1 |
Revision history for libev, a high-performance and full-featured event loop. |
2 |
|
3 |
TODO: for next ABI/API change, consider moving EV__IOFDSSET into io->fd instead and provide a getter. |
4 |
TODO: document EV_TSTAMP_T |
5 |
TODO: https://paste.debian.net/1268417/, stdfun |
6 |
|
7 |
- libev did not compile on systems lacking clock_gettime. |
8 |
- due to a logic bug, timerfd availability did not always translate |
9 |
into fewer wakeups - libev is now more aggressive in using |
10 |
longer sleeps, even when timerfd is not in use. |
11 |
|
12 |
4.33 Wed Mar 18 13:22:29 CET 2020 |
13 |
- no changes w.r.t. 4.32. |
14 |
|
15 |
4.32 (EV only) |
16 |
- the 4.31 timerfd code wrongly changed the priority of the signal |
17 |
fd watcher, which is usually harmless unless signal fds are |
18 |
also used (found via cpan tester service). |
19 |
- the documentation wrongly claimed that user may modify fd and events |
20 |
members in io watchers when the watcher was stopped |
21 |
(found by b_jonas). |
22 |
- new ev_io_modify mutator which changes only the events member, |
23 |
which can be faster. also added ev::io::set (int events) method |
24 |
to ev++.h. |
25 |
- officially allow a zero events mask for io watchers. this should |
26 |
work with older libev versions as well but was not officially |
27 |
allowed before. |
28 |
- do not wake up every minute when timerfd is used to detect timejumps. |
29 |
- do not wake up every minute when periodics are disabled and we have |
30 |
a monotonic clock. |
31 |
- support a lot more "uncommon" compile time configurations, |
32 |
such as ev_embed enabled but ev_timer disabled. |
33 |
- use a start/stop wrapper class to reduce code duplication in |
34 |
ev++.h and make it needlessly more c++-y. |
35 |
- the linux aio backend is no longer compiled in by default. |
36 |
- update to libecb version 0x00010008. |
37 |
|
38 |
4.31 Fri Dec 20 21:58:29 CET 2019 |
39 |
- handle backends with minimum wait time a bit better by not |
40 |
waiting in the presence of already-expired timers |
41 |
(behaviour reported by Felipe Gasper). |
42 |
- new feature: use timerfd to detect timejumps quickly, |
43 |
can be disabled with the new EVFLAG_NOTIMERFD loop flag. |
44 |
- document EV_USE_SIGNALFD feature macro. |
45 |
|
46 |
4.30 (EV only) |
47 |
- change non-autoconf test for __kernel_rwf_t by testing |
48 |
LINUX_VERSION_CODE, the most direct test I could find. |
49 |
- fix a bug in the io_uring backend that polled the wrong |
50 |
backend fd, causing it to not work in many cases. |
51 |
|
52 |
4.29 (EV only) |
53 |
- add io uring autoconf and non-autoconf detection. |
54 |
- disable io_uring when some header files are too old. |
55 |
|
56 |
4.28 (EV only) |
57 |
- linuxaio backend resulted in random memory corruption |
58 |
when loop is forked. |
59 |
- linuxaio backend might have tried to cancel an iocb |
60 |
multiple times (was unable to trigger this). |
61 |
- linuxaio backend now employs a generation counter to |
62 |
avoid handling spurious events from cancelled requests. |
63 |
- io_cancel can return EINTR, deal with it. also, assume |
64 |
io_submit also returns EINTR. |
65 |
- fix some other minor bugs in linuxaio backend. |
66 |
- ev_tstamp type can now be overriden by defining EV_TSTAMP_T. |
67 |
- cleanup: replace expect_true/false and noinline by their |
68 |
libecb counterparts. |
69 |
- move syscall infrastructure from ev_linuxaio.c to ev.c. |
70 |
- prepare io_uring integration. |
71 |
- tweak ev_floor. |
72 |
- epoll, poll, win32 Sleep and other places that use millisecond |
73 |
reslution now all try to round up times. |
74 |
- solaris port backend didn't compile. |
75 |
- abstract time constants into their macros, for more flexibility. |
76 |
|
77 |
4.27 Thu Jun 27 22:43:44 CEST 2019 |
78 |
- linux aio backend almost completely rewritten to work around its |
79 |
limitations. |
80 |
- linux aio backend now requires linux 4.19+. |
81 |
- epoll backend now mandatory for linux aio backend. |
82 |
- fail assertions more aggressively on invalid fd's detected |
83 |
in the event loop, do not just silently fd_kill in case of |
84 |
user error. |
85 |
- ev_io_start/ev_io_stop now verify the watcher fd using |
86 |
a syscall when EV_VERIFY is 2 or higher. |
87 |
|
88 |
4.26 (EV only) |
89 |
- update to libecb 0x00010006. |
90 |
- new experimental linux aio backend (linux 4.18+). |
91 |
- removed redundant 0-ptr check in ev_once. |
92 |
- updated/extended ev_set_allocator documentation. |
93 |
- replaced EMPTY2 macro by array_needsize_noinit. |
94 |
- minor code cleanups. |
95 |
- epoll backend now uses epoll_create1 also after fork. |
96 |
|
97 |
4.25 Fri Dec 21 07:49:20 CET 2018 |
98 |
- INCOMPATIBLE CHANGE: EV_THROW was renamed to EV_NOEXCEPT |
99 |
(EV_THROW still provided) and now uses noexcept on C++11 or newer. |
100 |
- move the darwin select workaround higher in ev.c, as newer versions of |
101 |
darwin managed to break their broken select even more. |
102 |
- ANDROID => __ANDROID__ (reported by enh@google.com). |
103 |
- disable epoll_create1 on android because it has broken header files |
104 |
and google is unwilling to fix them (reported by enh@google.com). |
105 |
- avoid a minor compilation warning on win32. |
106 |
- c++: remove deprecated dynamic throw() specifications. |
107 |
- c++: improve the (unsupported) bad_loop exception class. |
108 |
- backport perl ev_periodic example to C, untested. |
109 |
- update libecb, biggets change is to include a memory fence |
110 |
in ECB_MEMORY_FENCE_RELEASE on x86/amd64. |
111 |
- minor autoconf/automake modernisation. |
112 |
|
113 |
4.24 Wed Dec 28 05:19:55 CET 2016 |
114 |
- bump version to 4.24, as the release tarball inexplicably |
115 |
didn't have the right version in ev.h, even though the cvs-tagged |
116 |
version did have the right one (reported by Ales Teska). |
117 |
|
118 |
4.23 Wed Nov 16 18:23:41 CET 2016 |
119 |
- move some declarations at the beginning to help certain retarded |
120 |
microsoft compilers, even though their documentation claims |
121 |
otherwise (reported by Ruslan Osmanov). |
122 |
|
123 |
4.22 Sun Dec 20 22:11:50 CET 2015 |
124 |
- when epoll detects unremovable fds in the fd set, rebuild |
125 |
only the epoll descriptor, not the signal pipe, to avoid |
126 |
SIGPIPE in ev_async_send. This doesn't solve it on fork, |
127 |
so document what needs to be done in ev_loop_fork |
128 |
(analyzed by Benjamin Mahler). |
129 |
- remove superfluous sys/timeb.h include on win32 |
130 |
(analyzed by Jason Madden). |
131 |
- updated libecb. |
132 |
|
133 |
4.20 Sat Jun 20 13:01:43 CEST 2015 |
134 |
- prefer noexcept over throw () with C++ 11. |
135 |
- update ecb.h due to incompatibilities with c11. |
136 |
- fix a potential aliasing issue when reading and writing |
137 |
watcher callbacks. |
138 |
|
139 |
4.19 Thu Sep 25 08:18:25 CEST 2014 |
140 |
- ev.h wasn't valid C++ anymore, which tripped compilers other than |
141 |
clang, msvc or gcc (analyzed by Raphael 'kena' Poss). Unfortunately, |
142 |
C++ doesn't support typedefs for function pointers fully, so the affected |
143 |
declarations have to spell out the types each time. |
144 |
- when not using autoconf, tighten the check for clock_gettime and related |
145 |
functionality. |
146 |
|
147 |
4.18 Fri Sep 5 17:55:26 CEST 2014 |
148 |
- events on files were not always generated properly with the |
149 |
epoll backend (testcase by Assaf Inbal). |
150 |
- mark event pipe fd as cloexec after a fork (analyzed by Sami Farin). |
151 |
- (ecb) support m68k, m88k and sh (patch by Miod Vallat). |
152 |
- use a reasonable fallback for EV_NSIG instead of erroring out |
153 |
when we can't detect the signal set size. |
154 |
- in the absence of autoconf, do not use the clock syscall |
155 |
on glibc >= 2.17 (avoids the syscall AND -lrt on systems |
156 |
doing clock_gettime in userspace). |
157 |
- ensure extern "C" function pointers are used for externally-visible |
158 |
loop callbacks (not watcher callbacks yet). |
159 |
- (ecb) work around memory barriers and volatile apparently both being |
160 |
broken in visual studio 2008 and later (analysed and patch by Nicolas Noble). |
161 |
|
162 |
4.15 Fri Mar 1 12:04:50 CET 2013 |
163 |
- destroying a non-default loop would stop the global waitpid |
164 |
watcher (Denis Bilenko). |
165 |
- queueing pending watchers of higher priority from a watcher now invokes |
166 |
them in a timely fashion (reported by Denis Bilenko). |
167 |
- add throw() to all libev functions that cannot throw exceptions, for |
168 |
further code size decrease when compiling for C++. |
169 |
- add throw () to callbacks that must not throw exceptions (allocator, |
170 |
syserr, loop acquire/release, periodic reschedule cbs). |
171 |
- fix event_base_loop return code, add event_get_callback, event_base_new, |
172 |
event_base_get_method calls to improve libevent 1.x emulation and add |
173 |
some libevent 2.x functionality (based on a patch by Jeff Davey). |
174 |
- add more memory fences to fix a bug reported by Jeff Davey. Better |
175 |
be overfenced than underprotected. |
176 |
- ev_run now returns a boolean status (true meaning watchers are |
177 |
still active). |
178 |
- ev_once: undef EV_ERROR in ev_kqueue.c, to avoid clashing with |
179 |
libev's EV_ERROR (reported by 191919). |
180 |
- (ecb) add memory fence support for xlC (Darin McBride). |
181 |
- (ecb) add memory fence support for gcc-mips (Anton Kirilov). |
182 |
- (ecb) add memory fence support for gcc-alpha (Christian Weisgerber). |
183 |
- work around some kernels losing file descriptors by leaking |
184 |
the kqueue descriptor in the child. |
185 |
- work around linux inotify not reporting IN_ATTRIB changes for directories |
186 |
in many cases. |
187 |
- include sys/syscall.h instead of plain syscall.h. |
188 |
- check for io watcher loops in ev_verify, check for the most |
189 |
common reported usage bug in ev_io_start. |
190 |
- choose socket vs. WSASocket at compiletime using EV_USE_WSASOCKET. |
191 |
- always use WSASend/WSARecv directly on windows, hoping that this |
192 |
works in all cases (unlike read/write/send/recv...). |
193 |
- try to detect signals around a fork faster (test program by |
194 |
Denis Bilenko). |
195 |
- work around recent glibc versions that leak memory in realloc. |
196 |
- rename ev::embed::set to ev::embed::set_embed to avoid clashing |
197 |
the watcher base set (loop) method. |
198 |
- rewrite the async/signal pipe logic to always keep a valid fd, which |
199 |
simplifies (and hopefully correctifies :) the race checking |
200 |
on fork, at the cost of one extra fd. |
201 |
- add fat, msdos, jffs2, ramfs, ntfs and btrfs to the list of |
202 |
inotify-supporting filesystems. |
203 |
- move orig_CFLAGS assignment to after AC_INIT, as newer autoconf |
204 |
versions ignore it before |
205 |
(https://bugzilla.redhat.com/show_bug.cgi?id=908096). |
206 |
- add some untested android support. |
207 |
- enum expressions must be of type int (reported by Juan Pablo L). |
208 |
|
209 |
4.11 Sat Feb 4 19:52:39 CET 2012 |
210 |
- INCOMPATIBLE CHANGE: ev_timer_again now clears the pending status, as |
211 |
was documented already, but not implemented in the repeating case. |
212 |
- new compiletime symbols: EV_NO_SMP and EV_NO_THREADS. |
213 |
- fix a race where the workaround against the epoll fork bugs |
214 |
caused signals to not be handled anymore. |
215 |
- correct backend_fudge for most backends, and implement a windows |
216 |
specific workaround to avoid looping because we call both |
217 |
select and Sleep, both with different time resolutions. |
218 |
- document range and guarantees of ev_sleep. |
219 |
- document reasonable ranges for periodics interval and offset. |
220 |
- rename backend_fudge to backend_mintime to avoid future confusion :) |
221 |
- change the default periodic reschedule function to hopefully be more |
222 |
exact and correct even in corner cases or in the far future. |
223 |
- do not rely on -lm anymore: use it when available but use our |
224 |
own floor () if it is missing. This should make it easier to embed, |
225 |
as no external libraries are required. |
226 |
- strategically import macros from libecb and mark rarely-used functions |
227 |
as cache-cold (saving almost 2k code size on typical amd64 setups). |
228 |
- add Symbols.ev and Symbols.event files, that were missing. |
229 |
- fix backend_mintime value for epoll (was 1/1024, is 1/1000 now). |
230 |
- fix #3 "be smart about timeouts" to not "deadlock" when |
231 |
timeout == now, also improve the section overall. |
232 |
- avoid "AVOIDING FINISHING BEFORE RETURNING" idiom. |
233 |
- support new EV_API_STATIC mode to make all libev symbols |
234 |
static. |
235 |
- supply default CFLAGS of -g -O3 with gcc when original CFLAGS |
236 |
were empty. |
237 |
|
238 |
4.04 Wed Feb 16 09:01:51 CET 2011 |
239 |
- fix two problems in the native win32 backend, where reuse of fd's |
240 |
with different underlying handles caused handles not to be removed |
241 |
or added to the select set (analyzed and tested by Bert Belder). |
242 |
- do no rely on ceil() in ev_e?poll.c. |
243 |
- backport libev to HP-UX versions before 11 v3. |
244 |
- configure did not detect nanosleep and clock_gettime properly when |
245 |
they are available in the libc (as opposed to -lrt). |
246 |
|
247 |
4.03 Tue Jan 11 14:37:25 CET 2011 |
248 |
- officially support polling files with all backends. |
249 |
- support files, /dev/zero etc. the same way as select in the epoll |
250 |
backend, by generating events on our own. |
251 |
- ports backend: work around solaris bug 6874410 and many related ones |
252 |
(EINTR, maybe more), with no performance loss (note that the solaris |
253 |
bug report is actually wrong, reality is far more bizarre and broken |
254 |
than that). |
255 |
- define EV_READ/EV_WRITE as macros in event.h, as some programs use |
256 |
#ifdef to test for them. |
257 |
- new (experimental) function: ev_feed_signal. |
258 |
- new (to become default) EVFLAG_NOSIGMASK flag. |
259 |
- new EVBACKEND_MASK symbol. |
260 |
- updated COMMON IDIOMS SECTION. |
261 |
|
262 |
4.01 Fri Nov 5 21:51:29 CET 2010 |
263 |
- automake fucked it up, apparently, --add-missing -f is not quite enough |
264 |
to make it update its files, so 4.00 didn't install ev++.h and |
265 |
event.h on make install. grrr. |
266 |
- ev_loop(count|depth) didn't return anything (Robin Haberkorn). |
267 |
- change EV_UNDEF to 0xffffffff to silence some overzealous compilers. |
268 |
- use "(libev) " prefix for all libev error messages now. |
269 |
|
270 |
4.00 Mon Oct 25 12:32:12 CEST 2010 |
271 |
- "PORTING FROM LIBEV 3.X TO 4.X" (in ev.pod) is recommended reading. |
272 |
- ev_embed_stop did not correctly stop the watcher (very good |
273 |
testcase by Vladimir Timofeev). |
274 |
- ev_run will now always update the current loop time - it erroneously |
275 |
didn't when idle watchers were active, causing timers not to fire. |
276 |
- fix a bug where a timeout of zero caused the timer not to fire |
277 |
in the libevent emulation (testcase by Péter Szabó). |
278 |
- applied win32 fixes by Michael Lenaghan (also James Mansion). |
279 |
- replace EV_MINIMAL by EV_FEATURES. |
280 |
- prefer EPOLL_CTL_ADD over EPOLL_CTL_MOD in some more cases, as it |
281 |
seems the former is *much* faster than the latter. |
282 |
- linux kernel version detection (for inotify bug workarounds) |
283 |
did not work properly. |
284 |
- reduce the number of spurious wake-ups with the ports backend. |
285 |
- remove dependency on sys/queue.h on freebsd (patch by Vanilla Hsu). |
286 |
- do async init within ev_async_start, not ev_async_set, which avoids |
287 |
an API quirk where the set function must be called in the C++ API |
288 |
even when there is nothing to set. |
289 |
- add (undocumented) EV_ENABLE when adding events with kqueue, |
290 |
this might help with OS X, which seems to need it despite documenting |
291 |
not to need it (helpfully pointed out by Tilghman Lesher). |
292 |
- do not use poll by default on freebsd, it's broken (what isn't |
293 |
on freebsd...). |
294 |
- allow to embed epoll on kernels >= 2.6.32. |
295 |
- configure now prepends -O3, not appends it, so one can still |
296 |
override it. |
297 |
- ev.pod: greatly expanded the portability section, added a porting |
298 |
section, a description of watcher states and made lots of minor fixes. |
299 |
- disable poll backend on AIX, the poll header spams the namespace |
300 |
and it's not worth working around dead platforms (reported |
301 |
and analyzed by Aivars Kalvans). |
302 |
- improve header file compatibility of the standalone eventfd code |
303 |
in an obscure case. |
304 |
- implement EV_AVOID_STDIO option. |
305 |
- do not use sscanf to parse linux version number (smaller, faster, |
306 |
no sscanf dependency). |
307 |
- new EV_CHILD_ENABLE and EV_SIGNAL_ENABLE configurable settings. |
308 |
- update libev.m4 HAVE_CLOCK_SYSCALL test for newer glibcs. |
309 |
- add section on accept() problems to the manpage. |
310 |
- rename EV_TIMEOUT to EV_TIMER. |
311 |
- rename ev_loop_count/depth/verify/loop/unloop. |
312 |
- remove ev_default_destroy and ev_default_fork. |
313 |
- switch to two-digit minor version. |
314 |
- work around an apparent gentoo compiler bug. |
315 |
- define _DARWIN_UNLIMITED_SELECT. just so. |
316 |
- use enum instead of #define for most constants. |
317 |
- improve compatibility to older C++ compilers. |
318 |
- (experimental) ev_run/ev_default_loop/ev_break/ev_loop_new have now |
319 |
default arguments when compiled as C++. |
320 |
- enable automake dependency tracking. |
321 |
- ev_loop_new no longer leaks memory when loop creation failed. |
322 |
- new ev_cleanup watcher type. |
323 |
|
324 |
3.9 Thu Dec 31 07:59:59 CET 2009 |
325 |
- signalfd is no longer used by default and has to be requested |
326 |
explicitly - this means that easy to catch bugs become hard to |
327 |
catch race conditions, but the users have spoken. |
328 |
- point out the unspecified signal mask in the documentation, and |
329 |
that this is a race condition regardless of EV_SIGNALFD. |
330 |
- backport inotify code to C89. |
331 |
- inotify file descriptors could leak into child processes. |
332 |
- ev_stat watchers could keep an erroneous extra ref on the loop, |
333 |
preventing exit when unregistering all watchers (testcases |
334 |
provided by ry@tinyclouds.org). |
335 |
- implement EV_WIN32_HANDLE_TO_FD and EV_WIN32_CLOSE_FD configuration |
336 |
symbols to make it easier for apps to do their own fd management. |
337 |
- support EV_IDLE_ENABLE being disabled in ev++.h |
338 |
(patch by Didier Spezia). |
339 |
- take advantage of inotify_init1, if available, to set cloexec/nonblock |
340 |
on fd creation, to avoid races. |
341 |
- the signal handling pipe wasn't always initialised under windows |
342 |
(analysed by lekma). |
343 |
- changed minimum glibc requirement from glibc 2.9 to 2.7, for |
344 |
signalfd. |
345 |
- add missing string.h include (Denis F. Latypoff). |
346 |
- only replace ev_stat.prev when we detect an actual difference, |
347 |
so prev is (almost) always different to attr. this might |
348 |
have caused the problems with 04_stat.t. |
349 |
- add ev::timer->remaining () method to C++ API. |
350 |
|
351 |
3.8 Sun Aug 9 14:30:45 CEST 2009 |
352 |
- incompatible change: do not necessarily reset signal handler |
353 |
to SIG_DFL when a sighandler is stopped. |
354 |
- ev_default_destroy did not properly free or zero some members, |
355 |
potentially causing crashes and memory corruption on repeated |
356 |
ev_default_destroy/ev_default_loop calls. |
357 |
- take advantage of signalfd on GNU/Linux systems. |
358 |
- document that the signal mask might be in an unspecified |
359 |
state when using libev's signal handling. |
360 |
- take advantage of some GNU/Linux calls to set cloexec/nonblock |
361 |
on fd creation, to avoid race conditions. |
362 |
|
363 |
3.7 Fri Jul 17 16:36:32 CEST 2009 |
364 |
- ev_unloop and ev_loop wrongly used a global variable to exit loops, |
365 |
instead of using a per-loop variable (bug caught by accident...). |
366 |
- the ev_set_io_collect_interval interpretation has changed. |
367 |
- add new functionality: ev_set_userdata, ev_userdata, |
368 |
ev_set_invoke_pending_cb, ev_set_loop_release_cb, |
369 |
ev_invoke_pending, ev_pending_count, together with a long example |
370 |
about thread locking. |
371 |
- add ev_timer_remaining (as requested by Denis F. Latypoff). |
372 |
- add ev_loop_depth. |
373 |
- calling ev_unloop in fork/prepare watchers will no longer poll |
374 |
for new events. |
375 |
- Denis F. Latypoff corrected many typos in example code snippets. |
376 |
- honor autoconf detection of EV_USE_CLOCK_SYSCALL, also double- |
377 |
check that the syscall number is available before trying to |
378 |
use it (reported by ry@tinyclouds). |
379 |
- use GetSystemTimeAsFileTime instead of _timeb on windows, for |
380 |
slightly higher accuracy. |
381 |
- properly declare ev_loop_verify and ev_now_update even when |
382 |
!EV_MULTIPLICITY. |
383 |
- do not compile in any priority code when EV_MAXPRI == EV_MINPRI. |
384 |
- support EV_MINIMAL==2 for a reduced API. |
385 |
- actually 0-initialise struct sigaction when installing signals. |
386 |
- add section on hibernate and stopped processes to ev_timer docs. |
387 |
|
388 |
3.6 Tue Apr 28 02:49:30 CEST 2009 |
389 |
- multiple timers becoming ready within an event loop iteration |
390 |
will be invoked in the "correct" order now. |
391 |
- do not leave the event loop early just because we have no active |
392 |
watchers, fixing a problem when embedding a kqueue loop |
393 |
that has active kernel events but no registered watchers |
394 |
(reported by blacksand blacksand). |
395 |
- correctly zero the idx values for arrays, so destroying and |
396 |
reinitialising the default loop actually works (patch by |
397 |
Malek Hadj-Ali). |
398 |
- implement ev_suspend and ev_resume. |
399 |
- new EV_CUSTOM revents flag for use by applications. |
400 |
- add documentation section about priorities. |
401 |
- add a glossary to the documentation. |
402 |
- extend the ev_fork description slightly. |
403 |
- optimize a jump out of call_pending. |
404 |
|
405 |
3.53 Sun Feb 15 02:38:20 CET 2009 |
406 |
- fix a bug in event pipe creation on win32 that would cause a |
407 |
failed assertion on event loop creation (patch by Malek Hadj-Ali). |
408 |
- probe for CLOCK_REALTIME support at runtime as well and fall |
409 |
back to gettimeofday if there is an error, to support older |
410 |
operating systems with newer header files/libraries. |
411 |
- prefer gettimeofday over clock_gettime with USE_CLOCK_SYSCALL |
412 |
(default most everywhere), otherwise not. |
413 |
|
414 |
3.52 Wed Jan 7 21:43:02 CET 2009 |
415 |
- fix compilation of select backend in fd_set mode when NFDBITS is |
416 |
missing (to get it to compile on QNX, reported by Rodrigo Campos). |
417 |
- better select-nfds handling when select backend is in fd_set mode. |
418 |
- diagnose fd_set overruns when select backend is in fd_set mode. |
419 |
- due to a thinko, instead of disabling everything but |
420 |
select on the borked OS X platform, everything but select was |
421 |
allowed (reported by Emanuele Giaquinta). |
422 |
- actually verify that local and remote port are matching in |
423 |
libev's socketpair emulation, which makes denial-of-service |
424 |
attacks harder (but not impossible - it's windows). Make sure |
425 |
it even works under vista, which thinks that getpeer/sockname |
426 |
should return fantasy port numbers. |
427 |
- include "libev" in all assertion messages for potentially |
428 |
clearer diagnostics. |
429 |
- event_get_version (libevent compatibility) returned |
430 |
a useless string instead of the expected version string |
431 |
(patch by W.C.A. Wijngaards). |
432 |
|
433 |
3.51 Wed Dec 24 23:00:11 CET 2008 |
434 |
- fix a bug where an inotify watcher was added twice, causing |
435 |
freezes on hash collisions (reported and analysed by Graham Leggett). |
436 |
- new config symbol, EV_USE_CLOCK_SYSCALL, to make libev use |
437 |
a direct syscall - slower, but no dependency on librt et al. |
438 |
- assume negative return values != -1 signals success of port_getn |
439 |
(http://cvs.epicsol.org/cgi/viewcvs.cgi/epic5/source/newio.c?rev=1.52) |
440 |
(no known failure reports, but it doesn't hurt). |
441 |
- fork detection in ev_embed now stops and restarts the watcher |
442 |
automatically. |
443 |
- EXPERIMENTAL: default the method to operator () in ev++.h, |
444 |
to make it nicer to use functors (requested by Benedek László). |
445 |
- fixed const object callbacks in ev++.h. |
446 |
- replaced loop_ref argument of watcher.set (loop) by a direct |
447 |
ev_loop * in ev++.h, to avoid clashes with functor patch. |
448 |
- do not try to watch the empty string via inotify. |
449 |
- inotify watchers could be leaked under certain circumstances. |
450 |
- OS X 10.5 is actually even more broken than earlier versions, |
451 |
so fall back to select on that piece of garbage. |
452 |
- fixed some weirdness in the ev_embed documentation. |
453 |
|
454 |
3.49 Wed Nov 19 11:26:53 CET 2008 |
455 |
- ev_stat watchers will now use inotify as a mere hint on |
456 |
kernels <2.6.25, or if the filesystem is not in the |
457 |
"known to be good" list. |
458 |
- better mingw32 compatibility (it's not as borked as native win32) |
459 |
(analysed by Roger Pack). |
460 |
- include stdio.h in the example program, as too many people are |
461 |
confused by the weird C language otherwise. I guess the next thing |
462 |
I get told is that the "..." ellipses in the examples don't compile |
463 |
with their C compiler. |
464 |
|
465 |
3.48 Thu Oct 30 09:02:37 CET 2008 |
466 |
- further optimise away the EPOLL_CTL_ADD/MOD combo in the epoll |
467 |
backend by assuming the kernel event mask hasn't changed if |
468 |
ADD fails with EEXIST. |
469 |
- work around spurious event notification bugs in epoll by using |
470 |
a 32-bit generation counter. recreate kernel state if we receive |
471 |
spurious notifications or unwanted events. this is very costly, |
472 |
but I didn't come up with this horrible design. |
473 |
- use memset to initialise most arrays now and do away with the |
474 |
init functions. |
475 |
- expand time-out strategies into a "Be smart about timeouts" section. |
476 |
- drop the "struct" from all ev_watcher declarations in the |
477 |
documentation and did other clarifications (yeah, it was a mistake |
478 |
to have a struct AND a function called ev_loop). |
479 |
- fix a bug where ev_default would not initialise the default |
480 |
loop again after it was destroyed with ev_default_destroy. |
481 |
- rename syserr to ev_syserr to avoid name clashes when embedding, |
482 |
do similar changes for event.c. |
483 |
|
484 |
3.45 Tue Oct 21 21:59:26 CEST 2008 |
485 |
- disable inotify usage on linux <2.6.25, as it is broken |
486 |
(reported by Yoann Vandoorselaere). |
487 |
- ev_stat erroneously would try to add inotify watchers |
488 |
even when inotify wasn't available (this should only |
489 |
have a performance impact). |
490 |
- ev_once now passes both timeout and io to the callback if both |
491 |
occur concurrently, instead of giving timeouts precedence. |
492 |
- disable EV_USE_INOTIFY when sys/inotify.h is too old. |
493 |
|
494 |
3.44 Mon Sep 29 05:18:39 CEST 2008 |
495 |
- embed watchers now automatically invoke ev_loop_fork on the |
496 |
embedded loop when the parent loop forks. |
497 |
- new function: ev_now_update (loop). |
498 |
- verify_watcher was not marked static. |
499 |
- improve the "associating..." manpage section. |
500 |
- documentation tweaks here and there. |
501 |
|
502 |
3.43 Sun Jul 6 05:34:41 CEST 2008 |
503 |
- include more include files on windows to get struct _stati64 |
504 |
(reported by Chris Hulbert, but doesn't quite fix his issue). |
505 |
- add missing #include <io.h> in ev.c on windows (reported by |
506 |
Matt Tolton). |
507 |
|
508 |
3.42 Tue Jun 17 12:12:07 CEST 2008 |
509 |
- work around yet another windows bug: FD_SET actually adds fd's |
510 |
multiple times to the fd_*SET*, despite official MSN docs claiming |
511 |
otherwise. Reported and well-analysed by Matt Tolton. |
512 |
- define NFDBITS to 0 when EV_SELECT_IS_WINSOCKET to make it compile |
513 |
(reported any analysed by Chris Hulbert). |
514 |
- fix a bug in ev_ebadf (this function is only used to catch |
515 |
programming errors in the libev user). reported by Matt Tolton. |
516 |
- fix a bug in fd_intern on win32 (could lead to compile errors |
517 |
under some circumstances, but would work correctly if it compiles). |
518 |
reported by Matt Tolton. |
519 |
- (try to) work around missing lstat on windows. |
520 |
- pass in the write fd set as except fd set under windows. windows |
521 |
is so uncontrollably lame that it requires this. this means that |
522 |
switching off oobinline is not supported (but tcp/ip doesn't |
523 |
have oob, so that would be stupid anyways. |
524 |
- use posix module symbol to auto-detect monotonic clock presence |
525 |
and some other default values. |
526 |
|
527 |
3.41 Fri May 23 18:42:54 CEST 2008 |
528 |
- work around an obscure bug in winsocket select: if you |
529 |
provide only empty fd sets then select returns WSAEINVAL. how sucky. |
530 |
- improve timer scheduling stability and reduce use of time_epsilon. |
531 |
- use 1-based 2-heap for EV_MINIMAL, simplifies code, reduces |
532 |
codesize and makes for better cache-efficiency. |
533 |
- use 3-based 4-heap for !EV_MINIMAL. this makes better use |
534 |
of cpu cache lines and gives better growth behaviour than |
535 |
2-based heaps. |
536 |
- cache timestamp within heap for !EV_MINIMAL, to avoid random |
537 |
memory accesses. |
538 |
- document/add EV_USE_4HEAP and EV_HEAP_CACHE_AT. |
539 |
- fix a potential aliasing issue in ev_timer_again. |
540 |
- add/document ev_periodic_at, retract direct access to ->at. |
541 |
- improve ev_stat docs. |
542 |
- add portability requirements section. |
543 |
- fix manpage headers etc. |
544 |
- normalise WSA error codes to lower range on windows. |
545 |
- add consistency check code that can be called automatically |
546 |
or on demand to check for internal structures (ev_loop_verify). |
547 |
|
548 |
3.31 Wed Apr 16 20:45:04 CEST 2008 |
549 |
- added last minute fix for ev_poll.c by Brandon Black. |
550 |
|
551 |
3.3 Wed Apr 16 19:04:10 CEST 2008 |
552 |
- event_base_loopexit should return 0 on success |
553 |
(W.C.A. Wijngaards). |
554 |
- added linux eventfd support. |
555 |
- try to autodetect epoll and inotify support |
556 |
by libc header version if not using autoconf. |
557 |
- new symbols: EV_DEFAULT_UC and EV_DEFAULT_UC_. |
558 |
- declare functions defined in ev.h as inline if |
559 |
C99 or gcc are available. |
560 |
- enable inlining with gcc versions 2 and 3. |
561 |
- work around broken poll implementations potentially |
562 |
not clearing revents field in ev_poll (Brandon Black) |
563 |
(no such systems are known at this time). |
564 |
- work around a bug in realloc on openbsd and darwin, |
565 |
also makes the erroneous valgrind complaints |
566 |
go away (noted by various people). |
567 |
- fix ev_async_pending, add c++ wrapper for ev_async |
568 |
(based on patch sent by Johannes Deisenhofer). |
569 |
- add sensible set method to ev::embed. |
570 |
- made integer constants type int in ev.h. |
571 |
|
572 |
3.2 Wed Apr 2 17:11:19 CEST 2008 |
573 |
- fix a 64 bit overflow issue in the select backend, |
574 |
by using fd_mask instead of int for the mask. |
575 |
- rename internal sighandler to avoid clash with very old perls. |
576 |
- entering ev_loop will not clear the ONESHOT or NONBLOCKING |
577 |
flags of any outer loops anymore. |
578 |
- add ev_async_pending. |
579 |
|
580 |
3.1 Thu Mar 13 13:45:22 CET 2008 |
581 |
- implement ev_async watchers. |
582 |
- only initialise signal pipe on demand. |
583 |
- make use of sig_atomic_t configurable. |
584 |
- improved documentation. |
585 |
|
586 |
3.0 Mon Jan 28 13:14:47 CET 2008 |
587 |
- API/ABI bump to version 3.0. |
588 |
- ev++.h includes "ev.h" by default now, not <ev.h>. |
589 |
- slightly improved documentation. |
590 |
- speed up signal detection after a fork. |
591 |
- only optionally return trace status changed in ev_child |
592 |
watchers. |
593 |
- experimental (and undocumented) loop wrappers for ev++.h. |
594 |
|
595 |
2.01 Tue Dec 25 08:04:41 CET 2007 |
596 |
- separate Changes file. |
597 |
- fix ev_path_set => ev_stat_set typo. |
598 |
- remove event_compat.h from the libev tarball. |
599 |
- change how include files are found. |
600 |
- doc updates. |
601 |
- update licenses, explicitly allow for GPL relicensing. |
602 |
|
603 |
2.0 Sat Dec 22 17:47:03 CET 2007 |
604 |
- new ev_sleep, ev_set_(io|timeout)_collect_interval. |
605 |
- removed epoll from embeddable fd set. |
606 |
- fix embed watchers. |
607 |
- renamed ev_embed.loop to other. |
608 |
- added exported Symbol tables. |
609 |
- undefine member wrapper macros at the end of ev.c. |
610 |
- respect EV_H in ev++.h. |
611 |
|
612 |
1.86 Tue Dec 18 02:36:57 CET 2007 |
613 |
- fix memleak on loop destroy (not relevant for perl). |
614 |
|
615 |
1.85 Fri Dec 14 20:32:40 CET 2007 |
616 |
- fix some aliasing issues w.r.t. timers and periodics |
617 |
(not relevant for perl). |
618 |
|
619 |
(for historic versions refer to EV/Changes, found in the Perl interface) |
620 |
|
621 |
0.1 Wed Oct 31 21:31:48 CET 2007 |
622 |
- original version; hacked together in <24h. |
623 |
|