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

Comparing libev/Changes (file contents):
Revision 1.75 by root, Mon Oct 27 11:08:29 2008 UTC vs.
Revision 1.159 by root, Thu Jan 14 20:23:36 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
3WISH? monotonic clocks times/GetTickCount for coarse corrections? 3 - applied win32 fixes by Michael Lenaghan.
4 - remove dependency on sys/queue.h on freebsd (patch by Vanilla Hsu).
5 - configure now prepends -O3, not appends it, so one can still
6 override it.
4 7
83.9 Thu Dec 31 07:59:59 CET 2009
9 - signalfd is no longer used by default and has to be requested
10 explicitly - this means that easy to catch bugs become hard to
11 catch race conditions, but the users have spoken.
12 - point out the unspecified signal mask in the documentation, and
13 that this is a race condition regardless of EV_SIGNALFD.
14 - backport inotify code to C89.
15 - inotify file descriptors could leak into child processes.
16 - ev_stat watchers could keep an errornous extra ref on the loop,
17 preventing exit when unregistering all watchers (testcases
18 provided by ry@tinyclouds.org).
19 - implement EV_WIN32_HANDLE_TO_FD and EV_WIN32_CLOSE_FD configuration
20 symbols to make it easier for apps to do their own fd management.
21 - support EV_IDLE_ENABLE being disabled in ev++.h
22 (patch by Didier Spezia).
23 - take advantage of inotify_init1, if available, to set cloexec/nonblock
24 on fd creation, to avoid races.
25 - the signal handling pipe wasn't always initialised under windows
26 (analysed by lekma).
27 - changed minimum glibc requirement from glibc 2.9 to 2.7, for
28 signalfd.
29 - add missing string.h include (Denis F. Latypoff).
30 - only replace ev_stat.prev when we detect an actual difference,
31 so prev is (almost) always different to attr. this might
32 have caused the probems with 04_stat.t.
33 - add ev::timer->remaining () method to C++ API.
34
353.8 Sun Aug 9 14:30:45 CEST 2009
36 - incompatible change: do not necessarily reset signal handler
37 to SIG_DFL when a sighandler is stopped.
38 - ev_default_destroy did not properly free or zero some members,
39 potentially causing crashes and memory corruption on repated
40 ev_default_destroy/ev_default_loop calls.
41 - take advantage of signalfd on GNU/Linux systems.
42 - document that the signal mask might be in an unspecified
43 state when using libev's signal handling.
44 - take advantage of some GNU/Linux calls to set cloexec/nonblock
45 on fd creation, to avoid race conditions.
46
473.7 Fri Jul 17 16:36:32 CEST 2009
48 - ev_unloop and ev_loop wrongly used a global variable to exit loops,
49 instead of using a per-loop variable (bug caught by accident...).
50 - the ev_set_io_collect_interval interpretation has changed.
51 - add new functionality: ev_set_userdata, ev_userdata,
52 ev_set_invoke_pending_cb, ev_set_loop_release_cb,
53 ev_invoke_pending, ev_pending_count, together with a long example
54 about thread locking.
55 - add ev_timer_remaining (as requested by Denis F. Latypoff).
56 - add ev_loop_depth.
57 - calling ev_unloop in fork/prepare watchers will no longer poll
58 for new events.
59 - Denis F. Latypoff corrected many typos in example code snippets.
60 - honor autoconf detection of EV_USE_CLOCK_SYSCALL, also double-
61 check that the syscall number is available before trying to
62 use it (reported by ry@tinyclouds).
63 - use GetSystemTimeAsFileTime instead of _timeb on windows, for
64 slightly higher accuracy.
65 - properly declare ev_loop_verify and ev_now_update even when
66 !EV_MULTIPLICITY.
67 - do not compile in any priority code when EV_MAXPRI == EV_MINPRI.
68 - support EV_MINIMAL==2 for a reduced API.
69 - actually 0-initialise struct sigaction when installing signals.
70 - add section on hibernate and stopped processes to ev_timer docs.
71
723.6 Tue Apr 28 02:49:30 CEST 2009
73 - multiple timers becoming ready within an event loop iteration
74 will be invoked in the "correct" order now.
75 - do not leave the event loop early just because we have no active
76 watchers, fixing a problem when embedding a kqueue loop
77 that has active kernel events but no registered watchers
78 (reported by blacksand blacksand).
79 - correctly zero the idx values for arrays, so destroying and
80 reinitialising the default loop actually works (patch by
81 Malek Hadj-Ali).
82 - implement ev_suspend and ev_resume.
83 - new EV_CUSTOM revents flag for use by applications.
84 - add documentation section about priorites.
85 - add a glossary to the dcoumentation.
86 - extend the ev_fork description slightly.
87 - optimize a jump out of call_pending.
88
893.53 Sun Feb 15 02:38:20 CET 2009
90 - fix a bug in event pipe creation on win32 that would cause a
91 failed assertion on event loop creation (patch by Malek Hadj-Ali).
92 - probe for CLOCK_REALTIME support at runtime as well and fall
93 back to gettimeofday if there is an error, to support older
94 operating systems with newer header files/libraries.
95 - prefer gettimeofday over clock_gettime with USE_CLOCK_SYSCALL
96 (default most everywhere), otherwise not.
97
983.52 Wed Jan 7 21:43:02 CET 2009
99 - fix compilation of select backend in fd_set mode when NFDBITS is
100 missing (to get it to compile on QNX, reported by Rodrigo Campos).
101 - better select-nfds handling when select backend is in fd_set mode.
102 - diagnose fd_set overruns when select backend is in fd_set mode.
103 - due to a thinko, instead of disabling everything but
104 select on the borked OS X platform, everything but select was
105 allowed (reported by Emanuele Giaquinta).
106 - actually verify that local and remote port are matching in
107 libev's socketpair emulation, which makes denial-of-service
108 attacks harder (but not impossible - it's windows). Make sure
109 it even works under vista, which thinks that getpeer/sockname
110 should return fantasy port numbers.
111 - include "libev" in all assertion messages for potentially
112 clearer diagnostics.
113 - event_get_version (libevent compatibility) returned
114 a useless string instead of the expected version string
115 (patch by W.C.A. Wijngaards).
116
1173.51 Wed Dec 24 23:00:11 CET 2008
118 - fix a bug where an inotify watcher was added twice, causing
119 freezes on hash collisions (reported and analysed by Graham Leggett).
120 - new config symbol, EV_USE_CLOCK_SYSCALL, to make libev use
121 a direct syscall - slower, but no dependency on librt et al.
122 - assume negative return values != -1 signals success of port_getn
123 (http://cvs.epicsol.org/cgi/viewcvs.cgi/epic5/source/newio.c?rev=1.52)
124 (no known failure reports, but it doesn't hurt).
125 - fork detection in ev_embed now stops and restarts the watcher
126 automatically.
127 - EXPERIMENTAL: default the method to operator () in ev++.h,
128 to make it nicer to use functors (requested by Benedek László).
129 - fixed const object callbacks in ev++.h.
130 - replaced loop_ref argument of watcher.set (loop) by a direct
131 ev_loop * in ev++.h, to avoid clashes with functor patch.
132 - do not try to watch the empty string via inotify.
133 - inotify watchers could be leaked under certain circumstances.
134 - OS X 10.5 is actually even more broken than earlier versions,
135 so fall back to select on that piece of garbage.
136 - fixed some weirdness in the ev_embed documentation.
137
1383.49 Wed Nov 19 11:26:53 CET 2008
139 - ev_stat watchers will now use inotify as a mere hint on
140 kernels <2.6.25, or if the filesystem is not in the
141 "known to be good" list.
142 - better mingw32 compatibility (it's not as borked as native win32)
143 (analysed by Roger Pack).
144 - include stdio.h in the example program, as too many people are
145 confused by the weird C language otherwise. I guess the next thing
146 I get told is that the "..." ellipses in the examples don't compile
147 with their C compiler.
148
1493.48 Thu Oct 30 09:02:37 CET 2008
5 - further optimise away the EPOLL_CTL_ADD/MOD combo in the epoll 150 - further optimise away the EPOLL_CTL_ADD/MOD combo in the epoll
6 backend by assuming the kernel event mask hasn't changed if 151 backend by assuming the kernel event mask hasn't changed if
7 ADD fails with EEXIST. 152 ADD fails with EEXIST.
8 - work around spurious event notification bugs in epoll by using 153 - work around spurious event notification bugs in epoll by using
9 an 8-bit generation counter. 154 a 32-bit generation counter. recreate kernel state if we receive
155 spurious notifications or unwanted events. this is very costly,
156 but I didn't come up with this horrible design.
10 - use memset to initialise most arrays now and do away with the 157 - use memset to initialise most arrays now and do away with the
11 init functions. 158 init functions.
12 - expand time-out strategies into a "Be smart about timeouts" section. 159 - expand time-out strategies into a "Be smart about timeouts" section.
13 - drop the "struct" from all ev_watcher declarations in the 160 - drop the "struct" from all ev_watcher declarations in the
14 documentation and did other clarifications (yeah, it was a mistake 161 documentation and did other clarifications (yeah, it was a mistake
15 to have a struct AND a function called ev_loop). 162 to have a struct AND a function called ev_loop).
16 - fix a bug where ev_default would not initialise the default 163 - fix a bug where ev_default would not initialise the default
17 loop again after it was destroyed with ev_default_destroy. 164 loop again after it was destroyed with ev_default_destroy.
165 - rename syserr to ev_syserr to avoid name clashes when embedding,
166 do similar changes for event.c.
18 167
193.45 Tue Oct 21 21:59:26 CEST 2008 1683.45 Tue Oct 21 21:59:26 CEST 2008
20 - disable inotify usage on linux <2.6.25, as it is broken 169 - disable inotify usage on linux <2.6.25, as it is broken
21 (reported by Yoann Vandoorselaere). 170 (reported by Yoann Vandoorselaere).
22 - ev_stat errornously would try to add inotify watchers 171 - ev_stat errornously would try to add inotify watchers

Diff Legend

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