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

Comparing libev/Changes (file contents):
Revision 1.91 by root, Fri Dec 12 20:35:21 2008 UTC vs.
Revision 1.180 by root, Thu Jun 24 19:22:26 2010 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 2
3TODO: ABI??? API????? Changes???
4TODO: win32 write() to socket for signal handling
5TODO: poll asert
6TODO: EV_USE__XXX default with config, document
7TODO: pendingall => idleall?
8TODO: portbality section, solaris errno rrentrant, aix, win32, linux 32 bit
9TODO: include ev_xyz_start in each example?
10TODO: section watcher states/lifetime
11 - "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
13 testcase by Vladimir Timofeev).
14 - fix a bug where a timeout of zero caused the timer not to fire
15 in the libevent emulation (testcase by Péter Szabó).
16 - applied win32 fixes by Michael Lenaghan (also James Mansion).
17 - replace EV_MINIMAL by EV_FEATURES.
18 - prefer EPOLL_CTL_ADD over EPOLL_CTL_MOD in some more cases, as it
19 seems the former is *much* faster than the latter.
20 - remove dependency on sys/queue.h on freebsd (patch by Vanilla Hsu).
21 - add (undocumented) EV_ENABLE when adding events with kqueue,
22 this might help with OS X, which seems to need it despite documenting
23 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
25 on freebsd...).
26 - configure now prepends -O3, not appends it, so one can still
27 override it.
28 - disable poll backend on AIX, the poll header spams the namespace
29 and it's not worth working around dead platforms (reported
30 and analyzed by Aivars Kalvans).
31 - improve header file compatibility of the standalone eventfd code
32 in an obscure case.
33 - implement EV_AVOID_STDIO option.
34 - do not use sscanf to parse linux version number (smaller, faster,
35 no sscanf dependency).
36 - new EV_CHILD_ENABLE and EV_SIGNAL_ENABLE configurable settings.
37 - update libev.m4 HAVE_CLOCK_SYSCALL test for newer glibcs.
38 - add section on accept() problems to the manpage.
39 - rename EV_TIMEOUT to EV_TIMER.
40 - rename ev_loop_count/depth/verify.
41 - switch to two-digit minor version.
42 - work around an apparent gentoo compiler bug.
43
443.9 Thu Dec 31 07:59:59 CET 2009
45 - signalfd is no longer used by default and has to be requested
46 explicitly - this means that easy to catch bugs become hard to
47 catch race conditions, but the users have spoken.
48 - point out the unspecified signal mask in the documentation, and
49 that this is a race condition regardless of EV_SIGNALFD.
50 - backport inotify code to C89.
51 - inotify file descriptors could leak into child processes.
52 - ev_stat watchers could keep an errornous extra ref on the loop,
53 preventing exit when unregistering all watchers (testcases
54 provided by ry@tinyclouds.org).
55 - implement EV_WIN32_HANDLE_TO_FD and EV_WIN32_CLOSE_FD configuration
56 symbols to make it easier for apps to do their own fd management.
57 - support EV_IDLE_ENABLE being disabled in ev++.h
58 (patch by Didier Spezia).
59 - take advantage of inotify_init1, if available, to set cloexec/nonblock
60 on fd creation, to avoid races.
61 - the signal handling pipe wasn't always initialised under windows
62 (analysed by lekma).
63 - changed minimum glibc requirement from glibc 2.9 to 2.7, for
64 signalfd.
65 - add missing string.h include (Denis F. Latypoff).
66 - only replace ev_stat.prev when we detect an actual difference,
67 so prev is (almost) always different to attr. this might
68 have caused the probems with 04_stat.t.
69 - add ev::timer->remaining () method to C++ API.
70
713.8 Sun Aug 9 14:30:45 CEST 2009
72 - incompatible change: do not necessarily reset signal handler
73 to SIG_DFL when a sighandler is stopped.
74 - ev_default_destroy did not properly free or zero some members,
75 potentially causing crashes and memory corruption on repated
76 ev_default_destroy/ev_default_loop calls.
77 - take advantage of signalfd on GNU/Linux systems.
78 - document that the signal mask might be in an unspecified
79 state when using libev's signal handling.
80 - take advantage of some GNU/Linux calls to set cloexec/nonblock
81 on fd creation, to avoid race conditions.
82
833.7 Fri Jul 17 16:36:32 CEST 2009
84 - ev_unloop and ev_loop wrongly used a global variable to exit loops,
85 instead of using a per-loop variable (bug caught by accident...).
86 - the ev_set_io_collect_interval interpretation has changed.
87 - add new functionality: ev_set_userdata, ev_userdata,
88 ev_set_invoke_pending_cb, ev_set_loop_release_cb,
89 ev_invoke_pending, ev_pending_count, together with a long example
90 about thread locking.
91 - add ev_timer_remaining (as requested by Denis F. Latypoff).
92 - add ev_loop_depth.
93 - calling ev_unloop in fork/prepare watchers will no longer poll
94 for new events.
95 - Denis F. Latypoff corrected many typos in example code snippets.
96 - honor autoconf detection of EV_USE_CLOCK_SYSCALL, also double-
97 check that the syscall number is available before trying to
98 use it (reported by ry@tinyclouds).
99 - use GetSystemTimeAsFileTime instead of _timeb on windows, for
100 slightly higher accuracy.
101 - properly declare ev_loop_verify and ev_now_update even when
102 !EV_MULTIPLICITY.
103 - do not compile in any priority code when EV_MAXPRI == EV_MINPRI.
104 - support EV_MINIMAL==2 for a reduced API.
105 - actually 0-initialise struct sigaction when installing signals.
106 - add section on hibernate and stopped processes to ev_timer docs.
107
1083.6 Tue Apr 28 02:49:30 CEST 2009
109 - multiple timers becoming ready within an event loop iteration
110 will be invoked in the "correct" order now.
111 - do not leave the event loop early just because we have no active
112 watchers, fixing a problem when embedding a kqueue loop
113 that has active kernel events but no registered watchers
114 (reported by blacksand blacksand).
115 - correctly zero the idx values for arrays, so destroying and
116 reinitialising the default loop actually works (patch by
117 Malek Hadj-Ali).
118 - implement ev_suspend and ev_resume.
119 - new EV_CUSTOM revents flag for use by applications.
120 - add documentation section about priorites.
121 - add a glossary to the dcoumentation.
122 - extend the ev_fork description slightly.
123 - optimize a jump out of call_pending.
124
1253.53 Sun Feb 15 02:38:20 CET 2009
126 - fix a bug in event pipe creation on win32 that would cause a
127 failed assertion on event loop creation (patch by Malek Hadj-Ali).
128 - probe for CLOCK_REALTIME support at runtime as well and fall
129 back to gettimeofday if there is an error, to support older
130 operating systems with newer header files/libraries.
131 - prefer gettimeofday over clock_gettime with USE_CLOCK_SYSCALL
132 (default most everywhere), otherwise not.
133
1343.52 Wed Jan 7 21:43:02 CET 2009
135 - fix compilation of select backend in fd_set mode when NFDBITS is
136 missing (to get it to compile on QNX, reported by Rodrigo Campos).
137 - better select-nfds handling when select backend is in fd_set mode.
138 - diagnose fd_set overruns when select backend is in fd_set mode.
139 - due to a thinko, instead of disabling everything but
140 select on the borked OS X platform, everything but select was
141 allowed (reported by Emanuele Giaquinta).
142 - actually verify that local and remote port are matching in
143 libev's socketpair emulation, which makes denial-of-service
144 attacks harder (but not impossible - it's windows). Make sure
145 it even works under vista, which thinks that getpeer/sockname
146 should return fantasy port numbers.
147 - include "libev" in all assertion messages for potentially
148 clearer diagnostics.
149 - event_get_version (libevent compatibility) returned
150 a useless string instead of the expected version string
151 (patch by W.C.A. Wijngaards).
152
1533.51 Wed Dec 24 23:00:11 CET 2008
3 - fix a bug where an inotify watcher was added twice, causing 154 - fix a bug where an inotify watcher was added twice, causing
4 freezes on hash collisions. 155 freezes on hash collisions (reported and analysed by Graham Leggett).
5 - new config symbol, EV_USE_CLOCK_SYSCALL, to make libev use 156 - new config symbol, EV_USE_CLOCK_SYSCALL, to make libev use
6 a direct syscall - slower, but no dependency on librt et al. 157 a direct syscall - slower, but no dependency on librt et al.
7 - assume negative return values != -1 signals success of port_getn 158 - assume negative return values != -1 signals success of port_getn
8 (http://cvs.epicsol.org/cgi/viewcvs.cgi/epic5/source/newio.c?rev=1.52) 159 (http://cvs.epicsol.org/cgi/viewcvs.cgi/epic5/source/newio.c?rev=1.52)
9 (no known failure reports, but it doesn't hurt). 160 (no known failure reports, but it doesn't hurt).
161 - fork detection in ev_embed now stops and restarts the watcher
162 automatically.
10 - EXPERIMENTAL: default the method to operator () in ev++.h, 163 - EXPERIMENTAL: default the method to operator () in ev++.h,
11 to make it nicer to use functors (requested by Benedek László). 164 to make it nicer to use functors (requested by Benedek László).
12 - fixed const object callbacks in ev++.h. 165 - fixed const object callbacks in ev++.h.
13 - replaced loop_ref argument of watcher.set (loop) by a direct 166 - replaced loop_ref argument of watcher.set (loop) by a direct
14 ev_loop * in ev++.h, to avoid clashes with functor patch. 167 ev_loop * in ev++.h, to avoid clashes with functor patch.
15 - do not try to watch the empty string via inotify. 168 - do not try to watch the empty string via inotify.
169 - inotify watchers could be leaked under certain circumstances.
170 - OS X 10.5 is actually even more broken than earlier versions,
171 so fall back to select on that piece of garbage.
172 - fixed some weirdness in the ev_embed documentation.
16 173
173.49 Wed Nov 19 11:26:53 CET 2008 1743.49 Wed Nov 19 11:26:53 CET 2008
18 - ev_stat watchers will now use inotify as a mere hint on 175 - ev_stat watchers will now use inotify as a mere hint on
19 kernels <2.6.25, or if the filesystem is not in the 176 kernels <2.6.25, or if the filesystem is not in the
20 "known to be good" list. 177 "known to be good" list.

Diff Legend

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