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

Comparing libev/Changes (file contents):
Revision 1.203 by root, Thu Oct 28 04:55:19 2010 UTC vs.
Revision 1.319 by root, Mon Oct 29 00:00:21 2018 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
3 - move the darwin select workaround highe rin ev.c, as newre versions of
4 darwin managed to break their broken select even more.
5 - ANDROID => __ANDROID__ (reported by enh@google.com).
6 - disable epoll_create1 on android because it has broken header files
7 and google is unwilling to fix them (reported by enh@google.com).
8 - avoid a minor compilation warning on win32.
9 - c++: rename EV_THROW to EV_NOEXCEPT.
10 - c++: remove deprecated dynamic throw() specifications.
11 - c++: improve the (unsupported) bad_loop exception class.
12 - backport perl ev_periodic example to C, untested.
13
144.24 Wed Dec 28 05:19:55 CET 2016
15 - bump version to 4.24, as the release tarball inexplicably
16 didn't have the right version in ev.h, even though the cvs-tagged
17 version did have the right one (reported by Ales Teska).
18
194.23 Wed Nov 16 18:23:41 CET 2016
20 - move some declarations at the beginning to help certain retarded
21 microsoft compilers, even though their documentation claims
22 otherwise (reported by Ruslan Osmanov).
23
244.22 Sun Dec 20 22:11:50 CET 2015
25 - when epoll detects unremovable fds in the fd set, rebuild
26 only the epoll descriptor, not the signal pipe, to avoid
27 SIGPIPE in ev_async_send. This doesn't solve it on fork,
28 so document what needs to be done in ev_loop_fork
29 (analyzed by Benjamin Mahler).
30 - remove superfluous sys/timeb.h include on win32
31 (analyzed by Jason Madden).
32 - updated libecb.
33
344.20 Sat Jun 20 13:01:43 CEST 2015
35 - prefer noexcept over throw () with C++ 11.
36 - update ecb.h due to incompatibilities with c11.
37 - fix a potential aliasing issue when reading and writing
38 watcher callbacks.
39
404.19 Thu Sep 25 08:18:25 CEST 2014
41 - ev.h wasn't valid C++ anymore, which tripped compilers other than
42 clang, msvc or gcc (analyzed by Raphael 'kena' Poss). Unfortunately,
43 C++ doesn't support typedefs for function pointers fully, so the affected
44 declarations have to spell out the types each time.
45 - when not using autoconf, tighten the check for clock_gettime and related
46 functionality.
47
484.18 Fri Sep 5 17:55:26 CEST 2014
49 - events on files were not always generated properly with the
50 epoll backend (testcase by Assaf Inbal).
51 - mark event pipe fd as cloexec after a fork (analyzed by Sami Farin).
52 - (ecb) support m68k, m88k and sh (patch by Miod Vallat).
53 - use a reasonable fallback for EV_NSIG instead of erroring out
54 when we can't detect the signal set size.
55 - in the absence of autoconf, do not use the clock syscall
56 on glibc >= 2.17 (avoids the syscall AND -lrt on systems
57 doing clock_gettime in userspace).
58 - ensure extern "C" function pointers are used for externally-visible
59 loop callbacks (not watcher callbacks yet).
60 - (ecb) work around memory barriers and volatile apparently both being
61 broken in visual studio 2008 and later (analysed and patch by Nicolas Noble).
62
634.15 Fri Mar 1 12:04:50 CET 2013
64 - destroying a non-default loop would stop the global waitpid
65 watcher (Denis Bilenko).
66 - queueing pending watchers of higher priority from a watcher now invokes
67 them in a timely fashion (reported by Denis Bilenko).
68 - add throw() to all libev functions that cannot throw exceptions, for
69 further code size decrease when compiling for C++.
70 - add throw () to callbacks that must not throw exceptions (allocator,
71 syserr, loop acquire/release, periodic reschedule cbs).
72 - fix event_base_loop return code, add event_get_callback, event_base_new,
73 event_base_get_method calls to improve libevent 1.x emulation and add
74 some libevent 2.x functionality (based on a patch by Jeff Davey).
75 - add more memory fences to fix a bug reported by Jeff Davey. Better
76 be overfenced than underprotected.
77 - ev_run now returns a boolean status (true meaning watchers are
78 still active).
79 - ev_once: undef EV_ERROR in ev_kqueue.c, to avoid clashing with
80 libev's EV_ERROR (reported by 191919).
81 - (ecb) add memory fence support for xlC (Darin McBride).
82 - (ecb) add memory fence support for gcc-mips (Anton Kirilov).
83 - (ecb) add memory fence support for gcc-alpha (Christian Weisgerber).
84 - work around some kernels losing file descriptors by leaking
85 the kqueue descriptor in the child.
86 - work around linux inotify not reporting IN_ATTRIB changes for directories
87 in many cases.
88 - include sys/syscall.h instead of plain syscall.h.
89 - check for io watcher loops in ev_verify, check for the most
90 common reported usage bug in ev_io_start.
91 - choose socket vs. WSASocket at compiletime using EV_USE_WSASOCKET.
92 - always use WSASend/WSARecv directly on windows, hoping that this
93 works in all cases (unlike read/write/send/recv...).
94 - try to detect signals around a fork faster (test program by
95 Denis Bilenko).
96 - work around recent glibc versions that leak memory in realloc.
97 - rename ev::embed::set to ev::embed::set_embed to avoid clashing
98 the watcher base set (loop) method.
99 - rewrite the async/signal pipe logic to always keep a valid fd, which
100 simplifies (and hopefully correctifies :) the race checking
101 on fork, at the cost of one extra fd.
102 - add fat, msdos, jffs2, ramfs, ntfs and btrfs to the list of
103 inotify-supporting filesystems.
104 - move orig_CFLAGS assignment to after AC_INIT, as newer autoconf
105 versions ignore it before
106 (https://bugzilla.redhat.com/show_bug.cgi?id=908096).
107 - add some untested android support.
108 - enum expressions must be of type int (reported by Juan Pablo L).
109
1104.11 Sat Feb 4 19:52:39 CET 2012
111 - INCOMPATIBLE CHANGE: ev_timer_again now clears the pending status, as
112 was documented already, but not implemented in the repeating case.
113 - new compiletime symbols: EV_NO_SMP and EV_NO_THREADS.
114 - fix a race where the workaround against the epoll fork bugs
115 caused signals to not be handled anymore.
116 - correct backend_fudge for most backends, and implement a windows
117 specific workaround to avoid looping because we call both
118 select and Sleep, both with different time resolutions.
119 - document range and guarantees of ev_sleep.
120 - document reasonable ranges for periodics interval and offset.
121 - rename backend_fudge to backend_mintime to avoid future confusion :)
122 - change the default periodic reschedule function to hopefully be more
123 exact and correct even in corner cases or in the far future.
124 - do not rely on -lm anymore: use it when available but use our
125 own floor () if it is missing. This should make it easier to embed,
126 as no external libraries are required.
127 - strategically import macros from libecb and mark rarely-used functions
128 as cache-cold (saving almost 2k code size on typical amd64 setups).
129 - add Symbols.ev and Symbols.event files, that were missing.
130 - fix backend_mintime value for epoll (was 1/1024, is 1/1000 now).
131 - fix #3 "be smart about timeouts" to not "deadlock" when
132 timeout == now, also improve the section overall.
133 - avoid "AVOIDING FINISHING BEFORE RETURNING" idiom.
134 - support new EV_API_STATIC mode to make all libev symbols
135 static.
136 - supply default CFLAGS of -g -O3 with gcc when original CFLAGS
137 were empty.
138
1394.04 Wed Feb 16 09:01:51 CET 2011
140 - fix two problems in the native win32 backend, where reuse of fd's
141 with different underlying handles caused handles not to be removed
142 or added to the select set (analyzed and tested by Bert Belder).
143 - do no rely on ceil() in ev_e?poll.c.
144 - backport libev to HP-UX versions before 11 v3.
145 - configure did not detect nanosleep and clock_gettime properly when
146 they are available in the libc (as opposed to -lrt).
147
1484.03 Tue Jan 11 14:37:25 CET 2011
149 - officially support polling files with all backends.
150 - support files, /dev/zero etc. the same way as select in the epoll
151 backend, by generating events on our own.
152 - ports backend: work around solaris bug 6874410 and many related ones
153 (EINTR, maybe more), with no performance loss (note that the solaris
154 bug report is actually wrong, reality is far more bizarre and broken
155 than that).
156 - define EV_READ/EV_WRITE as macros in event.h, as some programs use
157 #ifdef to test for them.
158 - new (experimental) function: ev_feed_signal.
159 - new (to become default) EVFLAG_NOSIGMASK flag.
160 - new EVBACKEND_MASK symbol.
161 - updated COMMON IDIOMS SECTION.
162
1634.01 Fri Nov 5 21:51:29 CET 2010
164 - automake fucked it up, apparently, --add-missing -f is not quite enough
165 to make it update its files, so 4.00 didn't install ev++.h and
166 event.h on make install. grrr.
167 - ev_loop(count|depth) didn't return anything (Robin Haberkorn).
3 - change EV_UNDEF to 0xffffffff to silence some overzealous compilers. 168 - change EV_UNDEF to 0xffffffff to silence some overzealous compilers.
169 - use "(libev) " prefix for all libev error messages now.
4 170
54.00 Mon Oct 25 12:32:12 CEST 2010 1714.00 Mon Oct 25 12:32:12 CEST 2010
6 - "PORTING FROM LIBEV 3.X TO 4.X" (in ev.pod) is recommended reading. 172 - "PORTING FROM LIBEV 3.X TO 4.X" (in ev.pod) is recommended reading.
7 - ev_embed_stop did not correctly stop the watcher (very good 173 - ev_embed_stop did not correctly stop the watcher (very good
8 testcase by Vladimir Timofeev). 174 testcase by Vladimir Timofeev).
… …
62 catch race conditions, but the users have spoken. 228 catch race conditions, but the users have spoken.
63 - point out the unspecified signal mask in the documentation, and 229 - point out the unspecified signal mask in the documentation, and
64 that this is a race condition regardless of EV_SIGNALFD. 230 that this is a race condition regardless of EV_SIGNALFD.
65 - backport inotify code to C89. 231 - backport inotify code to C89.
66 - inotify file descriptors could leak into child processes. 232 - inotify file descriptors could leak into child processes.
67 - ev_stat watchers could keep an errornous extra ref on the loop, 233 - ev_stat watchers could keep an erroneous extra ref on the loop,
68 preventing exit when unregistering all watchers (testcases 234 preventing exit when unregistering all watchers (testcases
69 provided by ry@tinyclouds.org). 235 provided by ry@tinyclouds.org).
70 - implement EV_WIN32_HANDLE_TO_FD and EV_WIN32_CLOSE_FD configuration 236 - implement EV_WIN32_HANDLE_TO_FD and EV_WIN32_CLOSE_FD configuration
71 symbols to make it easier for apps to do their own fd management. 237 symbols to make it easier for apps to do their own fd management.
72 - support EV_IDLE_ENABLE being disabled in ev++.h 238 - support EV_IDLE_ENABLE being disabled in ev++.h
… …
130 - correctly zero the idx values for arrays, so destroying and 296 - correctly zero the idx values for arrays, so destroying and
131 reinitialising the default loop actually works (patch by 297 reinitialising the default loop actually works (patch by
132 Malek Hadj-Ali). 298 Malek Hadj-Ali).
133 - implement ev_suspend and ev_resume. 299 - implement ev_suspend and ev_resume.
134 - new EV_CUSTOM revents flag for use by applications. 300 - new EV_CUSTOM revents flag for use by applications.
135 - add documentation section about priorites. 301 - add documentation section about priorities.
136 - add a glossary to the dcoumentation. 302 - add a glossary to the documentation.
137 - extend the ev_fork description slightly. 303 - extend the ev_fork description slightly.
138 - optimize a jump out of call_pending. 304 - optimize a jump out of call_pending.
139 305
1403.53 Sun Feb 15 02:38:20 CET 2009 3063.53 Sun Feb 15 02:38:20 CET 2009
141 - fix a bug in event pipe creation on win32 that would cause a 307 - fix a bug in event pipe creation on win32 that would cause a

Diff Legend

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