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

Comparing libev/Changes (file contents):
Revision 1.245 by root, Tue Dec 20 04:08:35 2011 UTC vs.
Revision 1.340 by root, Mon Jul 8 02:45:43 2019 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: ev_loop_wakeup 3TODO: revisit 59.x timer in the light of mdoenr powersaving
4TODO: maybe use timerfd to detect time jumps on linux
5 - linuxaio backend resulted in random memory corruption
6 when loop is forked.
7 - linuxaio backend might have tried to cancel an iocb
8 multiple times (was unable to trigger this).
9 - io_cancel can return EINTR, deal with it. also, assume
10 io_asubmit also retursn EINTR.
11 - fix some othe rminor bugs in linuxaio backend.
12 - cleanup: replace expect_true/false and noinline by their
13 libecb counterparts.
14 - move syscall infrastructure from ev_linuxaio.c to ev.c.
15 - prepare io_uring integration.
16 - tweak ev_floor.
4 17
5TODO: EV_NO_THREADS/NO_SMP and/or EV_USE_PTHREADS or so. also document. 184.27 Thu Jun 27 22:43:44 CEST 2019
6TODO: EV_STANDALONE == NO_HASSEL (do not use clock_gettime in ev_standalone) 19 - linux aio backend almost completely rewritten to work around its
7TODO: memory fences for clang 20 limitations.
8TODO: fix repeat=0 in http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#Be_smart_about_timeouts 21 - linux aio backend now requires linux 4.19+.
94.10 22 - epoll backend now mandatory for linux aio backend.
23 - fail assertions more aggressively on invalid fd's detected
24 in the event loop, do not just silently fd_kill in case of
25 user error.
26 - ev_io_start/ev_io_stop now verify the watcher fd using
27 a syscall when EV_VERIFY is 2 or higher.
28
294.26 (EV only)
30 - update to libecb 0x00010006.
31 - new experimental linux aio backend (linux 4.18+).
32 - removed redundant 0-ptr check in ev_once.
33 - updated/extended ev_set_allocator documentation.
34 - replaced EMPTY2 macro by array_needsize_noinit.
35 - minor code cleanups.
36 - epoll backend now uses epoll_create1 also after fork.
37
384.25 Fri Dec 21 07:49:20 CET 2018
39 - INCOMPATIBLE CHANGE: EV_THROW was renamed to EV_NOEXCEPT
40 (EV_THROW still provided) and now uses noexcept on C++11 or newer.
41 - move the darwin select workaround higher in ev.c, as newer versions of
42 darwin managed to break their broken select even more.
43 - ANDROID => __ANDROID__ (reported by enh@google.com).
44 - disable epoll_create1 on android because it has broken header files
45 and google is unwilling to fix them (reported by enh@google.com).
46 - avoid a minor compilation warning on win32.
47 - c++: remove deprecated dynamic throw() specifications.
48 - c++: improve the (unsupported) bad_loop exception class.
49 - backport perl ev_periodic example to C, untested.
50 - update libecb, biggets change is to include a memory fence
51 in ECB_MEMORY_FENCE_RELEASE on x86/amd64.
52 - minor autoconf/automake modernisation.
53
544.24 Wed Dec 28 05:19:55 CET 2016
55 - bump version to 4.24, as the release tarball inexplicably
56 didn't have the right version in ev.h, even though the cvs-tagged
57 version did have the right one (reported by Ales Teska).
58
594.23 Wed Nov 16 18:23:41 CET 2016
60 - move some declarations at the beginning to help certain retarded
61 microsoft compilers, even though their documentation claims
62 otherwise (reported by Ruslan Osmanov).
63
644.22 Sun Dec 20 22:11:50 CET 2015
65 - when epoll detects unremovable fds in the fd set, rebuild
66 only the epoll descriptor, not the signal pipe, to avoid
67 SIGPIPE in ev_async_send. This doesn't solve it on fork,
68 so document what needs to be done in ev_loop_fork
69 (analyzed by Benjamin Mahler).
70 - remove superfluous sys/timeb.h include on win32
71 (analyzed by Jason Madden).
72 - updated libecb.
73
744.20 Sat Jun 20 13:01:43 CEST 2015
75 - prefer noexcept over throw () with C++ 11.
76 - update ecb.h due to incompatibilities with c11.
77 - fix a potential aliasing issue when reading and writing
78 watcher callbacks.
79
804.19 Thu Sep 25 08:18:25 CEST 2014
81 - ev.h wasn't valid C++ anymore, which tripped compilers other than
82 clang, msvc or gcc (analyzed by Raphael 'kena' Poss). Unfortunately,
83 C++ doesn't support typedefs for function pointers fully, so the affected
84 declarations have to spell out the types each time.
85 - when not using autoconf, tighten the check for clock_gettime and related
86 functionality.
87
884.18 Fri Sep 5 17:55:26 CEST 2014
89 - events on files were not always generated properly with the
90 epoll backend (testcase by Assaf Inbal).
91 - mark event pipe fd as cloexec after a fork (analyzed by Sami Farin).
92 - (ecb) support m68k, m88k and sh (patch by Miod Vallat).
93 - use a reasonable fallback for EV_NSIG instead of erroring out
94 when we can't detect the signal set size.
95 - in the absence of autoconf, do not use the clock syscall
96 on glibc >= 2.17 (avoids the syscall AND -lrt on systems
97 doing clock_gettime in userspace).
98 - ensure extern "C" function pointers are used for externally-visible
99 loop callbacks (not watcher callbacks yet).
100 - (ecb) work around memory barriers and volatile apparently both being
101 broken in visual studio 2008 and later (analysed and patch by Nicolas Noble).
102
1034.15 Fri Mar 1 12:04:50 CET 2013
104 - destroying a non-default loop would stop the global waitpid
105 watcher (Denis Bilenko).
106 - queueing pending watchers of higher priority from a watcher now invokes
107 them in a timely fashion (reported by Denis Bilenko).
108 - add throw() to all libev functions that cannot throw exceptions, for
109 further code size decrease when compiling for C++.
110 - add throw () to callbacks that must not throw exceptions (allocator,
111 syserr, loop acquire/release, periodic reschedule cbs).
112 - fix event_base_loop return code, add event_get_callback, event_base_new,
113 event_base_get_method calls to improve libevent 1.x emulation and add
114 some libevent 2.x functionality (based on a patch by Jeff Davey).
115 - add more memory fences to fix a bug reported by Jeff Davey. Better
116 be overfenced than underprotected.
117 - ev_run now returns a boolean status (true meaning watchers are
118 still active).
119 - ev_once: undef EV_ERROR in ev_kqueue.c, to avoid clashing with
120 libev's EV_ERROR (reported by 191919).
121 - (ecb) add memory fence support for xlC (Darin McBride).
122 - (ecb) add memory fence support for gcc-mips (Anton Kirilov).
123 - (ecb) add memory fence support for gcc-alpha (Christian Weisgerber).
124 - work around some kernels losing file descriptors by leaking
125 the kqueue descriptor in the child.
126 - work around linux inotify not reporting IN_ATTRIB changes for directories
127 in many cases.
128 - include sys/syscall.h instead of plain syscall.h.
129 - check for io watcher loops in ev_verify, check for the most
130 common reported usage bug in ev_io_start.
131 - choose socket vs. WSASocket at compiletime using EV_USE_WSASOCKET.
132 - always use WSASend/WSARecv directly on windows, hoping that this
133 works in all cases (unlike read/write/send/recv...).
134 - try to detect signals around a fork faster (test program by
135 Denis Bilenko).
136 - work around recent glibc versions that leak memory in realloc.
137 - rename ev::embed::set to ev::embed::set_embed to avoid clashing
138 the watcher base set (loop) method.
139 - rewrite the async/signal pipe logic to always keep a valid fd, which
140 simplifies (and hopefully correctifies :) the race checking
141 on fork, at the cost of one extra fd.
142 - add fat, msdos, jffs2, ramfs, ntfs and btrfs to the list of
143 inotify-supporting filesystems.
144 - move orig_CFLAGS assignment to after AC_INIT, as newer autoconf
145 versions ignore it before
146 (https://bugzilla.redhat.com/show_bug.cgi?id=908096).
147 - add some untested android support.
148 - enum expressions must be of type int (reported by Juan Pablo L).
149
1504.11 Sat Feb 4 19:52:39 CET 2012
151 - INCOMPATIBLE CHANGE: ev_timer_again now clears the pending status, as
152 was documented already, but not implemented in the repeating case.
153 - new compiletime symbols: EV_NO_SMP and EV_NO_THREADS.
10 - fix a race where the workaround against the epoll fork bugs 154 - fix a race where the workaround against the epoll fork bugs
11 caused signals to not be handled anymore. 155 caused signals to not be handled anymore.
12 - correct backend_fudge for most backends, and implement a windows 156 - correct backend_fudge for most backends, and implement a windows
13 specific workaround to avoid looping because we call both 157 specific workaround to avoid looping because we call both
14 select and Sleep, both with different time resolutions. 158 select and Sleep, both with different time resolutions.
27 - fix #3 "be smart about timeouts" to not "deadlock" when 171 - fix #3 "be smart about timeouts" to not "deadlock" when
28 timeout == now, also improve the section overall. 172 timeout == now, also improve the section overall.
29 - avoid "AVOIDING FINISHING BEFORE RETURNING" idiom. 173 - avoid "AVOIDING FINISHING BEFORE RETURNING" idiom.
30 - support new EV_API_STATIC mode to make all libev symbols 174 - support new EV_API_STATIC mode to make all libev symbols
31 static. 175 static.
176 - supply default CFLAGS of -g -O3 with gcc when original CFLAGS
177 were empty.
32 178
334.04 Wed Feb 16 09:01:51 CET 2011 1794.04 Wed Feb 16 09:01:51 CET 2011
34 - fix two problems in the native win32 backend, where reuse of fd's 180 - fix two problems in the native win32 backend, where reuse of fd's
35 with different underlying handles caused handles not to be removed 181 with different underlying handles caused handles not to be removed
36 or added to the select set (analyzed and tested by Bert Belder). 182 or added to the select set (analyzed and tested by Bert Belder).
191 reinitialising the default loop actually works (patch by 337 reinitialising the default loop actually works (patch by
192 Malek Hadj-Ali). 338 Malek Hadj-Ali).
193 - implement ev_suspend and ev_resume. 339 - implement ev_suspend and ev_resume.
194 - new EV_CUSTOM revents flag for use by applications. 340 - new EV_CUSTOM revents flag for use by applications.
195 - add documentation section about priorities. 341 - add documentation section about priorities.
196 - add a glossary to the dcoumentation. 342 - add a glossary to the documentation.
197 - extend the ev_fork description slightly. 343 - extend the ev_fork description slightly.
198 - optimize a jump out of call_pending. 344 - optimize a jump out of call_pending.
199 345
2003.53 Sun Feb 15 02:38:20 CET 2009 3463.53 Sun Feb 15 02:38:20 CET 2009
201 - fix a bug in event pipe creation on win32 that would cause a 347 - fix a bug in event pipe creation on win32 that would cause a

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines