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

Comparing libev/Changes (file contents):
Revision 1.39 by root, Sun May 11 11:47:27 2008 UTC vs.
Revision 1.119 by root, Wed Jul 8 02:46:05 2009 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
33.33 3TODO: ev_walk
4TODO: child wactehr execution order w.r.t. childsigcb reg. priorities
5TODO: remain
6 - ev_unloop and ev_loop wrongly used a global variable to exit loops,
7 instead of using a per-loop variable (bug caught by accident...).
8 - Denis F. Latypoff corrected many typos in example code snippets.
9 - the ev_set_io_collect_interval interpretation has changed.
10 - honor autoconf detection of EV_USE_CLOCK_SYSCALL, also double-
11 check that the syscall number is available before trying to
12 use it (reported by ry@tinyclouds).
13 - use GetSystemTimeAsFileTime instead of _timeb on windows, for
14 slightly higher accuracy.
15 - properly declare ev_loop_verify even when !EV_MULTIPLICITY.
16 - implement ev_loop_depth.
17
183.6 Tue Apr 28 02:49:30 CEST 2009
19 - multiple timers becoming ready within an event loop iteration
20 will be invoked in the "correct" order now.
21 - do not leave the event loop early just because we have no active
22 watchers, fixing a problem when embedding a kqueue loop
23 that has active kernel events but no registered watchers
24 (reported by blacksand blacksand).
25 - correctly zero the idx values for arrays, so destroying and
26 reinitialising the default loop actually works (patch by
27 Malek Hadj-Ali).
28 - implement ev_suspend and ev_resume.
29 - new EV_CUSTOM revents flag for use by applications.
30 - add documentation section about priorites.
31 - add a glossary to the dcoumentation.
32 - extend the ev_fork description slightly.
33 - optimize a jump out of call_pending.
34
353.53 Sun Feb 15 02:38:20 CET 2009
36 - fix a bug in event pipe creation on win32 that would cause a
37 failed assertion on event loop creation (patch by Malek Hadj-Ali).
38 - probe for CLOCK_REALTIME support at runtime as well and fall
39 back to gettimeofday if there is an error, to support older
40 operating systems with newer header files/libraries.
41 - prefer gettimeofday over clock_gettime with USE_CLOCK_SYSCALL
42 (default most everywhere), otherwise not.
43
443.52 Wed Jan 7 21:43:02 CET 2009
45 - fix compilation of select backend in fd_set mode when NFDBITS is
46 missing (to get it to compile on QNX, reported by Rodrigo Campos).
47 - better select-nfds handling when select backend is in fd_set mode.
48 - diagnose fd_set overruns when select backend is in fd_set mode.
49 - due to a thinko, instead of disabling everything but
50 select on the borked OS X platform, everything but select was
51 allowed (reported by Emanuele Giaquinta).
52 - actually verify that local and remote port are matching in
53 libev's socketpair emulation, which makes denial-of-service
54 attacks harder (but not impossible - it's windows). Make sure
55 it even works under vista, which thinks that getpeer/sockname
56 should return fantasy port numbers.
57 - include "libev" in all assertion messages for potentially
58 clearer diagnostics.
59 - event_get_version (libevent compatibility) returned
60 a useless string instead of the expected version string
61 (patch by W.C.A. Wijngaards).
62
633.51 Wed Dec 24 23:00:11 CET 2008
64 - fix a bug where an inotify watcher was added twice, causing
65 freezes on hash collisions (reported and analysed by Graham Leggett).
66 - new config symbol, EV_USE_CLOCK_SYSCALL, to make libev use
67 a direct syscall - slower, but no dependency on librt et al.
68 - assume negative return values != -1 signals success of port_getn
69 (http://cvs.epicsol.org/cgi/viewcvs.cgi/epic5/source/newio.c?rev=1.52)
70 (no known failure reports, but it doesn't hurt).
71 - fork detection in ev_embed now stops and restarts the watcher
72 automatically.
73 - EXPERIMENTAL: default the method to operator () in ev++.h,
74 to make it nicer to use functors (requested by Benedek László).
75 - fixed const object callbacks in ev++.h.
76 - replaced loop_ref argument of watcher.set (loop) by a direct
77 ev_loop * in ev++.h, to avoid clashes with functor patch.
78 - do not try to watch the empty string via inotify.
79 - inotify watchers could be leaked under certain circumstances.
80 - OS X 10.5 is actually even more broken than earlier versions,
81 so fall back to select on that piece of garbage.
82 - fixed some weirdness in the ev_embed documentation.
83
843.49 Wed Nov 19 11:26:53 CET 2008
85 - ev_stat watchers will now use inotify as a mere hint on
86 kernels <2.6.25, or if the filesystem is not in the
87 "known to be good" list.
88 - better mingw32 compatibility (it's not as borked as native win32)
89 (analysed by Roger Pack).
90 - include stdio.h in the example program, as too many people are
91 confused by the weird C language otherwise. I guess the next thing
92 I get told is that the "..." ellipses in the examples don't compile
93 with their C compiler.
94
953.48 Thu Oct 30 09:02:37 CET 2008
96 - further optimise away the EPOLL_CTL_ADD/MOD combo in the epoll
97 backend by assuming the kernel event mask hasn't changed if
98 ADD fails with EEXIST.
99 - work around spurious event notification bugs in epoll by using
100 a 32-bit generation counter. recreate kernel state if we receive
101 spurious notifications or unwanted events. this is very costly,
102 but I didn't come up with this horrible design.
103 - use memset to initialise most arrays now and do away with the
104 init functions.
105 - expand time-out strategies into a "Be smart about timeouts" section.
106 - drop the "struct" from all ev_watcher declarations in the
107 documentation and did other clarifications (yeah, it was a mistake
108 to have a struct AND a function called ev_loop).
109 - fix a bug where ev_default would not initialise the default
110 loop again after it was destroyed with ev_default_destroy.
111 - rename syserr to ev_syserr to avoid name clashes when embedding,
112 do similar changes for event.c.
113
1143.45 Tue Oct 21 21:59:26 CEST 2008
115 - disable inotify usage on linux <2.6.25, as it is broken
116 (reported by Yoann Vandoorselaere).
117 - ev_stat errornously would try to add inotify watchers
118 even when inotify wasn't available (this should only
119 have a performance impact).
120 - ev_once now passes both timeout and io to the callback if both
121 occur concurrently, instead of giving timeouts precedence.
122 - disable EV_USE_INOTIFY when sys/inotify.h is too old.
123
1243.44 Mon Sep 29 05:18:39 CEST 2008
125 - embed watchers now automatically invoke ev_loop_fork on the
126 embedded loop when the parent loop forks.
127 - new function: ev_now_update (loop).
128 - verify_watcher was not marked static.
129 - improve the "associating..." manpage section.
130 - documentation tweaks here and there.
131
1323.43 Sun Jul 6 05:34:41 CEST 2008
133 - include more include files on windows to get struct _stati64
134 (reported by Chris Hulbert, but doesn't quite fix his issue).
135 - add missing #include <io.h> in ev.c on windows (reported by
136 Matt Tolton).
137
1383.42 Tue Jun 17 12:12:07 CEST 2008
139 - work around yet another windows bug: FD_SET actually adds fd's
140 multiple times to the fd_*SET*, despite official MSN docs claiming
141 otherwise. Reported and well-analysed by Matt Tolton.
142 - define NFDBITS to 0 when EV_SELECT_IS_WINSOCKET to make it compile
143 (reported any analysed by Chris Hulbert).
144 - fix a bug in ev_ebadf (this function is only used to catch
145 programming errors in the libev user). reported by Matt Tolton.
146 - fix a bug in fd_intern on win32 (could lead to compile errors
147 under some circumstances, but would work correctly if it compiles).
148 reported by Matt Tolton.
149 - (try to) work around missing lstat on windows.
150 - pass in the write fd set as except fd set under windows. windows
151 is so uncontrollably lame that it requires this. this means that
152 switching off oobinline is not supported (but tcp/ip doesn't
153 have oob, so that would be stupid anyways.
154 - use posix module symbol to auto-detect monotonic clock presence
155 and some other default values.
156
1573.41 Fri May 23 18:42:54 CEST 2008
158 - work around an obscure bug in winsocket select: if you
159 provide only empty fd sets then select returns WSAEINVAL. how sucky.
160 - improve timer scheduling stability and reduce use of time_epsilon.
4 - use 1-based 2-heap for EV_MINIMAL, simplifies code, reduces 161 - use 1-based 2-heap for EV_MINIMAL, simplifies code, reduces
5 codesize and makes for better cache-efficiency. 162 codesize and makes for better cache-efficiency.
6 - use 3-based 4-heap for !EV_MINIMAL. this makes better use 163 - use 3-based 4-heap for !EV_MINIMAL. this makes better use
7 of cpu cache lines and gives better growth behaviour than 164 of cpu cache lines and gives better growth behaviour than
8 2-based heaps. 165 2-based heaps.
… …
12 - fix a potential aliasing issue in ev_timer_again. 169 - fix a potential aliasing issue in ev_timer_again.
13 - add/document ev_periodic_at, retract direct access to ->at. 170 - add/document ev_periodic_at, retract direct access to ->at.
14 - improve ev_stat docs. 171 - improve ev_stat docs.
15 - add portability requirements section. 172 - add portability requirements section.
16 - fix manpage headers etc. 173 - fix manpage headers etc.
174 - normalise WSA error codes to lower range on windows.
175 - add consistency check code that can be called automatically
176 or on demand to check for internal structures (ev_loop_verify).
17 177
183.31 Wed Apr 16 20:45:04 CEST 2008 1783.31 Wed Apr 16 20:45:04 CEST 2008
19 - added last minute fix for ev_poll.c by Brandon Black. 179 - added last minute fix for ev_poll.c by Brandon Black.
20 180
213.3 Wed Apr 16 19:04:10 CEST 2008 1813.3 Wed Apr 16 19:04:10 CEST 2008

Diff Legend

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