| 1 | Revision history for libev, a high-performance and full-featured event loop. |
1 | Revision history for libev, a high-performance and full-featured event loop. |
| 2 | |
2 | |
| 3 | TODO: ev_loop_wakeup |
3 | - ANDROID => __ANDROID__ (reported by enh@google.com). |
| 4 | TODO: EV_STANDALONE == NO_HASSEL (do not use clock_gettime in ev_standalone) |
4 | |
|
|
5 | 4.24 Wed Dec 28 05:19:55 CET 2016 |
|
|
6 | - bump version to 4.24, as the release tarball inexplicably |
|
|
7 | didn't have the right version in ev.h, even though the cvs-tagged |
|
|
8 | version did have the right one (reported by Ales Teska). |
|
|
9 | |
|
|
10 | 4.23 Wed Nov 16 18:23:41 CET 2016 |
|
|
11 | - move some declarations at the beginning to help certain retarded |
|
|
12 | microsoft compilers, even though their documentation claims |
|
|
13 | otherwise (reported by Ruslan Osmanov). |
|
|
14 | |
|
|
15 | 4.22 Sun Dec 20 22:11:50 CET 2015 |
|
|
16 | - when epoll detects unremovable fds in the fd set, rebuild |
|
|
17 | only the epoll descriptor, not the signal pipe, to avoid |
|
|
18 | SIGPIPE in ev_async_send. This doesn't solve it on fork, |
|
|
19 | so document what needs to be done in ev_loop_fork |
|
|
20 | (analyzed by Benjamin Mahler). |
|
|
21 | - remove superfluous sys/timeb.h include on win32 |
|
|
22 | (analyzed by Jason Madden). |
|
|
23 | - updated libecb. |
|
|
24 | |
|
|
25 | 4.20 Sat Jun 20 13:01:43 CEST 2015 |
|
|
26 | - prefer noexcept over throw () with C++ 11. |
|
|
27 | - update ecb.h due to incompatibilities with c11. |
|
|
28 | - fix a potential aliasing issue when reading and writing |
|
|
29 | watcher callbacks. |
|
|
30 | |
|
|
31 | 4.19 Thu Sep 25 08:18:25 CEST 2014 |
|
|
32 | - ev.h wasn't valid C++ anymore, which tripped compilers other than |
|
|
33 | clang, msvc or gcc (analyzed by Raphael 'kena' Poss). Unfortunately, |
|
|
34 | C++ doesn't support typedefs for function pointers fully, so the affected |
|
|
35 | declarations have to spell out the types each time. |
|
|
36 | - when not using autoconf, tighten the check for clock_gettime and related |
|
|
37 | functionality. |
|
|
38 | |
|
|
39 | 4.18 Fri Sep 5 17:55:26 CEST 2014 |
|
|
40 | - events on files were not always generated properly with the |
|
|
41 | epoll backend (testcase by Assaf Inbal). |
|
|
42 | - mark event pipe fd as cloexec after a fork (analyzed by Sami Farin). |
|
|
43 | - (ecb) support m68k, m88k and sh (patch by Miod Vallat). |
|
|
44 | - use a reasonable fallback for EV_NSIG instead of erroring out |
|
|
45 | when we can't detect the signal set size. |
|
|
46 | - in the absence of autoconf, do not use the clock syscall |
|
|
47 | on glibc >= 2.17 (avoids the syscall AND -lrt on systems |
|
|
48 | doing clock_gettime in userspace). |
|
|
49 | - ensure extern "C" function pointers are used for externally-visible |
|
|
50 | loop callbacks (not watcher callbacks yet). |
|
|
51 | - (ecb) work around memory barriers and volatile apparently both being |
|
|
52 | broken in visual studio 2008 and later (analysed and patch by Nicolas Noble). |
|
|
53 | |
|
|
54 | 4.15 Fri Mar 1 12:04:50 CET 2013 |
|
|
55 | - destroying a non-default loop would stop the global waitpid |
|
|
56 | watcher (Denis Bilenko). |
|
|
57 | - queueing pending watchers of higher priority from a watcher now invokes |
|
|
58 | them in a timely fashion (reported by Denis Bilenko). |
|
|
59 | - add throw() to all libev functions that cannot throw exceptions, for |
|
|
60 | further code size decrease when compiling for C++. |
|
|
61 | - add throw () to callbacks that must not throw exceptions (allocator, |
|
|
62 | syserr, loop acquire/release, periodic reschedule cbs). |
|
|
63 | - fix event_base_loop return code, add event_get_callback, event_base_new, |
|
|
64 | event_base_get_method calls to improve libevent 1.x emulation and add |
|
|
65 | some libevent 2.x functionality (based on a patch by Jeff Davey). |
|
|
66 | - add more memory fences to fix a bug reported by Jeff Davey. Better |
|
|
67 | be overfenced than underprotected. |
|
|
68 | - ev_run now returns a boolean status (true meaning watchers are |
|
|
69 | still active). |
|
|
70 | - ev_once: undef EV_ERROR in ev_kqueue.c, to avoid clashing with |
|
|
71 | libev's EV_ERROR (reported by 191919). |
|
|
72 | - (ecb) add memory fence support for xlC (Darin McBride). |
|
|
73 | - (ecb) add memory fence support for gcc-mips (Anton Kirilov). |
|
|
74 | - (ecb) add memory fence support for gcc-alpha (Christian Weisgerber). |
|
|
75 | - work around some kernels losing file descriptors by leaking |
|
|
76 | the kqueue descriptor in the child. |
|
|
77 | - work around linux inotify not reporting IN_ATTRIB changes for directories |
|
|
78 | in many cases. |
|
|
79 | - include sys/syscall.h instead of plain syscall.h. |
|
|
80 | - check for io watcher loops in ev_verify, check for the most |
|
|
81 | common reported usage bug in ev_io_start. |
|
|
82 | - choose socket vs. WSASocket at compiletime using EV_USE_WSASOCKET. |
|
|
83 | - always use WSASend/WSARecv directly on windows, hoping that this |
|
|
84 | works in all cases (unlike read/write/send/recv...). |
|
|
85 | - try to detect signals around a fork faster (test program by |
|
|
86 | Denis Bilenko). |
|
|
87 | - work around recent glibc versions that leak memory in realloc. |
|
|
88 | - rename ev::embed::set to ev::embed::set_embed to avoid clashing |
|
|
89 | the watcher base set (loop) method. |
|
|
90 | - rewrite the async/signal pipe logic to always keep a valid fd, which |
|
|
91 | simplifies (and hopefully correctifies :) the race checking |
|
|
92 | on fork, at the cost of one extra fd. |
|
|
93 | - add fat, msdos, jffs2, ramfs, ntfs and btrfs to the list of |
|
|
94 | inotify-supporting filesystems. |
|
|
95 | - move orig_CFLAGS assignment to after AC_INIT, as newer autoconf |
|
|
96 | versions ignore it before |
|
|
97 | (https://bugzilla.redhat.com/show_bug.cgi?id=908096). |
|
|
98 | - add some untested android support. |
|
|
99 | - enum expressions must be of type int (reported by Juan Pablo L). |
| 5 | |
100 | |
| 6 | 4.11 Sat Feb 4 19:52:39 CET 2012 |
101 | 4.11 Sat Feb 4 19:52:39 CET 2012 |
| 7 | - INCOMPATIBLE CHANGE: ev_timer_again now clears the pending status, as |
102 | - INCOMPATIBLE CHANGE: ev_timer_again now clears the pending status, as |
| 8 | was documented already, but not implemented in the repeating case. |
103 | was documented already, but not implemented in the repeating case. |
| 9 | - new compiletime symbols: EV_NO_SMP and EV_NO_THREADS. |
104 | - new compiletime symbols: EV_NO_SMP and EV_NO_THREADS. |
| … | |
… | |
| 193 | reinitialising the default loop actually works (patch by |
288 | reinitialising the default loop actually works (patch by |
| 194 | Malek Hadj-Ali). |
289 | Malek Hadj-Ali). |
| 195 | - implement ev_suspend and ev_resume. |
290 | - implement ev_suspend and ev_resume. |
| 196 | - new EV_CUSTOM revents flag for use by applications. |
291 | - new EV_CUSTOM revents flag for use by applications. |
| 197 | - add documentation section about priorities. |
292 | - add documentation section about priorities. |
| 198 | - add a glossary to the dcoumentation. |
293 | - add a glossary to the documentation. |
| 199 | - extend the ev_fork description slightly. |
294 | - extend the ev_fork description slightly. |
| 200 | - optimize a jump out of call_pending. |
295 | - optimize a jump out of call_pending. |
| 201 | |
296 | |
| 202 | 3.53 Sun Feb 15 02:38:20 CET 2009 |
297 | 3.53 Sun Feb 15 02:38:20 CET 2009 |
| 203 | - fix a bug in event pipe creation on win32 that would cause a |
298 | - fix a bug in event pipe creation on win32 that would cause a |
