| 1 | Revision history for libev, a high-performance and full-featured event loop. |
1 | Revision history for libev, a high-performance and full-featured event loop. |
| 2 | |
2 | |
| 3 | TODO: ev_walk |
3 | TODO: ev_loop_wakeup |
| 4 | TODO: signal handling per loop |
4 | |
| 5 | TODO: NSIG etc. |
5 | 4.10 |
| 6 | TODO: NOSIGFD/NOEVTFD/NOINOTIFY? |
6 | - fix a race where the workaround against the epoll fork bugs |
| 7 | TODO: gotsig per-loop |
7 | caused signals to not be handled anymore. |
|
|
8 | - correct backend_fudge for most backends, and implement a windows |
|
|
9 | specific workaround to avoid looping because we call both |
|
|
10 | select and Sleep, both with different time resolutions. |
|
|
11 | - document range and guarantees of ev_sleep. |
|
|
12 | - document reasonable ranges for periodics interval and offset. |
|
|
13 | - rename backend_fudge to backend_mintime to avoid future confusion :) |
|
|
14 | - change the default periodic reschedule function to hopefully be more |
|
|
15 | exact and correct even in corner cases or in the far future. |
|
|
16 | - do not rely on -lm anymore: use it when available but use our |
|
|
17 | own floor () if it is missing. This should make it easier to embed, |
|
|
18 | as no external libraries are required. |
|
|
19 | - strategically import macros from libecb and mark rarely-used functions |
|
|
20 | as cache-chold (saving almost 2k code size on typical amd64 setups). |
|
|
21 | |
|
|
22 | 4.04 Wed Feb 16 09:01:51 CET 2011 |
|
|
23 | - fix two problems in the native win32 backend, where reuse of fd's |
|
|
24 | with different underlying handles caused handles not to be removed |
|
|
25 | or added to the select set (analyzed and tested by Bert Belder). |
|
|
26 | - do no rely on ceil() in ev_e?poll.c. |
|
|
27 | - backport libev to HP-UX versions before 11 v3. |
|
|
28 | - configure did not detect nanosleep and clock_gettime properly when |
|
|
29 | they are available in the libc (as opposed to -lrt). |
|
|
30 | |
|
|
31 | 4.03 Tue Jan 11 14:37:25 CET 2011 |
|
|
32 | - officially support polling files with all backends. |
|
|
33 | - support files, /dev/zero etc. the same way as select in the epoll |
|
|
34 | backend, by generating events on our own. |
|
|
35 | - ports backend: work around solaris bug 6874410 and many related ones |
|
|
36 | (EINTR, maybe more), with no performance loss (note that the solaris |
|
|
37 | bug report is actually wrong, reality is far more bizarre and broken |
|
|
38 | than that). |
|
|
39 | - define EV_READ/EV_WRITE as macros in event.h, as some programs use |
|
|
40 | #ifdef to test for them. |
|
|
41 | - new (experimental) function: ev_feed_signal. |
|
|
42 | - new (to become default) EVFLAG_NOSIGMASK flag. |
|
|
43 | - new EVBACKEND_MASK symbol. |
|
|
44 | - updated COMMON IDIOMS SECTION. |
|
|
45 | |
|
|
46 | 4.01 Fri Nov 5 21:51:29 CET 2010 |
|
|
47 | - automake fucked it up, apparently, --add-missing -f is not quite enough |
|
|
48 | to make it update its files, so 4.00 didn't install ev++.h and |
|
|
49 | event.h on make install. grrr. |
|
|
50 | - ev_loop(count|depth) didn't return anything (Robin Haberkorn). |
|
|
51 | - change EV_UNDEF to 0xffffffff to silence some overzealous compilers. |
|
|
52 | - use "(libev) " prefix for all libev error messages now. |
|
|
53 | |
|
|
54 | 4.00 Mon Oct 25 12:32:12 CEST 2010 |
|
|
55 | - "PORTING FROM LIBEV 3.X TO 4.X" (in ev.pod) is recommended reading. |
|
|
56 | - ev_embed_stop did not correctly stop the watcher (very good |
|
|
57 | testcase by Vladimir Timofeev). |
|
|
58 | - ev_run will now always update the current loop time - it erroneously |
|
|
59 | didn't when idle watchers were active, causing timers not to fire. |
|
|
60 | - fix a bug where a timeout of zero caused the timer not to fire |
|
|
61 | in the libevent emulation (testcase by Péter Szabó). |
|
|
62 | - applied win32 fixes by Michael Lenaghan (also James Mansion). |
|
|
63 | - replace EV_MINIMAL by EV_FEATURES. |
|
|
64 | - prefer EPOLL_CTL_ADD over EPOLL_CTL_MOD in some more cases, as it |
|
|
65 | seems the former is *much* faster than the latter. |
|
|
66 | - linux kernel version detection (for inotify bug workarounds) |
|
|
67 | did not work properly. |
|
|
68 | - reduce the number of spurious wake-ups with the ports backend. |
|
|
69 | - remove dependency on sys/queue.h on freebsd (patch by Vanilla Hsu). |
|
|
70 | - do async init within ev_async_start, not ev_async_set, which avoids |
|
|
71 | an API quirk where the set function must be called in the C++ API |
|
|
72 | even when there is nothing to set. |
|
|
73 | - add (undocumented) EV_ENABLE when adding events with kqueue, |
|
|
74 | this might help with OS X, which seems to need it despite documenting |
|
|
75 | not to need it (helpfully pointed out by Tilghman Lesher). |
|
|
76 | - do not use poll by default on freebsd, it's broken (what isn't |
|
|
77 | on freebsd...). |
|
|
78 | - allow to embed epoll on kernels >= 2.6.32. |
|
|
79 | - configure now prepends -O3, not appends it, so one can still |
|
|
80 | override it. |
|
|
81 | - ev.pod: greatly expanded the portability section, added a porting |
|
|
82 | section, a description of watcher states and made lots of minor fixes. |
|
|
83 | - disable poll backend on AIX, the poll header spams the namespace |
|
|
84 | and it's not worth working around dead platforms (reported |
|
|
85 | and analyzed by Aivars Kalvans). |
|
|
86 | - improve header file compatibility of the standalone eventfd code |
|
|
87 | in an obscure case. |
|
|
88 | - implement EV_AVOID_STDIO option. |
|
|
89 | - do not use sscanf to parse linux version number (smaller, faster, |
|
|
90 | no sscanf dependency). |
|
|
91 | - new EV_CHILD_ENABLE and EV_SIGNAL_ENABLE configurable settings. |
|
|
92 | - update libev.m4 HAVE_CLOCK_SYSCALL test for newer glibcs. |
|
|
93 | - add section on accept() problems to the manpage. |
|
|
94 | - rename EV_TIMEOUT to EV_TIMER. |
|
|
95 | - rename ev_loop_count/depth/verify/loop/unloop. |
|
|
96 | - remove ev_default_destroy and ev_default_fork. |
|
|
97 | - switch to two-digit minor version. |
|
|
98 | - work around an apparent gentoo compiler bug. |
|
|
99 | - define _DARWIN_UNLIMITED_SELECT. just so. |
|
|
100 | - use enum instead of #define for most constants. |
|
|
101 | - improve compatibility to older C++ compilers. |
|
|
102 | - (experimental) ev_run/ev_default_loop/ev_break/ev_loop_new have now |
|
|
103 | default arguments when compiled as C++. |
|
|
104 | - enable automake dependency tracking. |
|
|
105 | - ev_loop_new no longer leaks memory when loop creation failed. |
|
|
106 | - new ev_cleanup watcher type. |
|
|
107 | |
|
|
108 | 3.9 Thu Dec 31 07:59:59 CET 2009 |
|
|
109 | - signalfd is no longer used by default and has to be requested |
|
|
110 | explicitly - this means that easy to catch bugs become hard to |
|
|
111 | catch race conditions, but the users have spoken. |
|
|
112 | - point out the unspecified signal mask in the documentation, and |
|
|
113 | that this is a race condition regardless of EV_SIGNALFD. |
|
|
114 | - backport inotify code to C89. |
|
|
115 | - inotify file descriptors could leak into child processes. |
|
|
116 | - ev_stat watchers could keep an errornous extra ref on the loop, |
|
|
117 | preventing exit when unregistering all watchers (testcases |
|
|
118 | provided by ry@tinyclouds.org). |
|
|
119 | - implement EV_WIN32_HANDLE_TO_FD and EV_WIN32_CLOSE_FD configuration |
|
|
120 | symbols to make it easier for apps to do their own fd management. |
|
|
121 | - support EV_IDLE_ENABLE being disabled in ev++.h |
|
|
122 | (patch by Didier Spezia). |
|
|
123 | - take advantage of inotify_init1, if available, to set cloexec/nonblock |
|
|
124 | on fd creation, to avoid races. |
|
|
125 | - the signal handling pipe wasn't always initialised under windows |
|
|
126 | (analysed by lekma). |
|
|
127 | - changed minimum glibc requirement from glibc 2.9 to 2.7, for |
|
|
128 | signalfd. |
|
|
129 | - add missing string.h include (Denis F. Latypoff). |
|
|
130 | - only replace ev_stat.prev when we detect an actual difference, |
|
|
131 | so prev is (almost) always different to attr. this might |
|
|
132 | have caused the problems with 04_stat.t. |
|
|
133 | - add ev::timer->remaining () method to C++ API. |
|
|
134 | |
|
|
135 | 3.8 Sun Aug 9 14:30:45 CEST 2009 |
| 8 | - incompatible change: do not necessarily reset signal handler |
136 | - incompatible change: do not necessarily reset signal handler |
| 9 | to SIG_DFL when a sighandler is stopped. |
137 | to SIG_DFL when a sighandler is stopped. |
| 10 | - ev_default_destroy did not properly free or zero some members, |
138 | - ev_default_destroy did not properly free or zero some members, |
| 11 | potentially causing crashes and memory corruption on repated |
139 | potentially causing crashes and memory corruption on repeated |
| 12 | ev_default_destroy/ev_default_loop calls. |
140 | ev_default_destroy/ev_default_loop calls. |
| 13 | - take advantage of signalfd on GNU/Linux systems. |
141 | - take advantage of signalfd on GNU/Linux systems. |
| 14 | - document that the signal mask might be in an unspecified |
142 | - document that the signal mask might be in an unspecified |
| 15 | state when using libev's signal handling. |
143 | state when using libev's signal handling. |
| 16 | - take advantage of some GNU/Linux calls to set cloexec/nonblock |
144 | - take advantage of some GNU/Linux calls to set cloexec/nonblock |
| … | |
… | |
| 138 | do similar changes for event.c. |
266 | do similar changes for event.c. |
| 139 | |
267 | |
| 140 | 3.45 Tue Oct 21 21:59:26 CEST 2008 |
268 | 3.45 Tue Oct 21 21:59:26 CEST 2008 |
| 141 | - disable inotify usage on linux <2.6.25, as it is broken |
269 | - disable inotify usage on linux <2.6.25, as it is broken |
| 142 | (reported by Yoann Vandoorselaere). |
270 | (reported by Yoann Vandoorselaere). |
| 143 | - ev_stat errornously would try to add inotify watchers |
271 | - ev_stat erroneously would try to add inotify watchers |
| 144 | even when inotify wasn't available (this should only |
272 | even when inotify wasn't available (this should only |
| 145 | have a performance impact). |
273 | have a performance impact). |
| 146 | - ev_once now passes both timeout and io to the callback if both |
274 | - ev_once now passes both timeout and io to the callback if both |
| 147 | occur concurrently, instead of giving timeouts precedence. |
275 | occur concurrently, instead of giving timeouts precedence. |
| 148 | - disable EV_USE_INOTIFY when sys/inotify.h is too old. |
276 | - disable EV_USE_INOTIFY when sys/inotify.h is too old. |
| … | |
… | |
| 216 | - enable inlining with gcc versions 2 and 3. |
344 | - enable inlining with gcc versions 2 and 3. |
| 217 | - work around broken poll implementations potentially |
345 | - work around broken poll implementations potentially |
| 218 | not clearing revents field in ev_poll (Brandon Black) |
346 | not clearing revents field in ev_poll (Brandon Black) |
| 219 | (no such systems are known at this time). |
347 | (no such systems are known at this time). |
| 220 | - work around a bug in realloc on openbsd and darwin, |
348 | - work around a bug in realloc on openbsd and darwin, |
| 221 | also makes the errornous valgrind complaints |
349 | also makes the erroneous valgrind complaints |
| 222 | go away (noted by various people). |
350 | go away (noted by various people). |
| 223 | - fix ev_async_pending, add c++ wrapper for ev_async |
351 | - fix ev_async_pending, add c++ wrapper for ev_async |
| 224 | (based on patch sent by Johannes Deisenhofer. |
352 | (based on patch sent by Johannes Deisenhofer). |
| 225 | - add sensible set method to ev::embed. |
353 | - add sensible set method to ev::embed. |
| 226 | - made integer constants type int in ev.h. |
354 | - made integer constants type int in ev.h. |
| 227 | |
355 | |
| 228 | 3.2 Wed Apr 2 17:11:19 CEST 2008 |
356 | 3.2 Wed Apr 2 17:11:19 CEST 2008 |
| 229 | - fix a 64 bit overflow issue in the select backend, |
357 | - fix a 64 bit overflow issue in the select backend, |
