| 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: ABI??? API????? Changes??? |
3 | TODO: move some other examples to common idioms? combining watchers, |
| 4 | TODO: win32 write() to socket for signal handling |
4 | thread usage, coroutine switch? |
| 5 | TODO: poll asert |
5 | |
| 6 | TODO: EV_USE__XXX default with config, document |
6 | TODO: /dev/null epoll breakage, maybe catch EPERM? |
| 7 | TODO: pendingall => idleall? |
7 | TODO: frankenpoll |
| 8 | TODO: portbality section, solaris errno rrentrant, aix, win32, linux 32 bit |
8 | 4.03 |
| 9 | TODO: include ev_xyz_start in each example? |
9 | - support files, /dev/zero etc. the same way as select in the epoll |
| 10 | TODO: section watcher states/lifetime |
10 | backend, by generating events on our own. |
|
|
11 | - define EV_READ/EV_WRITE as macros in event.h, as some programs use |
|
|
12 | #ifdef to test for them. |
|
|
13 | |
|
|
14 | 4.01 Fri Nov 5 21:51:29 CET 2010 |
|
|
15 | - automake fucked it up, apparently, --add-missing -f is not quite enough |
|
|
16 | to make it update its files, so 4.00 didn't install ev++.h and |
|
|
17 | event.h on make install. grrr. |
|
|
18 | - ev_loop(count|depth) didn't return anything (Robin Haberkorn). |
|
|
19 | - change EV_UNDEF to 0xffffffff to silence some overzealous compilers. |
|
|
20 | - use "(libev) " prefix for all libev error messages now. |
|
|
21 | |
|
|
22 | 4.00 Mon Oct 25 12:32:12 CEST 2010 |
| 11 | - "PORTING FROM LIBEV 3.X TO 4.X" (in ev.pod) is recommended reading. |
23 | - "PORTING FROM LIBEV 3.X TO 4.X" (in ev.pod) is recommended reading. |
| 12 | - ev_embed_stop did not correctly stop the watcher (very good |
24 | - ev_embed_stop did not correctly stop the watcher (very good |
| 13 | testcase by Vladimir Timofeev). |
25 | testcase by Vladimir Timofeev). |
|
|
26 | - ev_run will now always update the current loop time - it erroneously |
|
|
27 | didn't when idle watchers were active, causing timers not to fire. |
| 14 | - fix a bug where a timeout of zero caused the timer not to fire |
28 | - fix a bug where a timeout of zero caused the timer not to fire |
| 15 | in the libevent emulation (testcase by Péter Szabó). |
29 | in the libevent emulation (testcase by Péter Szabó). |
| 16 | - applied win32 fixes by Michael Lenaghan (also James Mansion). |
30 | - applied win32 fixes by Michael Lenaghan (also James Mansion). |
| 17 | - replace EV_MINIMAL by EV_FEATURES. |
31 | - replace EV_MINIMAL by EV_FEATURES. |
| 18 | - prefer EPOLL_CTL_ADD over EPOLL_CTL_MOD in some more cases, as it |
32 | - prefer EPOLL_CTL_ADD over EPOLL_CTL_MOD in some more cases, as it |
| 19 | seems the former is *much* faster than the latter. |
33 | seems the former is *much* faster than the latter. |
|
|
34 | - linux kernel version detection (for inotify bug workarounds) |
|
|
35 | did not work properly. |
|
|
36 | - reduce the number of spurious wake-ups with the ports backend. |
| 20 | - remove dependency on sys/queue.h on freebsd (patch by Vanilla Hsu). |
37 | - remove dependency on sys/queue.h on freebsd (patch by Vanilla Hsu). |
|
|
38 | - do async init within ev_async_start, not ev_async_set, which avoids |
|
|
39 | an API quirk where the set function must be called in the C++ API |
|
|
40 | even when there is nothing to set. |
| 21 | - add (undocumented) EV_ENABLE when adding events with kqueue, |
41 | - add (undocumented) EV_ENABLE when adding events with kqueue, |
| 22 | this might help with OS X, which seems to need it despite documenting |
42 | this might help with OS X, which seems to need it despite documenting |
| 23 | not to need it (helpfully pointed out by Tilghman Lesher). |
43 | not to need it (helpfully pointed out by Tilghman Lesher). |
| 24 | - do not use poll by default on freebsd, it's broken (what isn't |
44 | - do not use poll by default on freebsd, it's broken (what isn't |
| 25 | on freebsd...). |
45 | on freebsd...). |
|
|
46 | - allow to embed epoll on kernels >= 2.6.32. |
| 26 | - configure now prepends -O3, not appends it, so one can still |
47 | - configure now prepends -O3, not appends it, so one can still |
| 27 | override it. |
48 | override it. |
|
|
49 | - ev.pod: greatly expanded the portability section, added a porting |
|
|
50 | section, a description of watcher states and made lots of minor fixes. |
| 28 | - disable poll backend on AIX, the poll header spams the namespace |
51 | - disable poll backend on AIX, the poll header spams the namespace |
| 29 | and it's not worth working around dead platforms (reported |
52 | and it's not worth working around dead platforms (reported |
| 30 | and analyzed by Aivars Kalvans). |
53 | and analyzed by Aivars Kalvans). |
| 31 | - improve header file compatibility of the standalone eventfd code |
54 | - improve header file compatibility of the standalone eventfd code |
| 32 | in an obscure case. |
55 | in an obscure case. |
| … | |
… | |
| 35 | no sscanf dependency). |
58 | no sscanf dependency). |
| 36 | - new EV_CHILD_ENABLE and EV_SIGNAL_ENABLE configurable settings. |
59 | - new EV_CHILD_ENABLE and EV_SIGNAL_ENABLE configurable settings. |
| 37 | - update libev.m4 HAVE_CLOCK_SYSCALL test for newer glibcs. |
60 | - update libev.m4 HAVE_CLOCK_SYSCALL test for newer glibcs. |
| 38 | - add section on accept() problems to the manpage. |
61 | - add section on accept() problems to the manpage. |
| 39 | - rename EV_TIMEOUT to EV_TIMER. |
62 | - rename EV_TIMEOUT to EV_TIMER. |
| 40 | - rename ev_loop_count/depth/verify. |
63 | - rename ev_loop_count/depth/verify/loop/unloop. |
|
|
64 | - remove ev_default_destroy and ev_default_fork. |
| 41 | - switch to two-digit minor version. |
65 | - switch to two-digit minor version. |
| 42 | - work around an apparent gentoo compiler bug. |
66 | - work around an apparent gentoo compiler bug. |
|
|
67 | - define _DARWIN_UNLIMITED_SELECT. just so. |
|
|
68 | - use enum instead of #define for most constants. |
|
|
69 | - improve compatibility to older C++ compilers. |
|
|
70 | - (experimental) ev_run/ev_default_loop/ev_break/ev_loop_new have now |
|
|
71 | default arguments when compiled as C++. |
|
|
72 | - enable automake dependency tracking. |
|
|
73 | - ev_loop_new no longer leaks memory when loop creation failed. |
|
|
74 | - new ev_cleanup watcher type. |
| 43 | |
75 | |
| 44 | 3.9 Thu Dec 31 07:59:59 CET 2009 |
76 | 3.9 Thu Dec 31 07:59:59 CET 2009 |
| 45 | - signalfd is no longer used by default and has to be requested |
77 | - signalfd is no longer used by default and has to be requested |
| 46 | explicitly - this means that easy to catch bugs become hard to |
78 | explicitly - this means that easy to catch bugs become hard to |
| 47 | catch race conditions, but the users have spoken. |
79 | catch race conditions, but the users have spoken. |
| … | |
… | |
| 63 | - changed minimum glibc requirement from glibc 2.9 to 2.7, for |
95 | - changed minimum glibc requirement from glibc 2.9 to 2.7, for |
| 64 | signalfd. |
96 | signalfd. |
| 65 | - add missing string.h include (Denis F. Latypoff). |
97 | - add missing string.h include (Denis F. Latypoff). |
| 66 | - only replace ev_stat.prev when we detect an actual difference, |
98 | - only replace ev_stat.prev when we detect an actual difference, |
| 67 | so prev is (almost) always different to attr. this might |
99 | so prev is (almost) always different to attr. this might |
| 68 | have caused the probems with 04_stat.t. |
100 | have caused the problems with 04_stat.t. |
| 69 | - add ev::timer->remaining () method to C++ API. |
101 | - add ev::timer->remaining () method to C++ API. |
| 70 | |
102 | |
| 71 | 3.8 Sun Aug 9 14:30:45 CEST 2009 |
103 | 3.8 Sun Aug 9 14:30:45 CEST 2009 |
| 72 | - incompatible change: do not necessarily reset signal handler |
104 | - incompatible change: do not necessarily reset signal handler |
| 73 | to SIG_DFL when a sighandler is stopped. |
105 | to SIG_DFL when a sighandler is stopped. |
| 74 | - ev_default_destroy did not properly free or zero some members, |
106 | - ev_default_destroy did not properly free or zero some members, |
| 75 | potentially causing crashes and memory corruption on repated |
107 | potentially causing crashes and memory corruption on repeated |
| 76 | ev_default_destroy/ev_default_loop calls. |
108 | ev_default_destroy/ev_default_loop calls. |
| 77 | - take advantage of signalfd on GNU/Linux systems. |
109 | - take advantage of signalfd on GNU/Linux systems. |
| 78 | - document that the signal mask might be in an unspecified |
110 | - document that the signal mask might be in an unspecified |
| 79 | state when using libev's signal handling. |
111 | state when using libev's signal handling. |
| 80 | - take advantage of some GNU/Linux calls to set cloexec/nonblock |
112 | - take advantage of some GNU/Linux calls to set cloexec/nonblock |
| … | |
… | |
| 202 | do similar changes for event.c. |
234 | do similar changes for event.c. |
| 203 | |
235 | |
| 204 | 3.45 Tue Oct 21 21:59:26 CEST 2008 |
236 | 3.45 Tue Oct 21 21:59:26 CEST 2008 |
| 205 | - disable inotify usage on linux <2.6.25, as it is broken |
237 | - disable inotify usage on linux <2.6.25, as it is broken |
| 206 | (reported by Yoann Vandoorselaere). |
238 | (reported by Yoann Vandoorselaere). |
| 207 | - ev_stat errornously would try to add inotify watchers |
239 | - ev_stat erroneously would try to add inotify watchers |
| 208 | even when inotify wasn't available (this should only |
240 | even when inotify wasn't available (this should only |
| 209 | have a performance impact). |
241 | have a performance impact). |
| 210 | - ev_once now passes both timeout and io to the callback if both |
242 | - ev_once now passes both timeout and io to the callback if both |
| 211 | occur concurrently, instead of giving timeouts precedence. |
243 | occur concurrently, instead of giving timeouts precedence. |
| 212 | - disable EV_USE_INOTIFY when sys/inotify.h is too old. |
244 | - disable EV_USE_INOTIFY when sys/inotify.h is too old. |
| … | |
… | |
| 280 | - enable inlining with gcc versions 2 and 3. |
312 | - enable inlining with gcc versions 2 and 3. |
| 281 | - work around broken poll implementations potentially |
313 | - work around broken poll implementations potentially |
| 282 | not clearing revents field in ev_poll (Brandon Black) |
314 | not clearing revents field in ev_poll (Brandon Black) |
| 283 | (no such systems are known at this time). |
315 | (no such systems are known at this time). |
| 284 | - work around a bug in realloc on openbsd and darwin, |
316 | - work around a bug in realloc on openbsd and darwin, |
| 285 | also makes the errornous valgrind complaints |
317 | also makes the erroneous valgrind complaints |
| 286 | go away (noted by various people). |
318 | go away (noted by various people). |
| 287 | - fix ev_async_pending, add c++ wrapper for ev_async |
319 | - fix ev_async_pending, add c++ wrapper for ev_async |
| 288 | (based on patch sent by Johannes Deisenhofer. |
320 | (based on patch sent by Johannes Deisenhofer). |
| 289 | - add sensible set method to ev::embed. |
321 | - add sensible set method to ev::embed. |
| 290 | - made integer constants type int in ev.h. |
322 | - made integer constants type int in ev.h. |
| 291 | |
323 | |
| 292 | 3.2 Wed Apr 2 17:11:19 CEST 2008 |
324 | 3.2 Wed Apr 2 17:11:19 CEST 2008 |
| 293 | - fix a 64 bit overflow issue in the select backend, |
325 | - fix a 64 bit overflow issue in the select backend, |
