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

Comparing libev/Changes (file contents):
Revision 1.265 by root, Thu May 3 16:00:47 2012 UTC vs.
Revision 1.331 by root, Wed Jun 26 07:37:02 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: EV_STANDALONE == NO_HASSEL (do not use clock_gettime in ev_standalone) 4 - linux aio backend almost complete rewritten to work around its
5 limitations.
6 - epoll backend now mandatory for linux aio backend.
7 - fail assertions more aggressively on invalid fd's detected
8 in the event loop, do not just silently fd_kill in case of
9 user error.
10 - ev_io_start/ev_io_stop now verify the watcher fd using
11 a syscall when EV_VERIFY is 2 or higher.
5 12
6TODO: document WSA_EV_USE_SOCKET in win32 part 134.26 (EV only)
7TODO: ^ OR use WSASend/WSARecv on the handle, which always works 14 - update to libecb 0x00010006.
8TODO: assert on fd watcher linked list pointing to itself 15 - new experimental linux aio backend (linux 4.18+).
9TODO: faq, process a thing in each iteration 16 - removed redundant 0-ptr check in ev_once.
10TODO: abuse idle watchers for their side effect 17 - updated/extended ev_set_allocator documentation.
11TODO: abuse check watchers for their side effect 18 - replaced EMPTY2 macro by array_needsize_noinit.
19 - minor code cleanups.
20 - epoll backend now uses epoll_create1 also after fork.
21
224.25 Fri Dec 21 07:49:20 CET 2018
23 - INCOMPATIBLE CHANGE: EV_THROW was renamed to EV_NOEXCEPT
24 (EV_THROW still provided) and now uses noexcept on C++11 or newer.
25 - move the darwin select workaround highe rin ev.c, as newer versions of
26 darwin managed to break their broken select even more.
27 - ANDROID => __ANDROID__ (reported by enh@google.com).
28 - disable epoll_create1 on android because it has broken header files
29 and google is unwilling to fix them (reported by enh@google.com).
30 - avoid a minor compilation warning on win32.
31 - c++: remove deprecated dynamic throw() specifications.
32 - c++: improve the (unsupported) bad_loop exception class.
33 - backport perl ev_periodic example to C, untested.
34 - update libecb, biggets change is to include a memory fence
35 in ECB_MEMORY_FENCE_RELEASE on x86/amd64.
36 - minor autoconf/automake modernisation.
37
384.24 Wed Dec 28 05:19:55 CET 2016
39 - bump version to 4.24, as the release tarball inexplicably
40 didn't have the right version in ev.h, even though the cvs-tagged
41 version did have the right one (reported by Ales Teska).
42
434.23 Wed Nov 16 18:23:41 CET 2016
44 - move some declarations at the beginning to help certain retarded
45 microsoft compilers, even though their documentation claims
46 otherwise (reported by Ruslan Osmanov).
47
484.22 Sun Dec 20 22:11:50 CET 2015
49 - when epoll detects unremovable fds in the fd set, rebuild
50 only the epoll descriptor, not the signal pipe, to avoid
51 SIGPIPE in ev_async_send. This doesn't solve it on fork,
52 so document what needs to be done in ev_loop_fork
53 (analyzed by Benjamin Mahler).
54 - remove superfluous sys/timeb.h include on win32
55 (analyzed by Jason Madden).
56 - updated libecb.
57
584.20 Sat Jun 20 13:01:43 CEST 2015
59 - prefer noexcept over throw () with C++ 11.
60 - update ecb.h due to incompatibilities with c11.
61 - fix a potential aliasing issue when reading and writing
62 watcher callbacks.
63
644.19 Thu Sep 25 08:18:25 CEST 2014
65 - ev.h wasn't valid C++ anymore, which tripped compilers other than
66 clang, msvc or gcc (analyzed by Raphael 'kena' Poss). Unfortunately,
67 C++ doesn't support typedefs for function pointers fully, so the affected
68 declarations have to spell out the types each time.
69 - when not using autoconf, tighten the check for clock_gettime and related
70 functionality.
71
724.18 Fri Sep 5 17:55:26 CEST 2014
73 - events on files were not always generated properly with the
74 epoll backend (testcase by Assaf Inbal).
75 - mark event pipe fd as cloexec after a fork (analyzed by Sami Farin).
76 - (ecb) support m68k, m88k and sh (patch by Miod Vallat).
77 - use a reasonable fallback for EV_NSIG instead of erroring out
78 when we can't detect the signal set size.
79 - in the absence of autoconf, do not use the clock syscall
80 on glibc >= 2.17 (avoids the syscall AND -lrt on systems
81 doing clock_gettime in userspace).
82 - ensure extern "C" function pointers are used for externally-visible
83 loop callbacks (not watcher callbacks yet).
84 - (ecb) work around memory barriers and volatile apparently both being
85 broken in visual studio 2008 and later (analysed and patch by Nicolas Noble).
86
874.15 Fri Mar 1 12:04:50 CET 2013
88 - destroying a non-default loop would stop the global waitpid
89 watcher (Denis Bilenko).
90 - queueing pending watchers of higher priority from a watcher now invokes
91 them in a timely fashion (reported by Denis Bilenko).
12 - add throw() to all libev functions that cannot throw exceptions, for 92 - add throw() to all libev functions that cannot throw exceptions, for
13 further code size decrease when compiling for C++. 93 further code size decrease when compiling for C++.
14 - add throw () to callbacks that must not throw exceptions (allocator, 94 - add throw () to callbacks that must not throw exceptions (allocator,
15 syserr, loop acquire/release, periodic reschedule cbs). 95 syserr, loop acquire/release, periodic reschedule cbs).
16 - fix event_base_loop return code, add event_get_callback, event_base_new, 96 - fix event_base_loop return code, add event_get_callback, event_base_new,
17 event_base_get_method calls to improve libevent 1.x emulation and add 97 event_base_get_method calls to improve libevent 1.x emulation and add
18 some libevent 2.x funcitonality (based on a patch by Jeff Davey). 98 some libevent 2.x functionality (based on a patch by Jeff Davey).
19 - add more memory fences to fix a bug reported by Jeff Davey. Better 99 - add more memory fences to fix a bug reported by Jeff Davey. Better
20 be overfenced than underprotected. 100 be overfenced than underprotected.
21 - ev_run now returns a boolean status (true meaning watchers are 101 - ev_run now returns a boolean status (true meaning watchers are
22 still active). 102 still active).
23 - ev_once: undef EV_ERROR in ev_kqueue.c, to avoid clashing with 103 - ev_once: undef EV_ERROR in ev_kqueue.c, to avoid clashing with
… …
25 - (ecb) add memory fence support for xlC (Darin McBride). 105 - (ecb) add memory fence support for xlC (Darin McBride).
26 - (ecb) add memory fence support for gcc-mips (Anton Kirilov). 106 - (ecb) add memory fence support for gcc-mips (Anton Kirilov).
27 - (ecb) add memory fence support for gcc-alpha (Christian Weisgerber). 107 - (ecb) add memory fence support for gcc-alpha (Christian Weisgerber).
28 - work around some kernels losing file descriptors by leaking 108 - work around some kernels losing file descriptors by leaking
29 the kqueue descriptor in the child. 109 the kqueue descriptor in the child.
110 - work around linux inotify not reporting IN_ATTRIB changes for directories
111 in many cases.
30 - include sys/syscall.h instead of plain syscall.h. 112 - include sys/syscall.h instead of plain syscall.h.
113 - check for io watcher loops in ev_verify, check for the most
114 common reported usage bug in ev_io_start.
115 - choose socket vs. WSASocket at compiletime using EV_USE_WSASOCKET.
116 - always use WSASend/WSARecv directly on windows, hoping that this
117 works in all cases (unlike read/write/send/recv...).
118 - try to detect signals around a fork faster (test program by
119 Denis Bilenko).
120 - work around recent glibc versions that leak memory in realloc.
121 - rename ev::embed::set to ev::embed::set_embed to avoid clashing
122 the watcher base set (loop) method.
123 - rewrite the async/signal pipe logic to always keep a valid fd, which
124 simplifies (and hopefully correctifies :) the race checking
125 on fork, at the cost of one extra fd.
126 - add fat, msdos, jffs2, ramfs, ntfs and btrfs to the list of
127 inotify-supporting filesystems.
128 - move orig_CFLAGS assignment to after AC_INIT, as newer autoconf
129 versions ignore it before
130 (https://bugzilla.redhat.com/show_bug.cgi?id=908096).
131 - add some untested android support.
132 - enum expressions must be of type int (reported by Juan Pablo L).
31 133
324.11 Sat Feb 4 19:52:39 CET 2012 1344.11 Sat Feb 4 19:52:39 CET 2012
33 - INCOMPATIBLE CHANGE: ev_timer_again now clears the pending status, as 135 - INCOMPATIBLE CHANGE: ev_timer_again now clears the pending status, as
34 was documented already, but not implemented in the repeating case. 136 was documented already, but not implemented in the repeating case.
35 - new compiletime symbols: EV_NO_SMP and EV_NO_THREADS. 137 - new compiletime symbols: EV_NO_SMP and EV_NO_THREADS.
… …
219 reinitialising the default loop actually works (patch by 321 reinitialising the default loop actually works (patch by
220 Malek Hadj-Ali). 322 Malek Hadj-Ali).
221 - implement ev_suspend and ev_resume. 323 - implement ev_suspend and ev_resume.
222 - new EV_CUSTOM revents flag for use by applications. 324 - new EV_CUSTOM revents flag for use by applications.
223 - add documentation section about priorities. 325 - add documentation section about priorities.
224 - add a glossary to the dcoumentation. 326 - add a glossary to the documentation.
225 - extend the ev_fork description slightly. 327 - extend the ev_fork description slightly.
226 - optimize a jump out of call_pending. 328 - optimize a jump out of call_pending.
227 329
2283.53 Sun Feb 15 02:38:20 CET 2009 3303.53 Sun Feb 15 02:38:20 CET 2009
229 - fix a bug in event pipe creation on win32 that would cause a 331 - fix a bug in event pipe creation on win32 that would cause a

Diff Legend

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