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

Comparing libev/Changes (file contents):
Revision 1.246 by root, Tue Dec 20 04:09:13 2011 UTC vs.
Revision 1.343 by root, Sat Aug 17 05:30:16 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 modern powersaving
4TODO: maybe use timerfd to detect time jumps on linux
5TODO: document EV_TSTAMP_T
6 - linuxaio backend resulted in random memory corruption
7 when loop is forked.
8 - linuxaio backend might have tried to cancel an iocb
9 multiple times (was unable to trigger this).
10 - linuxaio backend now employs a generation counter to
11 avoid handling spurious events from cancelled requests.
12 - io_cancel can return EINTR, deal with it. also, assume
13 io_submit also returns EINTR.
14 - fix some other minor bugs in linuxaio backend.
15 - ev_tstamp type can now be overriden by defining EV_TSTAMP_T.
16 - cleanup: replace expect_true/false and noinline by their
17 libecb counterparts.
18 - move syscall infrastructure from ev_linuxaio.c to ev.c.
19 - prepare io_uring integration.
20 - tweak ev_floor.
21 - epoll, poll, win32 Sleep and other places that use millisecond
22 reslution now all try to round up times.
23 - solaris port backend didn't compile.
24 - abstract time constants into their macros, for more flexibility.
4 25
5TODO: EV_NO_THREADS/NO_SMP and/or EV_USE_PTHREADS or so. also document. 264.27 Thu Jun 27 22:43:44 CEST 2019
6TODO: EV_STANDALONE == NO_HASSEL (do not use clock_gettime in ev_standalone) 27 - linux aio backend almost completely rewritten to work around its
7TODO: memory fences for clang 28 limitations.
8ASKEXG!TODO: #define EV_API_DECL static? ev_default_loop_ptr? ev_rt_now? 29 - linux aio backend now requires linux 4.19+.
9ASKEXG!TODO: fix repeat=0 in http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod#Be_smart_about_timeouts 30 - epoll backend now mandatory for linux aio backend.
104.10 31 - fail assertions more aggressively on invalid fd's detected
32 in the event loop, do not just silently fd_kill in case of
33 user error.
34 - ev_io_start/ev_io_stop now verify the watcher fd using
35 a syscall when EV_VERIFY is 2 or higher.
36
374.26 (EV only)
38 - update to libecb 0x00010006.
39 - new experimental linux aio backend (linux 4.18+).
40 - removed redundant 0-ptr check in ev_once.
41 - updated/extended ev_set_allocator documentation.
42 - replaced EMPTY2 macro by array_needsize_noinit.
43 - minor code cleanups.
44 - epoll backend now uses epoll_create1 also after fork.
45
464.25 Fri Dec 21 07:49:20 CET 2018
47 - INCOMPATIBLE CHANGE: EV_THROW was renamed to EV_NOEXCEPT
48 (EV_THROW still provided) and now uses noexcept on C++11 or newer.
49 - move the darwin select workaround higher in ev.c, as newer versions of
50 darwin managed to break their broken select even more.
51 - ANDROID => __ANDROID__ (reported by enh@google.com).
52 - disable epoll_create1 on android because it has broken header files
53 and google is unwilling to fix them (reported by enh@google.com).
54 - avoid a minor compilation warning on win32.
55 - c++: remove deprecated dynamic throw() specifications.
56 - c++: improve the (unsupported) bad_loop exception class.
57 - backport perl ev_periodic example to C, untested.
58 - update libecb, biggets change is to include a memory fence
59 in ECB_MEMORY_FENCE_RELEASE on x86/amd64.
60 - minor autoconf/automake modernisation.
61
624.24 Wed Dec 28 05:19:55 CET 2016
63 - bump version to 4.24, as the release tarball inexplicably
64 didn't have the right version in ev.h, even though the cvs-tagged
65 version did have the right one (reported by Ales Teska).
66
674.23 Wed Nov 16 18:23:41 CET 2016
68 - move some declarations at the beginning to help certain retarded
69 microsoft compilers, even though their documentation claims
70 otherwise (reported by Ruslan Osmanov).
71
724.22 Sun Dec 20 22:11:50 CET 2015
73 - when epoll detects unremovable fds in the fd set, rebuild
74 only the epoll descriptor, not the signal pipe, to avoid
75 SIGPIPE in ev_async_send. This doesn't solve it on fork,
76 so document what needs to be done in ev_loop_fork
77 (analyzed by Benjamin Mahler).
78 - remove superfluous sys/timeb.h include on win32
79 (analyzed by Jason Madden).
80 - updated libecb.
81
824.20 Sat Jun 20 13:01:43 CEST 2015
83 - prefer noexcept over throw () with C++ 11.
84 - update ecb.h due to incompatibilities with c11.
85 - fix a potential aliasing issue when reading and writing
86 watcher callbacks.
87
884.19 Thu Sep 25 08:18:25 CEST 2014
89 - ev.h wasn't valid C++ anymore, which tripped compilers other than
90 clang, msvc or gcc (analyzed by Raphael 'kena' Poss). Unfortunately,
91 C++ doesn't support typedefs for function pointers fully, so the affected
92 declarations have to spell out the types each time.
93 - when not using autoconf, tighten the check for clock_gettime and related
94 functionality.
95
964.18 Fri Sep 5 17:55:26 CEST 2014
97 - events on files were not always generated properly with the
98 epoll backend (testcase by Assaf Inbal).
99 - mark event pipe fd as cloexec after a fork (analyzed by Sami Farin).
100 - (ecb) support m68k, m88k and sh (patch by Miod Vallat).
101 - use a reasonable fallback for EV_NSIG instead of erroring out
102 when we can't detect the signal set size.
103 - in the absence of autoconf, do not use the clock syscall
104 on glibc >= 2.17 (avoids the syscall AND -lrt on systems
105 doing clock_gettime in userspace).
106 - ensure extern "C" function pointers are used for externally-visible
107 loop callbacks (not watcher callbacks yet).
108 - (ecb) work around memory barriers and volatile apparently both being
109 broken in visual studio 2008 and later (analysed and patch by Nicolas Noble).
110
1114.15 Fri Mar 1 12:04:50 CET 2013
112 - destroying a non-default loop would stop the global waitpid
113 watcher (Denis Bilenko).
114 - queueing pending watchers of higher priority from a watcher now invokes
115 them in a timely fashion (reported by Denis Bilenko).
116 - add throw() to all libev functions that cannot throw exceptions, for
117 further code size decrease when compiling for C++.
118 - add throw () to callbacks that must not throw exceptions (allocator,
119 syserr, loop acquire/release, periodic reschedule cbs).
120 - fix event_base_loop return code, add event_get_callback, event_base_new,
121 event_base_get_method calls to improve libevent 1.x emulation and add
122 some libevent 2.x functionality (based on a patch by Jeff Davey).
123 - add more memory fences to fix a bug reported by Jeff Davey. Better
124 be overfenced than underprotected.
125 - ev_run now returns a boolean status (true meaning watchers are
126 still active).
127 - ev_once: undef EV_ERROR in ev_kqueue.c, to avoid clashing with
128 libev's EV_ERROR (reported by 191919).
129 - (ecb) add memory fence support for xlC (Darin McBride).
130 - (ecb) add memory fence support for gcc-mips (Anton Kirilov).
131 - (ecb) add memory fence support for gcc-alpha (Christian Weisgerber).
132 - work around some kernels losing file descriptors by leaking
133 the kqueue descriptor in the child.
134 - work around linux inotify not reporting IN_ATTRIB changes for directories
135 in many cases.
136 - include sys/syscall.h instead of plain syscall.h.
137 - check for io watcher loops in ev_verify, check for the most
138 common reported usage bug in ev_io_start.
139 - choose socket vs. WSASocket at compiletime using EV_USE_WSASOCKET.
140 - always use WSASend/WSARecv directly on windows, hoping that this
141 works in all cases (unlike read/write/send/recv...).
142 - try to detect signals around a fork faster (test program by
143 Denis Bilenko).
144 - work around recent glibc versions that leak memory in realloc.
145 - rename ev::embed::set to ev::embed::set_embed to avoid clashing
146 the watcher base set (loop) method.
147 - rewrite the async/signal pipe logic to always keep a valid fd, which
148 simplifies (and hopefully correctifies :) the race checking
149 on fork, at the cost of one extra fd.
150 - add fat, msdos, jffs2, ramfs, ntfs and btrfs to the list of
151 inotify-supporting filesystems.
152 - move orig_CFLAGS assignment to after AC_INIT, as newer autoconf
153 versions ignore it before
154 (https://bugzilla.redhat.com/show_bug.cgi?id=908096).
155 - add some untested android support.
156 - enum expressions must be of type int (reported by Juan Pablo L).
157
1584.11 Sat Feb 4 19:52:39 CET 2012
159 - INCOMPATIBLE CHANGE: ev_timer_again now clears the pending status, as
160 was documented already, but not implemented in the repeating case.
161 - new compiletime symbols: EV_NO_SMP and EV_NO_THREADS.
11 - fix a race where the workaround against the epoll fork bugs 162 - fix a race where the workaround against the epoll fork bugs
12 caused signals to not be handled anymore. 163 caused signals to not be handled anymore.
13 - correct backend_fudge for most backends, and implement a windows 164 - correct backend_fudge for most backends, and implement a windows
14 specific workaround to avoid looping because we call both 165 specific workaround to avoid looping because we call both
15 select and Sleep, both with different time resolutions. 166 select and Sleep, both with different time resolutions.
… …
28 - fix #3 "be smart about timeouts" to not "deadlock" when 179 - fix #3 "be smart about timeouts" to not "deadlock" when
29 timeout == now, also improve the section overall. 180 timeout == now, also improve the section overall.
30 - avoid "AVOIDING FINISHING BEFORE RETURNING" idiom. 181 - avoid "AVOIDING FINISHING BEFORE RETURNING" idiom.
31 - support new EV_API_STATIC mode to make all libev symbols 182 - support new EV_API_STATIC mode to make all libev symbols
32 static. 183 static.
184 - supply default CFLAGS of -g -O3 with gcc when original CFLAGS
185 were empty.
33 186
344.04 Wed Feb 16 09:01:51 CET 2011 1874.04 Wed Feb 16 09:01:51 CET 2011
35 - fix two problems in the native win32 backend, where reuse of fd's 188 - fix two problems in the native win32 backend, where reuse of fd's
36 with different underlying handles caused handles not to be removed 189 with different underlying handles caused handles not to be removed
37 or added to the select set (analyzed and tested by Bert Belder). 190 or added to the select set (analyzed and tested by Bert Belder).
… …
192 reinitialising the default loop actually works (patch by 345 reinitialising the default loop actually works (patch by
193 Malek Hadj-Ali). 346 Malek Hadj-Ali).
194 - implement ev_suspend and ev_resume. 347 - implement ev_suspend and ev_resume.
195 - new EV_CUSTOM revents flag for use by applications. 348 - new EV_CUSTOM revents flag for use by applications.
196 - add documentation section about priorities. 349 - add documentation section about priorities.
197 - add a glossary to the dcoumentation. 350 - add a glossary to the documentation.
198 - extend the ev_fork description slightly. 351 - extend the ev_fork description slightly.
199 - optimize a jump out of call_pending. 352 - optimize a jump out of call_pending.
200 353
2013.53 Sun Feb 15 02:38:20 CET 2009 3543.53 Sun Feb 15 02:38:20 CET 2009
202 - fix a bug in event pipe creation on win32 that would cause a 355 - fix a bug in event pipe creation on win32 that would cause a

Diff Legend

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