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

Comparing libev/Changes (file contents):
Revision 1.99 by root, Wed Jan 7 20:45:52 2009 UTC vs.
Revision 1.223 by root, Thu Feb 3 20:59:08 2011 UTC

1Revision history for libev, a high-performance and full-featured event loop. 1Revision history for libev, a high-performance and full-featured event loop.
2
3TODO: signal mask problem documentation
4TODO: ev_loop_wakeup
5TODO: windows handle caching, added/not removed in ev_select.c
6 - do no rely on ceil() in ev_e?poll.c.
7 - backport libev to HP-UX versions before 11 v3.
8 - configure did not detect nanosleep and clock_gettime properly when
9 they are available in the libc (as opposed to -lrt).
10
114.03 Tue Jan 11 14:37:25 CET 2011
12 - officially support polling files with all backends.
13 - support files, /dev/zero etc. the same way as select in the epoll
14 backend, by generating events on our own.
15 - ports backend: work around solaris bug 6874410 and many related ones
16 (EINTR, maybe more), with no performance loss (note that the solaris
17 bug report is actually wrong, reality is far more bizarre and broken
18 than that).
19 - define EV_READ/EV_WRITE as macros in event.h, as some programs use
20 #ifdef to test for them.
21 - new (experimental) function: ev_feed_signal.
22 - new (to become default) EVFLAG_NOSIGMASK flag.
23 - new EVBACKEND_MASK symbol.
24 - updated COMMON IDIOMS SECTION.
25
264.01 Fri Nov 5 21:51:29 CET 2010
27 - automake fucked it up, apparently, --add-missing -f is not quite enough
28 to make it update its files, so 4.00 didn't install ev++.h and
29 event.h on make install. grrr.
30 - ev_loop(count|depth) didn't return anything (Robin Haberkorn).
31 - change EV_UNDEF to 0xffffffff to silence some overzealous compilers.
32 - use "(libev) " prefix for all libev error messages now.
33
344.00 Mon Oct 25 12:32:12 CEST 2010
35 - "PORTING FROM LIBEV 3.X TO 4.X" (in ev.pod) is recommended reading.
36 - ev_embed_stop did not correctly stop the watcher (very good
37 testcase by Vladimir Timofeev).
38 - ev_run will now always update the current loop time - it erroneously
39 didn't when idle watchers were active, causing timers not to fire.
40 - fix a bug where a timeout of zero caused the timer not to fire
41 in the libevent emulation (testcase by Péter Szabó).
42 - applied win32 fixes by Michael Lenaghan (also James Mansion).
43 - replace EV_MINIMAL by EV_FEATURES.
44 - prefer EPOLL_CTL_ADD over EPOLL_CTL_MOD in some more cases, as it
45 seems the former is *much* faster than the latter.
46 - linux kernel version detection (for inotify bug workarounds)
47 did not work properly.
48 - reduce the number of spurious wake-ups with the ports backend.
49 - remove dependency on sys/queue.h on freebsd (patch by Vanilla Hsu).
50 - do async init within ev_async_start, not ev_async_set, which avoids
51 an API quirk where the set function must be called in the C++ API
52 even when there is nothing to set.
53 - add (undocumented) EV_ENABLE when adding events with kqueue,
54 this might help with OS X, which seems to need it despite documenting
55 not to need it (helpfully pointed out by Tilghman Lesher).
56 - do not use poll by default on freebsd, it's broken (what isn't
57 on freebsd...).
58 - allow to embed epoll on kernels >= 2.6.32.
59 - configure now prepends -O3, not appends it, so one can still
60 override it.
61 - ev.pod: greatly expanded the portability section, added a porting
62 section, a description of watcher states and made lots of minor fixes.
63 - disable poll backend on AIX, the poll header spams the namespace
64 and it's not worth working around dead platforms (reported
65 and analyzed by Aivars Kalvans).
66 - improve header file compatibility of the standalone eventfd code
67 in an obscure case.
68 - implement EV_AVOID_STDIO option.
69 - do not use sscanf to parse linux version number (smaller, faster,
70 no sscanf dependency).
71 - new EV_CHILD_ENABLE and EV_SIGNAL_ENABLE configurable settings.
72 - update libev.m4 HAVE_CLOCK_SYSCALL test for newer glibcs.
73 - add section on accept() problems to the manpage.
74 - rename EV_TIMEOUT to EV_TIMER.
75 - rename ev_loop_count/depth/verify/loop/unloop.
76 - remove ev_default_destroy and ev_default_fork.
77 - switch to two-digit minor version.
78 - work around an apparent gentoo compiler bug.
79 - define _DARWIN_UNLIMITED_SELECT. just so.
80 - use enum instead of #define for most constants.
81 - improve compatibility to older C++ compilers.
82 - (experimental) ev_run/ev_default_loop/ev_break/ev_loop_new have now
83 default arguments when compiled as C++.
84 - enable automake dependency tracking.
85 - ev_loop_new no longer leaks memory when loop creation failed.
86 - new ev_cleanup watcher type.
87
883.9 Thu Dec 31 07:59:59 CET 2009
89 - signalfd is no longer used by default and has to be requested
90 explicitly - this means that easy to catch bugs become hard to
91 catch race conditions, but the users have spoken.
92 - point out the unspecified signal mask in the documentation, and
93 that this is a race condition regardless of EV_SIGNALFD.
94 - backport inotify code to C89.
95 - inotify file descriptors could leak into child processes.
96 - ev_stat watchers could keep an errornous extra ref on the loop,
97 preventing exit when unregistering all watchers (testcases
98 provided by ry@tinyclouds.org).
99 - implement EV_WIN32_HANDLE_TO_FD and EV_WIN32_CLOSE_FD configuration
100 symbols to make it easier for apps to do their own fd management.
101 - support EV_IDLE_ENABLE being disabled in ev++.h
102 (patch by Didier Spezia).
103 - take advantage of inotify_init1, if available, to set cloexec/nonblock
104 on fd creation, to avoid races.
105 - the signal handling pipe wasn't always initialised under windows
106 (analysed by lekma).
107 - changed minimum glibc requirement from glibc 2.9 to 2.7, for
108 signalfd.
109 - add missing string.h include (Denis F. Latypoff).
110 - only replace ev_stat.prev when we detect an actual difference,
111 so prev is (almost) always different to attr. this might
112 have caused the problems with 04_stat.t.
113 - add ev::timer->remaining () method to C++ API.
114
1153.8 Sun Aug 9 14:30:45 CEST 2009
116 - incompatible change: do not necessarily reset signal handler
117 to SIG_DFL when a sighandler is stopped.
118 - ev_default_destroy did not properly free or zero some members,
119 potentially causing crashes and memory corruption on repeated
120 ev_default_destroy/ev_default_loop calls.
121 - take advantage of signalfd on GNU/Linux systems.
122 - document that the signal mask might be in an unspecified
123 state when using libev's signal handling.
124 - take advantage of some GNU/Linux calls to set cloexec/nonblock
125 on fd creation, to avoid race conditions.
126
1273.7 Fri Jul 17 16:36:32 CEST 2009
128 - ev_unloop and ev_loop wrongly used a global variable to exit loops,
129 instead of using a per-loop variable (bug caught by accident...).
130 - the ev_set_io_collect_interval interpretation has changed.
131 - add new functionality: ev_set_userdata, ev_userdata,
132 ev_set_invoke_pending_cb, ev_set_loop_release_cb,
133 ev_invoke_pending, ev_pending_count, together with a long example
134 about thread locking.
135 - add ev_timer_remaining (as requested by Denis F. Latypoff).
136 - add ev_loop_depth.
137 - calling ev_unloop in fork/prepare watchers will no longer poll
138 for new events.
139 - Denis F. Latypoff corrected many typos in example code snippets.
140 - honor autoconf detection of EV_USE_CLOCK_SYSCALL, also double-
141 check that the syscall number is available before trying to
142 use it (reported by ry@tinyclouds).
143 - use GetSystemTimeAsFileTime instead of _timeb on windows, for
144 slightly higher accuracy.
145 - properly declare ev_loop_verify and ev_now_update even when
146 !EV_MULTIPLICITY.
147 - do not compile in any priority code when EV_MAXPRI == EV_MINPRI.
148 - support EV_MINIMAL==2 for a reduced API.
149 - actually 0-initialise struct sigaction when installing signals.
150 - add section on hibernate and stopped processes to ev_timer docs.
151
1523.6 Tue Apr 28 02:49:30 CEST 2009
153 - multiple timers becoming ready within an event loop iteration
154 will be invoked in the "correct" order now.
155 - do not leave the event loop early just because we have no active
156 watchers, fixing a problem when embedding a kqueue loop
157 that has active kernel events but no registered watchers
158 (reported by blacksand blacksand).
159 - correctly zero the idx values for arrays, so destroying and
160 reinitialising the default loop actually works (patch by
161 Malek Hadj-Ali).
162 - implement ev_suspend and ev_resume.
163 - new EV_CUSTOM revents flag for use by applications.
164 - add documentation section about priorites.
165 - add a glossary to the dcoumentation.
166 - extend the ev_fork description slightly.
167 - optimize a jump out of call_pending.
168
1693.53 Sun Feb 15 02:38:20 CET 2009
170 - fix a bug in event pipe creation on win32 that would cause a
171 failed assertion on event loop creation (patch by Malek Hadj-Ali).
172 - probe for CLOCK_REALTIME support at runtime as well and fall
173 back to gettimeofday if there is an error, to support older
174 operating systems with newer header files/libraries.
175 - prefer gettimeofday over clock_gettime with USE_CLOCK_SYSCALL
176 (default most everywhere), otherwise not.
2 177
33.52 Wed Jan 7 21:43:02 CET 2009 1783.52 Wed Jan 7 21:43:02 CET 2009
4 - fix compilation of select backend in fd_set mode when NFDBITS is 179 - fix compilation of select backend in fd_set mode when NFDBITS is
5 missing (to get it to compile on QNX, reported by Rodrigo Campos). 180 missing (to get it to compile on QNX, reported by Rodrigo Campos).
6 - better select-nfds handling when select backend is in fd_set mode. 181 - better select-nfds handling when select backend is in fd_set mode.
… …
11 - actually verify that local and remote port are matching in 186 - actually verify that local and remote port are matching in
12 libev's socketpair emulation, which makes denial-of-service 187 libev's socketpair emulation, which makes denial-of-service
13 attacks harder (but not impossible - it's windows). Make sure 188 attacks harder (but not impossible - it's windows). Make sure
14 it even works under vista, which thinks that getpeer/sockname 189 it even works under vista, which thinks that getpeer/sockname
15 should return fantasy port numbers. 190 should return fantasy port numbers.
16 - include "libev" all assertion messages for potentially 191 - include "libev" in all assertion messages for potentially
17 clearer diagnostics. 192 clearer diagnostics.
18 - event_get_version (libevent compatibility) returned 193 - event_get_version (libevent compatibility) returned
19 a useless string instead of the expected version string 194 a useless string instead of the expected version string
20 (patch by W.C.A. Wijngaards). 195 (patch by W.C.A. Wijngaards).
21 196
… …
71 do similar changes for event.c. 246 do similar changes for event.c.
72 247
733.45 Tue Oct 21 21:59:26 CEST 2008 2483.45 Tue Oct 21 21:59:26 CEST 2008
74 - disable inotify usage on linux <2.6.25, as it is broken 249 - disable inotify usage on linux <2.6.25, as it is broken
75 (reported by Yoann Vandoorselaere). 250 (reported by Yoann Vandoorselaere).
76 - ev_stat errornously would try to add inotify watchers 251 - ev_stat erroneously would try to add inotify watchers
77 even when inotify wasn't available (this should only 252 even when inotify wasn't available (this should only
78 have a performance impact). 253 have a performance impact).
79 - ev_once now passes both timeout and io to the callback if both 254 - ev_once now passes both timeout and io to the callback if both
80 occur concurrently, instead of giving timeouts precedence. 255 occur concurrently, instead of giving timeouts precedence.
81 - disable EV_USE_INOTIFY when sys/inotify.h is too old. 256 - disable EV_USE_INOTIFY when sys/inotify.h is too old.
… …
149 - enable inlining with gcc versions 2 and 3. 324 - enable inlining with gcc versions 2 and 3.
150 - work around broken poll implementations potentially 325 - work around broken poll implementations potentially
151 not clearing revents field in ev_poll (Brandon Black) 326 not clearing revents field in ev_poll (Brandon Black)
152 (no such systems are known at this time). 327 (no such systems are known at this time).
153 - work around a bug in realloc on openbsd and darwin, 328 - work around a bug in realloc on openbsd and darwin,
154 also makes the errornous valgrind complaints 329 also makes the erroneous valgrind complaints
155 go away (noted by various people). 330 go away (noted by various people).
156 - fix ev_async_pending, add c++ wrapper for ev_async 331 - fix ev_async_pending, add c++ wrapper for ev_async
157 (based on patch sent by Johannes Deisenhofer. 332 (based on patch sent by Johannes Deisenhofer).
158 - add sensible set method to ev::embed. 333 - add sensible set method to ev::embed.
159 - made integer constants type int in ev.h. 334 - made integer constants type int in ev.h.
160 335
1613.2 Wed Apr 2 17:11:19 CEST 2008 3363.2 Wed Apr 2 17:11:19 CEST 2008
162 - fix a 64 bit overflow issue in the select backend, 337 - fix a 64 bit overflow issue in the select backend,

Diff Legend

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