[ViewVC] Annotation of: cvs/libev/Changes

Revision history for libev, a high-performance and full-featured event loop.

TODO: ABI??? API????? Changes???
TODO: win32 write() to socket for signal handling
TODO: poll asert
        - "PORTING FROM LIBEV 3.X TO 4.X" (in ev.pod) is recommended reading.
        - ev_embed_stop did not correctly stop the watcher (very good
          testcase by Vladimir Timofeev).
        - applied win32 fixes by Michael Lenaghan (also James Mansion).
        - replace EV_MINIMAL by EV_FEATURES.
        - remove dependency on sys/queue.h on freebsd (patch by Vanilla Hsu).
        - add (undocumented) EV_ENABLE when adding events with kqueue,
          this might help with OS X, which seems to need it despite documenting
          not to need it (helpfully pointed out by Tilghman Lesher).
        - do not use poll by default on freebsd, it's broken (what isn't
          on freebsd...).
        - configure now prepends -O3, not appends it, so one can still
          override it.
        - disable poll backend on AIX, the poll header spams the namespace
          and it's not worth working around dead platforms (reported
          and analyzed by Aivars Kalvans).
        - improve header file compatibility of the standalone eventfd code
          in an obscure case.
        - implement EV_AVOID_STDIO option.
        - do not use sscanf to parse linux version number (smaller, faster,
          no sscanf dependency).
        - new EV_CHILD_ENABLE and EV_SIGNAL_ENABLE configurable settings.
        - update libev.m4 HAVE_CLOCK_SYSCALL test for newer glibcs.
        - add section on accept() problems to the manpage.
        - rename EV_TIMEOUT to EV_TIMER.
        - rename ev_loop_count/depth/verify.
        - switch to two-digit minor version.

3.9  Thu Dec 31 07:59:59 CET 2009
        - signalfd is no longer used by default and has to be requested
          explicitly - this means that easy to catch bugs become hard to
          catch race conditions, but the users have spoken.
        - point out the unspecified signal mask in the documentation, and
          that this is a race condition regardless of EV_SIGNALFD.
        - backport inotify code to C89.
        - inotify file descriptors could leak into child processes.
        - ev_stat watchers could keep an errornous extra ref on the loop,
          preventing exit when unregistering all watchers (testcases
          provided by ry@tinyclouds.org).
        - implement EV_WIN32_HANDLE_TO_FD and EV_WIN32_CLOSE_FD configuration
          symbols to make it easier for apps to do their own fd management.
        - support EV_IDLE_ENABLE being disabled in ev++.h
          (patch by Didier Spezia).
        - take advantage of inotify_init1, if available, to set cloexec/nonblock
          on fd creation, to avoid races.
        - the signal handling pipe wasn't always initialised under windows
          (analysed by lekma).
        - changed minimum glibc requirement from glibc 2.9 to 2.7, for
          signalfd.
        - add missing string.h include (Denis F. Latypoff).
        - only replace ev_stat.prev when we detect an actual difference,
          so prev is (almost) always different to attr. this might
          have caused the probems with 04_stat.t.
        - add ev::timer->remaining () method to C++ API.

3.8  Sun Aug  9 14:30:45 CEST 2009
        - incompatible change: do not necessarily reset signal handler
          to SIG_DFL when a sighandler is stopped.
        - ev_default_destroy did not properly free or zero some members,
          potentially causing crashes and memory corruption on repated
          ev_default_destroy/ev_default_loop calls.
        - take advantage of signalfd on GNU/Linux systems.
        - document that the signal mask might be in an unspecified
          state when using libev's signal handling.
        - take advantage of some GNU/Linux calls to set cloexec/nonblock
          on fd creation, to avoid race conditions.

3.7  Fri Jul 17 16:36:32 CEST 2009
        - ev_unloop and ev_loop wrongly used a global variable to exit loops,
          instead of using a per-loop variable (bug caught by accident...).
        - the ev_set_io_collect_interval interpretation has changed.
        - add new functionality: ev_set_userdata, ev_userdata,
          ev_set_invoke_pending_cb, ev_set_loop_release_cb,
          ev_invoke_pending, ev_pending_count, together with a long example
          about thread locking.
        - add ev_timer_remaining (as requested by Denis F. Latypoff).
        - add ev_loop_depth.
        - calling ev_unloop in fork/prepare watchers will no longer poll
          for new events.
        - Denis F. Latypoff corrected many typos in example code snippets.
        - honor autoconf detection of EV_USE_CLOCK_SYSCALL, also double-
          check that the syscall number is available before trying to
          use it (reported by ry@tinyclouds).
        - use GetSystemTimeAsFileTime instead of _timeb on windows, for
          slightly higher accuracy.
        - properly declare ev_loop_verify and ev_now_update even when
          !EV_MULTIPLICITY.
        - do not compile in any priority code when EV_MAXPRI == EV_MINPRI.
        - support EV_MINIMAL==2 for a reduced API.
        - actually 0-initialise struct sigaction when installing signals.
        - add section on hibernate and stopped processes to ev_timer docs.

3.6  Tue Apr 28 02:49:30 CEST 2009
        - multiple timers becoming ready within an event loop iteration
          will be invoked in the "correct" order now.
        - do not leave the event loop early just because we have no active
          watchers, fixing a problem when embedding a kqueue loop
          that has active kernel events but no registered watchers
          (reported by blacksand blacksand).
        - correctly zero the idx values for arrays, so destroying and
          reinitialising the default loop actually works (patch by
          Malek Hadj-Ali).
        - implement ev_suspend and ev_resume.
        - new EV_CUSTOM revents flag for use by applications.
        - add documentation section about priorites.
        - add a glossary to the dcoumentation.
        - extend the ev_fork description slightly.
        - optimize a jump out of call_pending.

3.53 Sun Feb 15 02:38:20 CET 2009
        - fix a bug in event pipe creation on win32 that would cause a
          failed assertion on event loop creation (patch by Malek Hadj-Ali).
        - probe for CLOCK_REALTIME support at runtime as well and fall
          back to gettimeofday if there is an error, to support older
          operating systems with newer header files/libraries.
        - prefer gettimeofday over clock_gettime with USE_CLOCK_SYSCALL
          (default most everywhere), otherwise not.

3.52 Wed Jan  7 21:43:02 CET 2009
        - fix compilation of select backend in fd_set mode when NFDBITS is
          missing (to get it to compile on QNX, reported by Rodrigo Campos).
        - better select-nfds handling when select backend is in fd_set mode.
        - diagnose fd_set overruns when select backend is in fd_set mode.
        - due to a thinko, instead of disabling everything but
          select on the borked OS X platform, everything but select was
          allowed (reported by Emanuele Giaquinta).
        - actually verify that local and remote port are matching in
          libev's socketpair emulation, which makes denial-of-service
          attacks harder (but not impossible - it's windows). Make sure
          it even works under vista, which thinks that getpeer/sockname
          should return fantasy port numbers.
        - include "libev" in all assertion messages for potentially
          clearer diagnostics.
        - event_get_version (libevent compatibility) returned
          a useless string instead of the expected version string
          (patch by W.C.A. Wijngaards).

3.51 Wed Dec 24 23:00:11 CET 2008
        - fix a bug where an inotify watcher was added twice, causing
          freezes on hash collisions (reported and analysed by Graham Leggett).
        - new config symbol, EV_USE_CLOCK_SYSCALL, to make libev use
          a direct syscall - slower, but no dependency on librt et al.
        - assume negative return values != -1 signals success of port_getn
          (http://cvs.epicsol.org/cgi/viewcvs.cgi/epic5/source/newio.c?rev=1.52)
          (no known failure reports, but it doesn't hurt).
        - fork detection in ev_embed now stops and restarts the watcher
          automatically.
        - EXPERIMENTAL: default the method to operator () in ev++.h,
          to make it nicer to use functors (requested by Benedek László).
        - fixed const object callbacks in ev++.h.
        - replaced loop_ref argument of watcher.set (loop) by a direct
          ev_loop * in ev++.h, to avoid clashes with functor patch.
        - do not try to watch the empty string via inotify.
        - inotify watchers could be leaked under certain circumstances.
        - OS X 10.5 is actually even more broken than earlier versions,
          so fall back to select on that piece of garbage.
        - fixed some weirdness in the ev_embed documentation.

3.49 Wed Nov 19 11:26:53 CET 2008
        - ev_stat watchers will now use inotify as a mere hint on
          kernels <2.6.25, or if the filesystem is not in the
          "known to be good" list.
        - better mingw32 compatibility (it's not as borked as native win32)
          (analysed by Roger Pack).
        - include stdio.h in the example program, as too many people are
          confused by the weird C language otherwise. I guess the next thing
          I get told is that the "..." ellipses in the examples don't compile
          with their C compiler.

3.48 Thu Oct 30 09:02:37 CET 2008
        - further optimise away the EPOLL_CTL_ADD/MOD combo in the epoll
          backend by assuming the kernel event mask hasn't changed if
          ADD fails with EEXIST.
        - work around spurious event notification bugs in epoll by using
          a 32-bit generation counter. recreate kernel state if we receive
          spurious notifications or unwanted events. this is very costly,
          but I didn't come up with this horrible design.
        - use memset to initialise most arrays now and do away with the
          init functions.
        - expand time-out strategies into a "Be smart about timeouts" section.
        - drop the "struct" from all ev_watcher declarations in the
          documentation and did other clarifications (yeah, it was a mistake
          to have a struct AND a function called ev_loop).
        - fix a bug where ev_default would not initialise the default
          loop again after it was destroyed with ev_default_destroy.
        - rename syserr to ev_syserr to avoid name clashes when embedding,
          do similar changes for event.c.

3.45 Tue Oct 21 21:59:26 CEST 2008
        - disable inotify usage on linux <2.6.25, as it is broken
          (reported by Yoann Vandoorselaere).
        - ev_stat errornously would try to add inotify watchers
          even when inotify wasn't available (this should only
          have a performance impact).
        - ev_once now passes both timeout and io to the callback if both
          occur concurrently, instead of giving timeouts precedence.
        - disable EV_USE_INOTIFY when sys/inotify.h is too old.

3.44 Mon Sep 29 05:18:39 CEST 2008
        - embed watchers now automatically invoke ev_loop_fork on the
          embedded loop when the parent loop forks.
        - new function: ev_now_update (loop).
        - verify_watcher was not marked static.
        - improve the "associating..." manpage section.
        - documentation tweaks here and there.

3.43 Sun Jul  6 05:34:41 CEST 2008
        - include more include files on windows to get struct _stati64
          (reported by Chris Hulbert, but doesn't quite fix his issue).
        - add missing #include <io.h> in ev.c on windows (reported by
          Matt Tolton).

3.42 Tue Jun 17 12:12:07 CEST 2008
        - work around yet another windows bug: FD_SET actually adds fd's
          multiple times to the fd_*SET*, despite official MSN docs claiming
          otherwise. Reported and well-analysed by Matt Tolton.
        - define NFDBITS to 0 when EV_SELECT_IS_WINSOCKET to make it compile
          (reported any analysed by Chris Hulbert).
        - fix a bug in ev_ebadf (this function is only used to catch
          programming errors in the libev user). reported by Matt Tolton.
        - fix a bug in fd_intern on win32 (could lead to compile errors
          under some circumstances, but would work correctly if it compiles).
          reported by Matt Tolton.
        - (try to) work around missing lstat on windows.
        - pass in the write fd set as except fd set under windows. windows
          is so uncontrollably lame that it requires this. this means that
          switching off oobinline is not supported (but tcp/ip doesn't
          have oob, so that would be stupid anyways.
        - use posix module symbol to auto-detect monotonic clock presence
          and some other default values.

3.41 Fri May 23 18:42:54 CEST 2008
        - work around an obscure bug in winsocket select: if you
          provide only empty fd sets then select returns WSAEINVAL. how sucky.
        - improve timer scheduling stability and reduce use of time_epsilon.
        - use 1-based 2-heap for EV_MINIMAL, simplifies code, reduces
          codesize and makes for better cache-efficiency.
        - use 3-based 4-heap for !EV_MINIMAL. this makes better use
          of cpu cache lines and gives better growth behaviour than
          2-based heaps.
        - cache timestamp within heap for !EV_MINIMAL, to avoid random
          memory accesses.
        - document/add EV_USE_4HEAP and EV_HEAP_CACHE_AT.
        - fix a potential aliasing issue in ev_timer_again.
        - add/document ev_periodic_at, retract direct access to ->at.
        - improve ev_stat docs.
        - add portability requirements section.
        - fix manpage headers etc.
        - normalise WSA error codes to lower range on windows.
        - add consistency check code that can be called automatically
          or on demand to check for internal structures (ev_loop_verify).

3.31 Wed Apr 16 20:45:04 CEST 2008
        - added last minute fix for ev_poll.c by Brandon Black.

3.3  Wed Apr 16 19:04:10 CEST 2008
        - event_base_loopexit should return 0 on success
          (W.C.A. Wijngaards).
        - added linux eventfd support.
        - try to autodetect epoll and inotify support
          by libc header version if not using autoconf.
        - new symbols: EV_DEFAULT_UC and EV_DEFAULT_UC_.
        - declare functions defined in ev.h as inline if
          C99 or gcc are available.
        - enable inlining with gcc versions 2 and 3.
        - work around broken poll implementations potentially
          not clearing revents field in ev_poll (Brandon Black)
          (no such systems are known at this time).
        - work around a bug in realloc on openbsd and darwin,
          also makes the errornous valgrind complaints
          go away (noted by various people).
        - fix ev_async_pending, add c++ wrapper for ev_async
          (based on patch sent by Johannes Deisenhofer.
        - add sensible set method to ev::embed.
        - made integer constants type int in ev.h.

3.2  Wed Apr  2 17:11:19 CEST 2008
        - fix a 64 bit overflow issue in the select backend,
          by using fd_mask instead of int for the mask.
        - rename internal sighandler to avoid clash with very old perls.
        - entering ev_loop will not clear the ONESHOT or NONBLOCKING
          flags of any outer loops anymore.
        - add ev_async_pending.

3.1  Thu Mar 13 13:45:22 CET 2008
        - implement ev_async watchers.
        - only initialise signal pipe on demand.
        - make use of sig_atomic_t configurable.
        - improved documentation.

3.0  Mon Jan 28 13:14:47 CET 2008
        - API/ABI bump to version 3.0.
        - ev++.h includes "ev.h" by default now, not <ev.h>.
        - slightly improved documentation.
        - speed up signal detection after a fork.
        - only optionally return trace status changed in ev_child
          watchers.
        - experimental (and undocumented) loop wrappers for ev++.h.

2.01 Tue Dec 25 08:04:41 CET 2007
        - separate Changes file.
        - fix ev_path_set => ev_stat_set typo.
        - remove event_compat.h from the libev tarball.
        - change how include files are found.
        - doc updates.
        - update licenses, explicitly allow for GPL relicensing.

2.0  Sat Dec 22 17:47:03 CET 2007
        - new ev_sleep, ev_set_(io|timeout)_collect_interval.
        - removed epoll from embeddable fd set.
        - fix embed watchers.
        - renamed ev_embed.loop to other.
        - added exported Symbol tables.
        - undefine member wrapper macros at the end of ev.c.
        - respect EV_H in ev++.h.

1.86 Tue Dec 18 02:36:57 CET 2007
        - fix memleak on loop destroy (not relevant for perl).

1.85 Fri Dec 14 20:32:40 CET 2007
        - fix some aliasing issues w.r.t. timers and periodics
          (not relevant for perl).

(for historic versions refer to EV/Changes, found in the Perl interface)

0.1  Wed Oct 31 21:31:48 CET 2007
        - original version; hacked together in <24h.

Revision:	1.176
Committed:	Mon Mar 29 12:40:57 2010 UTC (14 years, 1 month ago) by root
Branch:	MAIN
Changes since 1.175:	+3 -0 lines
Log Message:	* empty log message *
#	User	Rev	Content
1	root	1.14	Revision history for libev, a high-performance and full-featured event loop.
2	root	1.1
3	root	1.166	TODO: ABI??? API????? Changes???
4	root	1.168	TODO: win32 write() to socket for signal handling
5	root	1.176	TODO: poll asert
6	root	1.171	- "PORTING FROM LIBEV 3.X TO 4.X" (in ev.pod) is recommended reading.
7	root	1.161	- ev_embed_stop did not correctly stop the watcher (very good
8			testcase by Vladimir Timofeev).
9	root	1.164	- applied win32 fixes by Michael Lenaghan (also James Mansion).
10	root	1.166	- replace EV_MINIMAL by EV_FEATURES.
11	root	1.158	- remove dependency on sys/queue.h on freebsd (patch by Vanilla Hsu).
12	root	1.174	- add (undocumented) EV_ENABLE when adding events with kqueue,
13	root	1.175	this might help with OS X, which seems to need it despite documenting
14			not to need it (helpfully pointed out by Tilghman Lesher).
15	root	1.176	- do not use poll by default on freebsd, it's broken (what isn't
16			on freebsd...).
17	root	1.159	- configure now prepends -O3, not appends it, so one can still
18			override it.
19	root	1.160	- disable poll backend on AIX, the poll header spams the namespace
20			and it's not worth working around dead platforms (reported
21			and analyzed by Aivars Kalvans).
22	root	1.162	- improve header file compatibility of the standalone eventfd code
23			in an obscure case.
24	root	1.163	- implement EV_AVOID_STDIO option.
25			- do not use sscanf to parse linux version number (smaller, faster,
26			no sscanf dependency).
27	root	1.164	- new EV_CHILD_ENABLE and EV_SIGNAL_ENABLE configurable settings.
28	sf-exg	1.172	- update libev.m4 HAVE_CLOCK_SYSCALL test for newer glibcs.
29	root	1.165	- add section on accept() problems to the manpage.
30	root	1.168	- rename EV_TIMEOUT to EV_TIMER.
31	root	1.173	- rename ev_loop_count/depth/verify.
32	root	1.168	- switch to two-digit minor version.
33	root	1.157
34	root	1.156	3.9 Thu Dec 31 07:59:59 CET 2009
35	root	1.155	- signalfd is no longer used by default and has to be requested
36			explicitly - this means that easy to catch bugs become hard to
37			catch race conditions, but the users have spoken.
38			- point out the unspecified signal mask in the documentation, and
39			that this is a race condition regardless of EV_SIGNALFD.
40	root	1.142	- backport inotify code to C89.
41	root	1.148	- inotify file descriptors could leak into child processes.
42	root	1.151	- ev_stat watchers could keep an errornous extra ref on the loop,
43			preventing exit when unregistering all watchers (testcases
44	root	1.150	provided by ry@tinyclouds.org).
45	root	1.143	- implement EV_WIN32_HANDLE_TO_FD and EV_WIN32_CLOSE_FD configuration
46	root	1.145	symbols to make it easier for apps to do their own fd management.
47	root	1.144	- support EV_IDLE_ENABLE being disabled in ev++.h
48			(patch by Didier Spezia).
49	root	1.147	- take advantage of inotify_init1, if available, to set cloexec/nonblock
50			on fd creation, to avoid races.
51	root	1.150	- the signal handling pipe wasn't always initialised under windows
52			(analysed by lekma).
53	root	1.146	- changed minimum glibc requirement from glibc 2.9 to 2.7, for
54			signalfd.
55	root	1.152	- add missing string.h include (Denis F. Latypoff).
56	root	1.153	- only replace ev_stat.prev when we detect an actual difference,
57			so prev is (almost) always different to attr. this might
58			have caused the probems with 04_stat.t.
59	root	1.154	- add ev::timer->remaining () method to C++ API.
60	root	1.142
61	root	1.141	3.8 Sun Aug 9 14:30:45 CEST 2009
62	root	1.133	- incompatible change: do not necessarily reset signal handler
63			to SIG_DFL when a sighandler is stopped.
64	root	1.135	- ev_default_destroy did not properly free or zero some members,
65	root	1.134	potentially causing crashes and memory corruption on repated
66			ev_default_destroy/ev_default_loop calls.
67	root	1.133	- take advantage of signalfd on GNU/Linux systems.
68			- document that the signal mask might be in an unspecified
69			state when using libev's signal handling.
70	root	1.136	- take advantage of some GNU/Linux calls to set cloexec/nonblock
71			on fd creation, to avoid race conditions.
72	root	1.132
73	root	1.131	3.7 Fri Jul 17 16:36:32 CEST 2009
74	root	1.128	- ev_unloop and ev_loop wrongly used a global variable to exit loops,
75			instead of using a per-loop variable (bug caught by accident...).
76			- the ev_set_io_collect_interval interpretation has changed.
77			- add new functionality: ev_set_userdata, ev_userdata,
78	root	1.126	ev_set_invoke_pending_cb, ev_set_loop_release_cb,
79	root	1.132	ev_invoke_pending, ev_pending_count, together with a long example
80			about thread locking.
81	root	1.128	- add ev_timer_remaining (as requested by Denis F. Latypoff).
82			- add ev_loop_depth.
83	root	1.124	- calling ev_unloop in fork/prepare watchers will no longer poll
84			for new events.
85	root	1.115	- Denis F. Latypoff corrected many typos in example code snippets.
86			- honor autoconf detection of EV_USE_CLOCK_SYSCALL, also double-
87			check that the syscall number is available before trying to
88			use it (reported by ry@tinyclouds).
89	root	1.117	- use GetSystemTimeAsFileTime instead of _timeb on windows, for
90	root	1.116	slightly higher accuracy.
91	root	1.122	- properly declare ev_loop_verify and ev_now_update even when
92			!EV_MULTIPLICITY.
93	root	1.121	- do not compile in any priority code when EV_MAXPRI == EV_MINPRI.
94	root	1.123	- support EV_MINIMAL==2 for a reduced API.
95	root	1.125	- actually 0-initialise struct sigaction when installing signals.
96	root	1.127	- add section on hibernate and stopped processes to ev_timer docs.
97	root	1.115
98	root	1.113	3.6 Tue Apr 28 02:49:30 CEST 2009
99	root	1.107	- multiple timers becoming ready within an event loop iteration
100			will be invoked in the "correct" order now.
101	root	1.105	- do not leave the event loop early just because we have no active
102			watchers, fixing a problem when embedding a kqueue loop
103			that has active kernel events but no registered watchers
104			(reported by blacksand blacksand).
105	root	1.102	- correctly zero the idx values for arrays, so destroying and
106			reinitialising the default loop actually works (patch by
107			Malek Hadj-Ali).
108	root	1.108	- implement ev_suspend and ev_resume.
109	root	1.106	- new EV_CUSTOM revents flag for use by applications.
110	root	1.109	- add documentation section about priorites.
111			- add a glossary to the dcoumentation.
112	root	1.111	- extend the ev_fork description slightly.
113	root	1.112	- optimize a jump out of call_pending.
114	root	1.102
115	root	1.101	3.53 Sun Feb 15 02:38:20 CET 2009
116			- fix a bug in event pipe creation on win32 that would cause a
117			failed assertion on event loop creation (patch by Malek Hadj-Ali).
118	root	1.100	- probe for CLOCK_REALTIME support at runtime as well and fall
119			back to gettimeofday if there is an error, to support older
120			operating systems with newer header files/libraries.
121			- prefer gettimeofday over clock_gettime with USE_CLOCK_SYSCALL
122			(default most everywhere), otherwise not.
123
124	root	1.99	3.52 Wed Jan 7 21:43:02 CET 2009
125	root	1.97	- fix compilation of select backend in fd_set mode when NFDBITS is
126			missing (to get it to compile on QNX, reported by Rodrigo Campos).
127			- better select-nfds handling when select backend is in fd_set mode.
128			- diagnose fd_set overruns when select backend is in fd_set mode.
129			- due to a thinko, instead of disabling everything but
130			select on the borked OS X platform, everything but select was
131			allowed (reported by Emanuele Giaquinta).
132	root	1.95	- actually verify that local and remote port are matching in
133			libev's socketpair emulation, which makes denial-of-service
134			attacks harder (but not impossible - it's windows). Make sure
135	root	1.96	it even works under vista, which thinks that getpeer/sockname
136			should return fantasy port numbers.
137	root	1.100	- include "libev" in all assertion messages for potentially
138	root	1.97	clearer diagnostics.
139	root	1.98	- event_get_version (libevent compatibility) returned
140			a useless string instead of the expected version string
141			(patch by W.C.A. Wijngaards).
142	root	1.95
143	root	1.94	3.51 Wed Dec 24 23:00:11 CET 2008
144	root	1.91	- fix a bug where an inotify watcher was added twice, causing
145	root	1.92	freezes on hash collisions (reported and analysed by Graham Leggett).
146	root	1.87	- new config symbol, EV_USE_CLOCK_SYSCALL, to make libev use
147			a direct syscall - slower, but no dependency on librt et al.
148	root	1.88	- assume negative return values != -1 signals success of port_getn
149			(http://cvs.epicsol.org/cgi/viewcvs.cgi/epic5/source/newio.c?rev=1.52)
150			(no known failure reports, but it doesn't hurt).
151	root	1.93	- fork detection in ev_embed now stops and restarts the watcher
152			automatically.
153	root	1.89	- EXPERIMENTAL: default the method to operator () in ev++.h,
154			to make it nicer to use functors (requested by Benedek László).
155			- fixed const object callbacks in ev++.h.
156	root	1.90	- replaced loop_ref argument of watcher.set (loop) by a direct
157			ev_loop * in ev++.h, to avoid clashes with functor patch.
158	root	1.91	- do not try to watch the empty string via inotify.
159	root	1.92	- inotify watchers could be leaked under certain circumstances.
160			- OS X 10.5 is actually even more broken than earlier versions,
161			so fall back to select on that piece of garbage.
162	root	1.93	- fixed some weirdness in the ev_embed documentation.
163	root	1.87
164	root	1.86	3.49 Wed Nov 19 11:26:53 CET 2008
165	root	1.83	- ev_stat watchers will now use inotify as a mere hint on
166			kernels <2.6.25, or if the filesystem is not in the
167			"known to be good" list.
168	root	1.84	- better mingw32 compatibility (it's not as borked as native win32)
169			(analysed by Roger Pack).
170	root	1.85	- include stdio.h in the example program, as too many people are
171			confused by the weird C language otherwise. I guess the next thing
172			I get told is that the "..." ellipses in the examples don't compile
173			with their C compiler.
174	root	1.83
175	root	1.80	3.48 Thu Oct 30 09:02:37 CET 2008
176	root	1.71	- further optimise away the EPOLL_CTL_ADD/MOD combo in the epoll
177			backend by assuming the kernel event mask hasn't changed if
178			ADD fails with EEXIST.
179	root	1.75	- work around spurious event notification bugs in epoll by using
180	root	1.77	a 32-bit generation counter. recreate kernel state if we receive
181	root	1.78	spurious notifications or unwanted events. this is very costly,
182			but I didn't come up with this horrible design.
183	root	1.71	- use memset to initialise most arrays now and do away with the
184			init functions.
185	root	1.72	- expand time-out strategies into a "Be smart about timeouts" section.
186			- drop the "struct" from all ev_watcher declarations in the
187	root	1.73	documentation and did other clarifications (yeah, it was a mistake
188			to have a struct AND a function called ev_loop).
189	root	1.74	- fix a bug where ev_default would not initialise the default
190			loop again after it was destroyed with ev_default_destroy.
191	root	1.79	- rename syserr to ev_syserr to avoid name clashes when embedding,
192			do similar changes for event.c.
193	root	1.71
194	root	1.69	3.45 Tue Oct 21 21:59:26 CEST 2008
195	root	1.67	- disable inotify usage on linux <2.6.25, as it is broken
196			(reported by Yoann Vandoorselaere).
197			- ev_stat errornously would try to add inotify watchers
198	root	1.68	even when inotify wasn't available (this should only
199	root	1.67	have a performance impact).
200	root	1.65	- ev_once now passes both timeout and io to the callback if both
201			occur concurrently, instead of giving timeouts precedence.
202	root	1.66	- disable EV_USE_INOTIFY when sys/inotify.h is too old.
203	root	1.65
204	root	1.62	3.44 Mon Sep 29 05:18:39 CEST 2008
205			- embed watchers now automatically invoke ev_loop_fork on the
206			embedded loop when the parent loop forks.
207	root	1.58	- new function: ev_now_update (loop).
208	root	1.57	- verify_watcher was not marked static.
209	root	1.59	- improve the "associating..." manpage section.
210	root	1.60	- documentation tweaks here and there.
211	root	1.57
212	root	1.56	3.43 Sun Jul 6 05:34:41 CEST 2008
213	root	1.54	- include more include files on windows to get struct _stati64
214	root	1.56	(reported by Chris Hulbert, but doesn't quite fix his issue).
215	root	1.53	- add missing #include <io.h> in ev.c on windows (reported by
216			Matt Tolton).
217
218	root	1.52	3.42 Tue Jun 17 12:12:07 CEST 2008
219	root	1.50	- work around yet another windows bug: FD_SET actually adds fd's
220			multiple times to the fd_SET, despite official MSN docs claiming
221	root	1.51	otherwise. Reported and well-analysed by Matt Tolton.
222			- define NFDBITS to 0 when EV_SELECT_IS_WINSOCKET to make it compile
223	root	1.48	(reported any analysed by Chris Hulbert).
224	root	1.49	- fix a bug in ev_ebadf (this function is only used to catch
225			programming errors in the libev user). reported by Matt Tolton.
226			- fix a bug in fd_intern on win32 (could lead to compile errors
227			under some circumstances, but would work correctly if it compiles).
228			reported by Matt Tolton.
229	root	1.50	- (try to) work around missing lstat on windows.
230	root	1.45	- pass in the write fd set as except fd set under windows. windows
231	root	1.46	is so uncontrollably lame that it requires this. this means that
232			switching off oobinline is not supported (but tcp/ip doesn't
233			have oob, so that would be stupid anyways.
234	root	1.47	- use posix module symbol to auto-detect monotonic clock presence
235			and some other default values.
236	root	1.45
237	root	1.44	3.41 Fri May 23 18:42:54 CEST 2008
238	root	1.46	- work around an obscure bug in winsocket select: if you
239	root	1.40	provide only empty fd sets then select returns WSAEINVAL. how sucky.
240	root	1.41	- improve timer scheduling stability and reduce use of time_epsilon.
241	root	1.36	- use 1-based 2-heap for EV_MINIMAL, simplifies code, reduces
242			codesize and makes for better cache-efficiency.
243			- use 3-based 4-heap for !EV_MINIMAL. this makes better use
244			of cpu cache lines and gives better growth behaviour than
245			2-based heaps.
246	root	1.38	- cache timestamp within heap for !EV_MINIMAL, to avoid random
247			memory accesses.
248			- document/add EV_USE_4HEAP and EV_HEAP_CACHE_AT.
249	root	1.33	- fix a potential aliasing issue in ev_timer_again.
250	root	1.35	- add/document ev_periodic_at, retract direct access to ->at.
251			- improve ev_stat docs.
252			- add portability requirements section.
253	root	1.39	- fix manpage headers etc.
254	root	1.40	- normalise WSA error codes to lower range on windows.
255	root	1.43	- add consistency check code that can be called automatically
256			or on demand to check for internal structures (ev_loop_verify).
257	root	1.31
258	root	1.30	3.31 Wed Apr 16 20:45:04 CEST 2008
259			- added last minute fix for ev_poll.c by Brandon Black.
260	root	1.29
261	root	1.28	3.3 Wed Apr 16 19:04:10 CEST 2008
262	root	1.23	- event_base_loopexit should return 0 on success
263			(W.C.A. Wijngaards).
264	root	1.20	- added linux eventfd support.
265			- try to autodetect epoll and inotify support
266			by libc header version if not using autoconf.
267	root	1.21	- new symbols: EV_DEFAULT_UC and EV_DEFAULT_UC_.
268			- declare functions defined in ev.h as inline if
269			C99 or gcc are available.
270			- enable inlining with gcc versions 2 and 3.
271	root	1.25	- work around broken poll implementations potentially
272			not clearing revents field in ev_poll (Brandon Black)
273			(no such systems are known at this time).
274	root	1.23	- work around a bug in realloc on openbsd and darwin,
275			also makes the errornous valgrind complaints
276			go away (noted by various people).
277	root	1.26	- fix ev_async_pending, add c++ wrapper for ev_async
278			(based on patch sent by Johannes Deisenhofer.
279			- add sensible set method to ev::embed.
280	root	1.27	- made integer constants type int in ev.h.
281	root	1.20
282	root	1.19	3.2 Wed Apr 2 17:11:19 CEST 2008
283	root	1.15	- fix a 64 bit overflow issue in the select backend,
284			by using fd_mask instead of int for the mask.
285	root	1.16	- rename internal sighandler to avoid clash with very old perls.
286	root	1.18	- entering ev_loop will not clear the ONESHOT or NONBLOCKING
287			flags of any outer loops anymore.
288	root	1.17	- add ev_async_pending.
289	root	1.15
290	root	1.14	3.1 Thu Mar 13 13:45:22 CET 2008
291	root	1.12	- implement ev_async watchers.
292	root	1.10	- only initialise signal pipe on demand.
293			- make use of sig_atomic_t configurable.
294	root	1.14	- improved documentation.
295	root	1.10
296	root	1.9	3.0 Mon Jan 28 13:14:47 CET 2008
297	root	1.6	- API/ABI bump to version 3.0.
298	root	1.4	- ev++.h includes "ev.h" by default now, not <ev.h>.
299			- slightly improved documentation.
300	root	1.5	- speed up signal detection after a fork.
301	root	1.6	- only optionally return trace status changed in ev_child
302			watchers.
303	root	1.8	- experimental (and undocumented) loop wrappers for ev++.h.
304	root	1.4
305	root	1.2	2.01 Tue Dec 25 08:04:41 CET 2007
306	root	1.3	- separate Changes file.
307	root	1.2	- fix ev_path_set => ev_stat_set typo.
308			- remove event_compat.h from the libev tarball.
309			- change how include files are found.
310			- doc updates.
311			- update licenses, explicitly allow for GPL relicensing.
312
313			2.0 Sat Dec 22 17:47:03 CET 2007
314			- new ev_sleep, ev_set_(io\|timeout)_collect_interval.
315			- removed epoll from embeddable fd set.
316			- fix embed watchers.
317			- renamed ev_embed.loop to other.
318			- added exported Symbol tables.
319			- undefine member wrapper macros at the end of ev.c.
320			- respect EV_H in ev++.h.
321
322			1.86 Tue Dec 18 02:36:57 CET 2007
323			- fix memleak on loop destroy (not relevant for perl).
324
325			1.85 Fri Dec 14 20:32:40 CET 2007
326			- fix some aliasing issues w.r.t. timers and periodics
327			(not relevant for perl).
328
329			(for historic versions refer to EV/Changes, found in the Perl interface)
330
331	root	1.1	0.1 Wed Oct 31 21:31:48 CET 2007
332			- original version; hacked together in <24h.
333