE_init</code> (ev_TYPE *watcher, callback, [args])</dt>
  611. <dd>
  612. <p>This convinience macro rolls both <code>ev_init</code> and <code>ev_TYPE_set</code> macro
  613. calls into a single call. This is the most convinient method to initialise
  614. a watcher. The same limitations apply, of course.</p>
  615. </dd>
  616. <dt><code>ev_TYPE_start</code> (loop *, ev_TYPE *watcher)</dt>
  617. <dd>
  618. <p>Starts (activates) the given watcher. Only active watchers will receive
  619. events. If the watcher is already active nothing will happen.</p>
  620. </dd>
  621. <dt><code>ev_TYPE_stop</code> (loop *, ev_TYPE *watcher)</dt>
  622. <dd>
  623. <p>Stops the given watcher again (if active) and clears the pending
  624. status. It is possible that stopped watchers are pending (for example,
  625. non-repeating timers are being stopped when they become pending), but
  626. <code>ev_TYPE_stop</code> ensures that the watcher is neither active nor pending. If
  627. you want to free or reuse the memory used by the watcher it is therefore a
  628. good idea to always call its <code>ev_TYPE_stop</code> function.</p>
  629. </dd>
  630. <dt>bool ev_is_active (ev_TYPE *watcher)</dt>
  631. <dd>
  632. <p>Returns a true value iff the watcher is active (i.e. it has been started
  633. and not yet been stopped). As long as a watcher is active you must not modify
  634. it.</p>
  635. </dd>
  636. <dt>bool ev_is_pending (ev_TYPE *watcher)</dt>
  637. <dd>
  638. <p>Returns a true value iff the watcher is pending, (i.e. it has outstanding
  639. events but its callback has not yet been invoked). As long as a watcher
  640. is pending (but not active) you must not call an init function on it (but
  641. <code>ev_TYPE_set</code> is safe) and you must make sure the watcher is available to
  642. libev (e.g. you cnanot <code>free ()</code> it).</p>
  643. </dd>
  644. <dt>callback = ev_cb (ev_TYPE *watcher)</dt>
  645. <dd>
  646. <p>Returns the callback currently set on the watcher.</p>
  647. </dd>
  648. <dt>ev_cb_set (ev_TYPE *watcher, callback)</dt>
  649. <dd>
  650. <p>Change the callback. You can change the callback at virtually any time
  651. (modulo threads).</p>
  652. </dd>
  653. </dl>
  654. </div>
  655. <h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
  656. <div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
  657. <p>Each watcher has, by default, a member <code>void *data</code> that you can change
  658. and read at any time, libev will completely ignore it. This can be used
  659. to associate arbitrary data with your watcher. If you need more data and
  660. don't want to allocate memory and store a pointer to it in that data
  661. member, you can also &quot;subclass&quot; the watcher type and provide your own
  662. data:</p>
  663. <pre> struct my_io
  664. {
  665. struct ev_io io;
  666. int otherfd;
  667. void *somedata;
  668. struct whatever *mostinteresting;
  669. }
  670. </pre>
  671. <p>And since your callback will be called with a pointer to the watcher, you
  672. can cast it back to your own type:</p>
  673. <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
  674. {
  675. struct my_io *w = (struct my_io *)w_;
  676. ...
  677. }
  678. </pre>
  679. <p>More interesting and less C-conformant ways of catsing your callback type
  680. have been omitted....</p>
  681. </div>
  682. <h1 id="WATCHER_TYPES">WATCHER TYPES</h1><p><a href="#TOP" class="toplink">Top</a></p>
  683. <div id="WATCHER_TYPES_CONTENT">
  684. <p>This section describes each watcher in detail, but will not repeat
  685. information given in the last section.</p>
  686. </div>
  687. <h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable</h2>
  688. <div id="code_ev_io_code_is_this_file_descrip-2">
  689. <p>I/O watchers check whether a file descriptor is readable or writable
  690. in each iteration of the event loop (This behaviour is called
  691. level-triggering because you keep receiving events as long as the
  692. condition persists. Remember you can stop the watcher if you don't want to
  693. act on the event and neither want to receive future events).</p>
  694. <p>In general you can register as many read and/or write event watchers per
  695. fd as you want (as long as you don't confuse yourself). Setting all file
  696. descriptors to non-blocking mode is also usually a good idea (but not
  697. required if you know what you are doing).</p>
  698. <p>You have to be careful with dup'ed file descriptors, though. Some backends
  699. (the linux epoll backend is a notable example) cannot handle dup'ed file
  700. descriptors correctly if you register interest in two or more fds pointing
  701. to the same underlying file/socket etc. description (that is, they share
  702. the same underlying &quot;file open&quot;).</p>
  703. <p>If you must do this, then force the use of a known-to-be-good backend
  704. (at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
  705. <code>EVBACKEND_POLL</code>).</p>
  706. <dl>
  707. <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
  708. <dt>ev_io_set (ev_io *, int fd, int events)</dt>
  709. <dd>
  710. <p>Configures an <code>ev_io</code> watcher. The fd is the file descriptor to rceeive
  711. events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_READ |
  712. EV_WRITE</code> to receive the given events.</p>
  713. <p>Please note that most of the more scalable backend mechanisms (for example
  714. epoll and solaris ports) can result in spurious readyness notifications
  715. for file descriptors, so you practically need to use non-blocking I/O (and
  716. treat callback invocation as hint only), or retest separately with a safe
  717. interface before doing I/O (XLib can do this), or force the use of either
  718. <code>EVBACKEND_SELECT</code> or <code>EVBACKEND_POLL</code>, which don't suffer from this
  719. problem. Also note that it is quite easy to have your callback invoked
  720. when the readyness condition is no longer valid even when employing
  721. typical ways of handling events, so its a good idea to use non-blocking
  722. I/O unconditionally.</p>
  723. </dd>
  724. </dl>
  725. <p>Example: call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
  726. readable, but only once. Since it is likely line-buffered, you could
  727. attempt to read a whole line in the callback:</p>
  728. <pre> static void
  729. stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
  730. {
  731. ev_io_stop (loop, w);
  732. .. read from stdin here (or from w-&gt;fd) and haqndle any I/O errors
  733. }
  734. ...
  735. struct ev_loop *loop = ev_default_init (0);
  736. struct ev_io stdin_readable;
  737. ev_io_init (&amp;stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
  738. ev_io_start (loop, &amp;stdin_readable);
  739. ev_loop (loop, 0);
  740. </pre>
  741. </div>
  742. <h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally recurring timeouts</h2>
  743. <div id="code_ev_timer_code_relative_and_opti-2">
  744. <p>Timer watchers are simple relative timers that generate an event after a
  745. given time, and optionally repeating in regular intervals after that.</p>
  746. <p>The timers are based on real time, that is, if you register an event that
  747. times out after an hour and you reset your system clock to last years
  748. time, it will still time out after (roughly) and hour. &quot;Roughly&quot; because
  749. detecting time jumps is hard, and some inaccuracies are unavoidable (the
  750. monotonic clock option helps a lot here).</p>
  751. <p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
  752. time. This is usually the right thing as this timestamp refers to the time
  753. of the event triggering whatever timeout you are modifying/starting. If
  754. you suspect event processing to be delayed and you <i>need</i> to base the timeout
  755. on the current time, use something like this to adjust for this:</p>
  756. <pre> ev_timer_set (&amp;timer, after + ev_now () - ev_time (), 0.);
  757. </pre>
  758. <p>The callback is guarenteed to be invoked only when its timeout has passed,
  759. but if multiple timers become ready during the same loop iteration then
  760. order of execution is undefined.</p>
  761. <dl>
  762. <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
  763. <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
  764. <dd>
  765. <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
  766. <code>0.</code>, then it will automatically be stopped. If it is positive, then the
  767. timer will automatically be configured to trigger again <code>repeat</code> seconds
  768. later, again, and again, until stopped manually.</p>
  769. <p>The timer itself will do a best-effort at avoiding drift, that is, if you
  770. configure a timer to trigger every 10 seconds, then it will trigger at
  771. exactly 10 second intervals. If, however, your program cannot keep up with
  772. the timer (because it takes longer than those 10 seconds to do stuff) the
  773. timer will not fire more than once per event loop iteration.</p>
  774. </dd>
  775. <dt>ev_timer_again (loop)</dt>
  776. <dd>
  777. <p>This will act as if the timer timed out and restart it again if it is
  778. repeating. The exact semantics are:</p>
  779. <p>If the timer is started but nonrepeating, stop it.</p>
  780. <p>If the timer is repeating, either start it if necessary (with the repeat
  781. value), or reset the running timer to the repeat value.</p>
  782. <p>This sounds a bit complicated, but here is a useful and typical
  783. example: Imagine you have a tcp connection and you want a so-called idle
  784. timeout, that is, you want to be called when there have been, say, 60
  785. seconds of inactivity on the socket. The easiest way to do this is to
  786. configure an <code>ev_timer</code> with after=repeat=60 and calling ev_timer_again each
  787. time you successfully read or write some data. If you go into an idle
  788. state where you do not expect data to travel on the socket, you can stop
  789. the timer, and again will automatically restart it if need be.</p>
  790. </dd>
  791. </dl>
  792. <p>Example: create a timer that fires after 60 seconds.</p>
  793. <pre> static void
  794. one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
  795. {
  796. .. one minute over, w is actually stopped right here
  797. }
  798. struct ev_timer mytimer;
  799. ev_timer_init (&amp;mytimer, one_minute_cb, 60., 0.);
  800. ev_timer_start (loop, &amp;mytimer);
  801. </pre>
  802. <p>Example: create a timeout timer that times out after 10 seconds of
  803. inactivity.</p>
  804. <pre> static void
  805. timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
  806. {
  807. .. ten seconds without any activity
  808. }
  809. struct ev_timer mytimer;
  810. ev_timer_init (&amp;mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
  811. ev_timer_again (&amp;mytimer); /* start timer */
  812. ev_loop (loop, 0);
  813. // and in some piece of code that gets executed on any &quot;activity&quot;:
  814. // reset the timeout to start ticking again at 10 seconds
  815. ev_timer_again (&amp;mytimer);
  816. </pre>
  817. </div>
  818. <h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron</h2>
  819. <div id="code_ev_periodic_code_to_cron_or_not-2">
  820. <p>Periodic watchers are also timers of a kind, but they are very versatile
  821. (and unfortunately a bit complex).</p>
  822. <p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
  823. but on wallclock time (absolute time). You can tell a periodic watcher
  824. to trigger &quot;at&quot; some specific point in time. For example, if you tell a
  825. periodic watcher to trigger in 10 seconds (by specifiying e.g. <code>ev_now ()
  826. + 10.</code>) and then reset your system clock to the last year, then it will
  827. take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
  828. roughly 10 seconds later and of course not if you reset your system time
  829. again).</p>
  830. <p>They can also be used to implement vastly more complex timers, such as
  831. triggering an event on eahc midnight, local time.</p>
  832. <p>As with timers, the callback is guarenteed to be invoked only when the
  833. time (<code>at</code>) has been passed, but if multiple periodic timers become ready
  834. during the same loop iteration then order of execution is undefined.</p>
  835. <dl>
  836. <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
  837. <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
  838. <dd>
  839. <p>Lots of arguments, lets sort it out... There are basically three modes of
  840. operation, and we will explain them from simplest to complex:</p>
  841. <p>
  842. <dl>
  843. <dt>* absolute timer (interval = reschedule_cb = 0)</dt>
  844. <dd>
  845. <p>In this configuration the watcher triggers an event at the wallclock time
  846. <code>at</code> and doesn't repeat. It will not adjust when a time jump occurs,
  847. that is, if it is to be run at January 1st 2011 then it will run when the
  848. system time reaches or surpasses this time.</p>
  849. </dd>
  850. <dt>* non-repeating interval timer (interval &gt; 0, reschedule_cb = 0)</dt>
  851. <dd>
  852. <p>In this mode the watcher will always be scheduled to time out at the next
  853. <code>at + N * interval</code> time (for some integer N) and then repeat, regardless
  854. of any time jumps.</p>
  855. <p>This can be used to create timers that do not drift with respect to system
  856. time:</p>
  857. <pre> ev_periodic_set (&amp;periodic, 0., 3600., 0);
  858. </pre>
  859. <p>This doesn't mean there will always be 3600 seconds in between triggers,
  860. but only that the the callback will be called when the system time shows a
  861. full hour (UTC), or more correctly, when the system time is evenly divisible
  862. by 3600.</p>
  863. <p>Another way to think about it (for the mathematically inclined) is that
  864. <code>ev_periodic</code> will try to run the callback in this mode at the next possible
  865. time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
  866. </dd>
  867. <dt>* manual reschedule mode (reschedule_cb = callback)</dt>
  868. <dd>
  869. <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
  870. ignored. Instead, each time the periodic watcher gets scheduled, the
  871. reschedule callback will be called with the watcher as first, and the
  872. current time as second argument.</p>
  873. <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
  874. ever, or make any event loop modifications</i>. If you need to stop it,
  875. return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
  876. starting a prepare watcher).</p>
  877. <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
  878. ev_tstamp now)</code>, e.g.:</p>
  879. <pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
  880. {
  881. return now + 60.;
  882. }
  883. </pre>
  884. <p>It must return the next time to trigger, based on the passed time value
  885. (that is, the lowest time value larger than to the second argument). It
  886. will usually be called just before the callback will be triggered, but
  887. might be called at other times, too.</p>
  888. <p>NOTE: <i>This callback must always return a time that is later than the
  889. passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
  890. <p>This can be used to create very complex timers, such as a timer that
  891. triggers on each midnight, local time. To do this, you would calculate the
  892. next midnight after <code>now</code> and return the timestamp value for this. How
  893. you do this is, again, up to you (but it is not trivial, which is the main
  894. reason I omitted it as an example).</p>
  895. </dd>
  896. </dl>
  897. </p>
  898. </dd>
  899. <dt>ev_periodic_again (loop, ev_periodic *)</dt>
  900. <dd>
  901. <p>Simply stops and restarts the periodic watcher again. This is only useful
  902. when you changed some parameters or the reschedule callback would return
  903. a different time than the last time it was called (e.g. in a crond like
  904. program when the crontabs have changed).</p>
  905. </dd>
  906. </dl>
  907. <p>Example: call a callback every hour, or, more precisely, whenever the
  908. system clock is divisible by 3600. The callback invocation times have
  909. potentially a lot of jittering, but good long-term stability.</p>
  910. <pre> static void
  911. clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
  912. {
  913. ... its now a full hour (UTC, or TAI or whatever your clock follows)
  914. }
  915. struct ev_periodic hourly_tick;
  916. ev_periodic_init (&amp;hourly_tick, clock_cb, 0., 3600., 0);
  917. ev_periodic_start (loop, &amp;hourly_tick);
  918. </pre>
  919. <p>Example: the same as above, but use a reschedule callback to do it:</p>
  920. <pre> #include &lt;math.h&gt;
  921. static ev_tstamp
  922. my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
  923. {
  924. return fmod (now, 3600.) + 3600.;
  925. }
  926. ev_periodic_init (&amp;hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
  927. </pre>
  928. <p>Example: call a callback every hour, starting now:</p>
  929. <pre> struct ev_periodic hourly_tick;
  930. ev_periodic_init (&amp;hourly_tick, clock_cb,
  931. fmod (ev_now (loop), 3600.), 3600., 0);
  932. ev_periodic_start (loop, &amp;hourly_tick);
  933. </pre>
  934. </div>
  935. <h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled</h2>
  936. <div id="code_ev_signal_code_signal_me_when_a-2">
  937. <p>Signal watchers will trigger an event when the process receives a specific
  938. signal one or more times. Even though signals are very asynchronous, libev
  939. will try it's best to deliver signals synchronously, i.e. as part of the
  940. normal event processing, like any other event.</p>
  941. <p>You can configure as many watchers as you like per signal. Only when the
  942. first watcher gets started will libev actually register a signal watcher
  943. with the kernel (thus it coexists with your own signal handlers as long
  944. as you don't register any with libev). Similarly, when the last signal
  945. watcher for a signal is stopped libev will reset the signal handler to
  946. SIG_DFL (regardless of what it was set to before).</p>
  947. <dl>
  948. <dt>ev_signal_init (ev_signal *, callback, int signum)</dt>
  949. <dt>ev_signal_set (ev_signal *, int signum)</dt>
  950. <dd>
  951. <p>Configures the watcher to trigger on the given signal number (usually one
  952. of the <code>SIGxxx</code> constants).</p>
  953. </dd>
  954. </dl>
  955. </div>
  956. <h2 id="code_ev_child_code_wait_for_pid_stat"><code>ev_child</code> - wait for pid status changes</h2>
  957. <div id="code_ev_child_code_wait_for_pid_stat-2">
  958. <p>Child watchers trigger when your process receives a SIGCHLD in response to
  959. some child status changes (most typically when a child of yours dies).</p>
  960. <dl>
  961. <dt>ev_child_init (ev_child *, callback, int pid)</dt>
  962. <dt>ev_child_set (ev_child *, int pid)</dt>
  963. <dd>
  964. <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
  965. <i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
  966. at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
  967. the status word (use the macros from <code>sys/wait.h</code> and see your systems
  968. <code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
  969. process causing the status change.</p>
  970. </dd>
  971. </dl>
  972. <p>Example: try to exit cleanly on SIGINT and SIGTERM.</p>
  973. <pre> static void
  974. sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
  975. {
  976. ev_unloop (loop, EVUNLOOP_ALL);
  977. }
  978. struct ev_signal signal_watcher;
  979. ev_signal_init (&amp;signal_watcher, sigint_cb, SIGINT);
  980. ev_signal_start (loop, &amp;sigint_cb);
  981. </pre>
  982. </div>
  983. <h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do</h2>
  984. <div id="code_ev_idle_code_when_you_ve_got_no-2">
  985. <p>Idle watchers trigger events when there are no other events are pending
  986. (prepare, check and other idle watchers do not count). That is, as long
  987. as your process is busy handling sockets or timeouts (or even signals,
  988. imagine) it will not be triggered. But when your process is idle all idle
  989. watchers are being called again and again, once per event loop iteration -
  990. until stopped, that is, or your process receives more events and becomes
  991. busy.</p>
  992. <p>The most noteworthy effect is that as long as any idle watchers are
  993. active, the process will not block when waiting for new events.</p>
  994. <p>Apart from keeping your process non-blocking (which is a useful
  995. effect on its own sometimes), idle watchers are a good place to do
  996. &quot;pseudo-background processing&quot;, or delay processing stuff to after the
  997. event loop has handled all outstanding events.</p>
  998. <dl>
  999. <dt>ev_idle_init (ev_signal *, callback)</dt>
  1000. <dd>
  1001. <p>Initialises and configures the idle watcher - it has no parameters of any
  1002. kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
  1003. believe me.</p>
  1004. </dd>
  1005. </dl>
  1006. <p>Example: dynamically allocate an <code>ev_idle</code>, start it, and in the
  1007. callback, free it. Alos, use no error checking, as usual.</p>
  1008. <pre> static void
  1009. idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
  1010. {
  1011. free (w);
  1012. // now do something you wanted to do when the program has
  1013. // no longer asnything immediate to do.
  1014. }
  1015. struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
  1016. ev_idle_init (idle_watcher, idle_cb);
  1017. ev_idle_start (loop, idle_cb);
  1018. </pre>
  1019. </div>
  1020. <h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop</h2>
  1021. <div id="code_ev_prepare_code_and_code_ev_che-2">
  1022. <p>Prepare and check watchers are usually (but not always) used in tandem:
  1023. prepare watchers get invoked before the process blocks and check watchers
  1024. afterwards.</p>
  1025. <p>Their main purpose is to integrate other event mechanisms into libev and
  1026. their use is somewhat advanced. This could be used, for example, to track
  1027. variable changes, implement your own watchers, integrate net-snmp or a
  1028. coroutine library and lots more.</p>
  1029. <p>This is done by examining in each prepare call which file descriptors need
  1030. to be watched by the other library, registering <code>ev_io</code> watchers for
  1031. them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
  1032. provide just this functionality). Then, in the check watcher you check for
  1033. any events that occured (by checking the pending status of all watchers
  1034. and stopping them) and call back into the library. The I/O and timer
  1035. callbacks will never actually be called (but must be valid nevertheless,
  1036. because you never know, you know?).</p>
  1037. <p>As another example, the Perl Coro module uses these hooks to integrate
  1038. coroutines into libev programs, by yielding to other active coroutines
  1039. during each prepare and only letting the process block if no coroutines
  1040. are ready to run (it's actually more complicated: it only runs coroutines
  1041. with priority higher than or equal to the event loop and one coroutine
  1042. of lower priority, but only once, using idle watchers to keep the event
  1043. loop from blocking if lower-priority coroutines are active, thus mapping
  1044. low-priority coroutines to idle/background tasks).</p>
  1045. <dl>
  1046. <dt>ev_prepare_init (ev_prepare *, callback)</dt>
  1047. <dt>ev_check_init (ev_check *, callback)</dt>
  1048. <dd>
  1049. <p>Initialises and configures the prepare or check watcher - they have no
  1050. parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
  1051. macros, but using them is utterly, utterly and completely pointless.</p>
  1052. </dd>
  1053. </dl>
  1054. <p>Example: *TODO*.</p>
  1055. </div>
  1056. <h2 id="code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough</h2>
  1057. <div id="code_ev_embed_code_when_one_backend_-2">
  1058. <p>This is a rather advanced watcher type that lets you embed one event loop
  1059. into another (currently only <code>ev_io</code> events are supported in the embedded
  1060. loop, other types of watchers might be handled in a delayed or incorrect
  1061. fashion and must not be used).</p>
  1062. <p>There are primarily two reasons you would want that: work around bugs and
  1063. prioritise I/O.</p>
  1064. <p>As an example for a bug workaround, the kqueue backend might only support
  1065. sockets on some platform, so it is unusable as generic backend, but you
  1066. still want to make use of it because you have many sockets and it scales
  1067. so nicely. In this case, you would create a kqueue-based loop and embed it
  1068. into your default loop (which might use e.g. poll). Overall operation will
  1069. be a bit slower because first libev has to poll and then call kevent, but
  1070. at least you can use both at what they are best.</p>
  1071. <p>As for prioritising I/O: rarely you have the case where some fds have
  1072. to be watched and handled very quickly (with low latency), and even
  1073. priorities and idle watchers might have too much overhead. In this case
  1074. you would put all the high priority stuff in one loop and all the rest in
  1075. a second one, and embed the second one in the first.</p>
  1076. <p>As long as the watcher is active, the callback will be invoked every time
  1077. there might be events pending in the embedded loop. The callback must then
  1078. call <code>ev_embed_sweep (mainloop, watcher)</code> to make a single sweep and invoke
  1079. their callbacks (you could also start an idle watcher to give the embedded
  1080. loop strictly lower priority for example). You can also set the callback
  1081. to <code>0</code>, in which case the embed watcher will automatically execute the
  1082. embedded loop sweep.</p>
  1083. <p>As long as the watcher is started it will automatically handle events. The
  1084. callback will be invoked whenever some events have been handled. You can
  1085. set the callback to <code>0</code> to avoid having to specify one if you are not
  1086. interested in that.</p>
  1087. <p>Also, there have not currently been made special provisions for forking:
  1088. when you fork, you not only have to call <code>ev_loop_fork</code> on both loops,
  1089. but you will also have to stop and restart any <code>ev_embed</code> watchers
  1090. yourself.</p>
  1091. <p>Unfortunately, not all backends are embeddable, only the ones returned by
  1092. <code>ev_embeddable_backends</code> are, which, unfortunately, does not include any
  1093. portable one.</p>
  1094. <p>So when you want to use this feature you will always have to be prepared
  1095. that you cannot get an embeddable loop. The recommended way to get around
  1096. this is to have a separate variables for your embeddable loop, try to
  1097. create it, and if that fails, use the normal loop for everything:</p>
  1098. <pre> struct ev_loop *loop_hi = ev_default_init (0);
  1099. struct ev_loop *loop_lo = 0;
  1100. struct ev_embed embed;
  1101. // see if there is a chance of getting one that works
  1102. // (remember that a flags value of 0 means autodetection)
  1103. loop_lo = ev_embeddable_backends () &amp; ev_recommended_backends ()
  1104. ? ev_loop_new (ev_embeddable_backends () &amp; ev_recommended_backends ())
  1105. : 0;
  1106. // if we got one, then embed it, otherwise default to loop_hi
  1107. if (loop_lo)
  1108. {
  1109. ev_embed_init (&amp;embed, 0, loop_lo);
  1110. ev_embed_start (loop_hi, &amp;embed);
  1111. }
  1112. else
  1113. loop_lo = loop_hi;
  1114. </pre>
  1115. <dl>
  1116. <dt>ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
  1117. <dt>ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
  1118. <dd>
  1119. <p>Configures the watcher to embed the given loop, which must be
  1120. embeddable. If the callback is <code>0</code>, then <code>ev_embed_sweep</code> will be
  1121. invoked automatically, otherwise it is the responsibility of the callback
  1122. to invoke it (it will continue to be called until the sweep has been done,
  1123. if you do not want thta, you need to temporarily stop the embed watcher).</p>
  1124. </dd>
  1125. <dt>ev_embed_sweep (loop, ev_embed *)</dt>
  1126. <dd>
  1127. <p>Make a single, non-blocking sweep over the embedded loop. This works
  1128. similarly to <code>ev_loop (embedded_loop, EVLOOP_NONBLOCK)</code>, but in the most
  1129. apropriate way for embedded loops.</p>
  1130. </dd>
  1131. </dl>
  1132. </div>
  1133. <h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1><p><a href="#TOP" class="toplink">Top</a></p>
  1134. <div id="OTHER_FUNCTIONS_CONTENT">
  1135. <p>There are some other functions of possible interest. Described. Here. Now.</p>
  1136. <dl>
  1137. <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
  1138. <dd>
  1139. <p>This function combines a simple timer and an I/O watcher, calls your
  1140. callback on whichever event happens first and automatically stop both
  1141. watchers. This is useful if you want to wait for a single event on an fd
  1142. or timeout without having to allocate/configure/start/stop/free one or
  1143. more watchers yourself.</p>
  1144. <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
  1145. is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
  1146. <code>events</code> set will be craeted and started.</p>
  1147. <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
  1148. started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
  1149. repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
  1150. dubious value.</p>
  1151. <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
  1152. passed an <code>revents</code> set like normal event callbacks (a combination of
  1153. <code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
  1154. value passed to <code>ev_once</code>:</p>
  1155. <pre> static void stdin_ready (int revents, void *arg)
  1156. {
  1157. if (revents &amp; EV_TIMEOUT)
  1158. /* doh, nothing entered */;
  1159. else if (revents &amp; EV_READ)
  1160. /* stdin might have data for us, joy! */;
  1161. }
  1162. ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
  1163. </pre>
  1164. </dd>
  1165. <dt>ev_feed_event (ev_loop *, watcher *, int revents)</dt>
  1166. <dd>
  1167. <p>Feeds the given event set into the event loop, as if the specified event
  1168. had happened for the specified watcher (which must be a pointer to an
  1169. initialised but not necessarily started event watcher).</p>
  1170. </dd>
  1171. <dt>ev_feed_fd_event (ev_loop *, int fd, int revents)</dt>
  1172. <dd>
  1173. <p>Feed an event on the given fd, as if a file descriptor backend detected
  1174. the given events it.</p>
  1175. </dd>
  1176. <dt>ev_feed_signal_event (ev_loop *loop, int signum)</dt>
  1177. <dd>
  1178. <p>Feed an event as if the given signal occured (<code>loop</code> must be the default
  1179. loop!).</p>
  1180. </dd>
  1181. </dl>
  1182. </div>
  1183. <h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1><p><a href="#TOP" class="toplink">Top</a></p>
  1184. <div id="LIBEVENT_EMULATION_CONTENT">
  1185. <p>Libev offers a compatibility emulation layer for libevent. It cannot
  1186. emulate the internals of libevent, so here are some usage hints:</p>
  1187. <dl>
  1188. <dt>* Use it by including &lt;event.h&gt;, as usual.</dt>
  1189. <dt>* The following members are fully supported: ev_base, ev_callback,
  1190. ev_arg, ev_fd, ev_res, ev_events.</dt>
  1191. <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
  1192. maintained by libev, it does not work exactly the same way as in libevent (consider
  1193. it a private API).</dt>
  1194. <dt>* Priorities are not currently supported. Initialising priorities
  1195. will fail and all watchers will have the same priority, even though there
  1196. is an ev_pri field.</dt>
  1197. <dt>* Other members are not supported.</dt>
  1198. <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
  1199. to use the libev header file and library.</dt>
  1200. </dl>
  1201. </div>
  1202. <h1 id="C_SUPPORT">C++ SUPPORT</h1><p><a href="#TOP" class="toplink">Top</a></p>
  1203. <div id="C_SUPPORT_CONTENT">
  1204. <p>Libev comes with some simplistic wrapper classes for C++ that mainly allow
  1205. you to use some convinience methods to start/stop watchers and also change
  1206. the callback model to a model using method callbacks on objects.</p>
  1207. <p>To use it,</p>
  1208. <pre> #include &lt;ev++.h&gt;
  1209. </pre>
  1210. <p>(it is not installed by default). This automatically includes <cite>ev.h</cite>
  1211. and puts all of its definitions (many of them macros) into the global
  1212. namespace. All C++ specific things are put into the <code>ev</code> namespace.</p>
  1213. <p>It should support all the same embedding options as <cite>ev.h</cite>, most notably
  1214. <code>EV_MULTIPLICITY</code>.</p>
  1215. <p>Here is a list of things available in the <code>ev</code> namespace:</p>
  1216. <dl>
  1217. <dt><code>ev::READ</code>, <code>ev::WRITE</code> etc.</dt>
  1218. <dd>
  1219. <p>These are just enum values with the same values as the <code>EV_READ</code> etc.
  1220. macros from <cite>ev.h</cite>.</p>
  1221. </dd>
  1222. <dt><code>ev::tstamp</code>, <code>ev::now</code></dt>
  1223. <dd>
  1224. <p>Aliases to the same types/functions as with the <code>ev_</code> prefix.</p>
  1225. </dd>
  1226. <dt><code>ev::io</code>, <code>ev::timer</code>, <code>ev::periodic</code>, <code>ev::idle</code>, <code>ev::sig</code> etc.</dt>
  1227. <dd>
  1228. <p>For each <code>ev_TYPE</code> watcher in <cite>ev.h</cite> there is a corresponding class of
  1229. the same name in the <code>ev</code> namespace, with the exception of <code>ev_signal</code>
  1230. which is called <code>ev::sig</code> to avoid clashes with the <code>signal</code> macro
  1231. defines by many implementations.</p>
  1232. <p>All of those classes have these methods:</p>
  1233. <p>
  1234. <dl>
  1235. <dt>ev::TYPE::TYPE (object *, object::method *)</dt>
  1236. <dt>ev::TYPE::TYPE (object *, object::method *, struct ev_loop *)</dt>
  1237. <dt>ev::TYPE::~TYPE</dt>
  1238. <dd>
  1239. <p>The constructor takes a pointer to an object and a method pointer to
  1240. the event handler callback to call in this class. The constructor calls
  1241. <code>ev_init</code> for you, which means you have to call the <code>set</code> method
  1242. before starting it. If you do not specify a loop then the constructor
  1243. automatically associates the default loop with this watcher.</p>
  1244. <p>The destructor automatically stops the watcher if it is active.</p>
  1245. </dd>
  1246. <dt>w-&gt;set (struct ev_loop *)</dt>
  1247. <dd>
  1248. <p>Associates a different <code>struct ev_loop</code> with this watcher. You can only
  1249. do this when the watcher is inactive (and not pending either).</p>
  1250. </dd>
  1251. <dt>w-&gt;set ([args])</dt>
  1252. <dd>
  1253. <p>Basically the same as <code>ev_TYPE_set</code>, with the same args. Must be
  1254. called at least once. Unlike the C counterpart, an active watcher gets
  1255. automatically stopped and restarted.</p>
  1256. </dd>
  1257. <dt>w-&gt;start ()</dt>
  1258. <dd>
  1259. <p>Starts the watcher. Note that there is no <code>loop</code> argument as the
  1260. constructor already takes the loop.</p>
  1261. </dd>
  1262. <dt>w-&gt;stop ()</dt>
  1263. <dd>
  1264. <p>Stops the watcher if it is active. Again, no <code>loop</code> argument.</p>
  1265. </dd>
  1266. <dt>w-&gt;again () <code>ev::timer</code>, <code>ev::periodic</code> only</dt>
  1267. <dd>
  1268. <p>For <code>ev::timer</code> and <code>ev::periodic</code>, this invokes the corresponding
  1269. <code>ev_TYPE_again</code> function.</p>
  1270. </dd>
  1271. <dt>w-&gt;sweep () <code>ev::embed</code> only</dt>
  1272. <dd>
  1273. <p>Invokes <code>ev_embed_sweep</code>.</p>
  1274. </dd>
  1275. </dl>
  1276. </p>
  1277. </dd>
  1278. </dl>
  1279. <p>Example: Define a class with an IO and idle watcher, start one of them in
  1280. the constructor.</p>
  1281. <pre> class myclass
  1282. {
  1283. ev_io io; void io_cb (ev::io &amp;w, int revents);
  1284. ev_idle idle void idle_cb (ev::idle &amp;w, int revents);
  1285. myclass ();
  1286. }
  1287. myclass::myclass (int fd)
  1288. : io (this, &amp;myclass::io_cb),
  1289. idle (this, &amp;myclass::idle_cb)
  1290. {
  1291. io.start (fd, ev::READ);
  1292. }
  1293. </pre>
  1294. </div>
  1295. <h1 id="EMBEDDING">EMBEDDING</h1><p><a href="#TOP" class="toplink">Top</a></p>
  1296. <div id="EMBEDDING_CONTENT">
  1297. <p>Libev can (and often is) directly embedded into host
  1298. applications. Examples of applications that embed it include the Deliantra
  1299. Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
  1300. and rxvt-unicode.</p>
  1301. <p>The goal is to enable you to just copy the neecssary files into your
  1302. source directory without having to change even a single line in them, so
  1303. you can easily upgrade by simply copying (or having a checked-out copy of
  1304. libev somewhere in your source tree).</p>
  1305. </div>
  1306. <h2 id="FILESETS">FILESETS</h2>
  1307. <div id="FILESETS_CONTENT">
  1308. <p>Depending on what features you need you need to include one or more sets of files
  1309. in your app.</p>
  1310. </div>
  1311. <h3 id="CORE_EVENT_LOOP">CORE EVENT LOOP</h3>
  1312. <div id="CORE_EVENT_LOOP_CONTENT">
  1313. <p>To include only the libev core (all the <code>ev_*</code> functions), with manual
  1314. configuration (no autoconf):</p>
  1315. <pre> #define EV_STANDALONE 1
  1316. #include &quot;ev.c&quot;
  1317. </pre>
  1318. <p>This will automatically include <cite>ev.h</cite>, too, and should be done in a
  1319. single C source file only to provide the function implementations. To use
  1320. it, do the same for <cite>ev.h</cite> in all files wishing to use this API (best
  1321. done by writing a wrapper around <cite>ev.h</cite> that you can include instead and
  1322. where you can put other configuration options):</p>
  1323. <pre> #define EV_STANDALONE 1
  1324. #include &quot;ev.h&quot;
  1325. </pre>
  1326. <p>Both header files and implementation files can be compiled with a C++
  1327. compiler (at least, thats a stated goal, and breakage will be treated
  1328. as a bug).</p>
  1329. <p>You need the following files in your source tree, or in a directory
  1330. in your include path (e.g. in libev/ when using -Ilibev):</p>
  1331. <pre> ev.h
  1332. ev.c
  1333. ev_vars.h
  1334. ev_wrap.h
  1335. ev_win32.c required on win32 platforms only
  1336. ev_select.c only when select backend is enabled (which is is by default)
  1337. ev_poll.c only when poll backend is enabled (disabled by default)
  1338. ev_epoll.c only when the epoll backend is enabled (disabled by default)
  1339. ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
  1340. ev_port.c only when the solaris port backend is enabled (disabled by default)
  1341. </pre>
  1342. <p><cite>ev.c</cite> includes the backend files directly when enabled, so you only need
  1343. to compile a single file.</p>
  1344. </div>
  1345. <h3 id="LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</h3>
  1346. <div id="LIBEVENT_COMPATIBILITY_API_CONTENT">
  1347. <p>To include the libevent compatibility API, also include:</p>
  1348. <pre> #include &quot;event.c&quot;
  1349. </pre>
  1350. <p>in the file including <cite>ev.c</cite>, and:</p>
  1351. <pre> #include &quot;event.h&quot;
  1352. </pre>
  1353. <p>in the files that want to use the libevent API. This also includes <cite>ev.h</cite>.</p>
  1354. <p>You need the following additional files for this:</p>
  1355. <pre> event.h
  1356. event.c
  1357. </pre>
  1358. </div>
  1359. <h3 id="AUTOCONF_SUPPORT">AUTOCONF SUPPORT</h3>
  1360. <div id="AUTOCONF_SUPPORT_CONTENT">
  1361. <p>Instead of using <code>EV_STANDALONE=1</code> and providing your config in
  1362. whatever way you want, you can also <code>m4_include([libev.m4])</code> in your
  1363. <cite>configure.ac</cite> and leave <code>EV_STANDALONE</code> off. <cite>ev.c</cite> will then include
  1364. <cite>config.h</cite> and configure itself accordingly.</p>
  1365. <p>For this of course you need the m4 file:</p>
  1366. <pre> libev.m4
  1367. </pre>
  1368. </div>
  1369. <h2 id="PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</h2>
  1370. <div id="PREPROCESSOR_SYMBOLS_MACROS_CONTENT">
  1371. <p>Libev can be configured via a variety of preprocessor symbols you have to define
  1372. before including any of its files. The default is not to build for multiplicity
  1373. and only include the select backend.</p>
  1374. <dl>
  1375. <dt>EV_STANDALONE</dt>
  1376. <dd>
  1377. <p>Must always be <code>1</code> if you do not use autoconf configuration, which
  1378. keeps libev from including <cite>config.h</cite>, and it also defines dummy
  1379. implementations for some libevent functions (such as logging, which is not
  1380. supported). It will also not define any of the structs usually found in
  1381. <cite>event.h</cite> that are not directly supported by the libev core alone.</p>
  1382. </dd>
  1383. <dt>EV_USE_MONOTONIC</dt>
  1384. <dd>
  1385. <p>If defined to be <code>1</code>, libev will try to detect the availability of the
  1386. monotonic clock option at both compiletime and runtime. Otherwise no use
  1387. of the monotonic clock option will be attempted. If you enable this, you
  1388. usually have to link against librt or something similar. Enabling it when
  1389. the functionality isn't available is safe, though, althoguh you have
  1390. to make sure you link against any libraries where the <code>clock_gettime</code>
  1391. function is hiding in (often <cite>-lrt</cite>).</p>
  1392. </dd>
  1393. <dt>EV_USE_REALTIME</dt>
  1394. <dd>
  1395. <p>If defined to be <code>1</code>, libev will try to detect the availability of the
  1396. realtime clock option at compiletime (and assume its availability at
  1397. runtime if successful). Otherwise no use of the realtime clock option will
  1398. be attempted. This effectively replaces <code>gettimeofday</code> by <code>clock_get
  1399. (CLOCK_REALTIME, ...)</code> and will not normally affect correctness. See tzhe note about libraries
  1400. in the description of <code>EV_USE_MONOTONIC</code>, though.</p>
  1401. </dd>
  1402. <dt>EV_USE_SELECT</dt>
  1403. <dd>
  1404. <p>If undefined or defined to be <code>1</code>, libev will compile in support for the
  1405. <code>select</code>(2) backend. No attempt at autodetection will be done: if no
  1406. other method takes over, select will be it. Otherwise the select backend
  1407. will not be compiled in.</p>
  1408. </dd>
  1409. <dt>EV_SELECT_USE_FD_SET</dt>
  1410. <dd>
  1411. <p>If defined to <code>1</code>, then the select backend will use the system <code>fd_set</code>
  1412. structure. This is useful if libev doesn't compile due to a missing
  1413. <code>NFDBITS</code> or <code>fd_mask</code> definition or it misguesses the bitset layout on
  1414. exotic systems. This usually limits the range of file descriptors to some
  1415. low limit such as 1024 or might have other limitations (winsocket only
  1416. allows 64 sockets). The <code>FD_SETSIZE</code> macro, set before compilation, might
  1417. influence the size of the <code>fd_set</code> used.</p>
  1418. </dd>
  1419. <dt>EV_SELECT_IS_WINSOCKET</dt>
  1420. <dd>
  1421. <p>When defined to <code>1</code>, the select backend will assume that
  1422. select/socket/connect etc. don't understand file descriptors but
  1423. wants osf handles on win32 (this is the case when the select to
  1424. be used is the winsock select). This means that it will call
  1425. <code>_get_osfhandle</code> on the fd to convert it to an OS handle. Otherwise,
  1426. it is assumed that all these functions actually work on fds, even
  1427. on win32. Should not be defined on non-win32 platforms.</p>
  1428. </dd>
  1429. <dt>EV_USE_POLL</dt>
  1430. <dd>
  1431. <p>If defined to be <code>1</code>, libev will compile in support for the <code>poll</code>(2)
  1432. backend. Otherwise it will be enabled on non-win32 platforms. It
  1433. takes precedence over select.</p>
  1434. </dd>
  1435. <dt>EV_USE_EPOLL</dt>
  1436. <dd>
  1437. <p>If defined to be <code>1</code>, libev will compile in support for the Linux
  1438. <code>epoll</code>(7) backend. Its availability will be detected at runtime,
  1439. otherwise another method will be used as fallback. This is the
  1440. preferred backend for GNU/Linux systems.</p>
  1441. </dd>
  1442. <dt>EV_USE_KQUEUE</dt>
  1443. <dd>
  1444. <p>If defined to be <code>1</code>, libev will compile in support for the BSD style
  1445. <code>kqueue</code>(2) backend. Its actual availability will be detected at runtime,
  1446. otherwise another method will be used as fallback. This is the preferred
  1447. backend for BSD and BSD-like systems, although on most BSDs kqueue only
  1448. supports some types of fds correctly (the only platform we found that
  1449. supports ptys for example was NetBSD), so kqueue might be compiled in, but
  1450. not be used unless explicitly requested. The best way to use it is to find
  1451. out whether kqueue supports your type of fd properly and use an embedded
  1452. kqueue loop.</p>
  1453. </dd>
  1454. <dt>EV_USE_PORT</dt>
  1455. <dd>
  1456. <p>If defined to be <code>1</code>, libev will compile in support for the Solaris
  1457. 10 port style backend. Its availability will be detected at runtime,
  1458. otherwise another method will be used as fallback. This is the preferred
  1459. backend for Solaris 10 systems.</p>
  1460. </dd>
  1461. <dt>EV_USE_DEVPOLL</dt>
  1462. <dd>
  1463. <p>reserved for future expansion, works like the USE symbols above.</p>
  1464. </dd>
  1465. <dt>EV_H</dt>
  1466. <dd>
  1467. <p>The name of the <cite>ev.h</cite> header file used to include it. The default if
  1468. undefined is <code>&lt;ev.h&gt;</code> in <cite>event.h</cite> and <code>&quot;ev.h&quot;</code> in <cite>ev.c</cite>. This
  1469. can be used to virtually rename the <cite>ev.h</cite> header file in case of conflicts.</p>
  1470. </dd>
  1471. <dt>EV_CONFIG_H</dt>
  1472. <dd>
  1473. <p>If <code>EV_STANDALONE</code> isn't <code>1</code>, this variable can be used to override
  1474. <cite>ev.c</cite>'s idea of where to find the <cite>config.h</cite> file, similarly to
  1475. <code>EV_H</code>, above.</p>
  1476. </dd>
  1477. <dt>EV_EVENT_H</dt>
  1478. <dd>
  1479. <p>Similarly to <code>EV_H</code>, this macro can be used to override <cite>event.c</cite>'s idea
  1480. of how the <cite>event.h</cite> header can be found.</p>
  1481. </dd>
  1482. <dt>EV_PROTOTYPES</dt>
  1483. <dd>
  1484. <p>If defined to be <code>0</code>, then <cite>ev.h</cite> will not define any function
  1485. prototypes, but still define all the structs and other symbols. This is
  1486. occasionally useful if you want to provide your own wrapper functions
  1487. around libev functions.</p>
  1488. </dd>
  1489. <dt>EV_MULTIPLICITY</dt>
  1490. <dd>
  1491. <p>If undefined or defined to <code>1</code>, then all event-loop-specific functions
  1492. will have the <code>struct ev_loop *</code> as first argument, and you can create
  1493. additional independent event loops. Otherwise there will be no support
  1494. for multiple event loops and there is no first event loop pointer
  1495. argument. Instead, all functions act on the single default loop.</p>
  1496. </dd>
  1497. <dt>EV_PERIODICS</dt>
  1498. <dd>
  1499. <p>If undefined or defined to be <code>1</code>, then periodic timers are supported,
  1500. otherwise not. This saves a few kb of code.</p>
  1501. </dd>
  1502. <dt>EV_COMMON</dt>
  1503. <dd>
  1504. <p>By default, all watchers have a <code>void *data</code> member. By redefining
  1505. this macro to a something else you can include more and other types of
  1506. members. You have to define it each time you include one of the files,
  1507. though, and it must be identical each time.</p>
  1508. <p>For example, the perl EV module uses something like this:</p>
  1509. <pre> #define EV_COMMON \
  1510. SV *self; /* contains this struct */ \
  1511. SV *cb_sv, *fh /* note no trailing &quot;;&quot; */
  1512. </pre>
  1513. </dd>
  1514. <dt>EV_CB_DECLARE(type)</dt>
  1515. <dt>EV_CB_INVOKE(watcher,revents)</dt>
  1516. <dt>ev_set_cb(ev,cb)</dt>
  1517. <dd>
  1518. <p>Can be used to change the callback member declaration in each watcher,
  1519. and the way callbacks are invoked and set. Must expand to a struct member
  1520. definition and a statement, respectively. See the <cite>ev.v</cite> header file for
  1521. their default definitions. One possible use for overriding these is to
  1522. avoid the ev_loop pointer as first argument in all cases, or to use method
  1523. calls instead of plain function calls in C++.</p>
  1524. </div>
  1525. <h2 id="EXAMPLES">EXAMPLES</h2>
  1526. <div id="EXAMPLES_CONTENT">
  1527. <p>For a real-world example of a program the includes libev
  1528. verbatim, you can have a look at the EV perl module
  1529. (<a href="http://software.schmorp.de/pkg/EV.html">http://software.schmorp.de/pkg/EV.html</a>). It has the libev files in
  1530. the <cite>libev/</cite> subdirectory and includes them in the <cite>EV/EVAPI.h</cite> (public
  1531. interface) and <cite>EV.xs</cite> (implementation) files. Only the <cite>EV.xs</cite> file
  1532. will be compiled. It is pretty complex because it provides its own header
  1533. file.</p>
  1534. <p>The usage in rxvt-unicode is simpler. It has a <cite>ev_cpp.h</cite> header file
  1535. that everybody includes and which overrides some autoconf choices:</p>
  1536. <pre> #define EV_USE_POLL 0
  1537. #define EV_MULTIPLICITY 0
  1538. #define EV_PERIODICS 0
  1539. #define EV_CONFIG_H &lt;config.h&gt;
  1540. #include &quot;ev++.h&quot;
  1541. </pre>
  1542. <p>And a <cite>ev_cpp.C</cite> implementation file that contains libev proper and is compiled:</p>
  1543. <pre> #include &quot;ev_cpp.h&quot;
  1544. #include &quot;ev.c&quot;
  1545. </pre>
  1546. </div>
  1547. <h1 id="AUTHOR">AUTHOR</h1><p><a href="#TOP" class="toplink">Top</a></p>
  1548. <div id="AUTHOR_CONTENT">
  1549. <p>Marc Lehmann &lt;libev@schmorp.de&gt;.</p>
  1550. </div>
  1551. </div></body>
  1552. </html>