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

Comparing libev/Changes (file contents):
Revision 1.46 by root, Mon May 26 04:14:32 2008 UTC vs.
Revision 1.92 by root, Sun Dec 14 13:03:54 2008 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
3 - fix a bug where an inotify watcher was added twice, causing
4 freezes on hash collisions (reported and analysed by Graham Leggett).
5 - new config symbol, EV_USE_CLOCK_SYSCALL, to make libev use
6 a direct syscall - slower, but no dependency on librt et al.
7 - assume negative return values != -1 signals success of port_getn
8 (http://cvs.epicsol.org/cgi/viewcvs.cgi/epic5/source/newio.c?rev=1.52)
9 (no known failure reports, but it doesn't hurt).
10 - EXPERIMENTAL: default the method to operator () in ev++.h,
11 to make it nicer to use functors (requested by Benedek László).
12 - fixed const object callbacks in ev++.h.
13 - replaced loop_ref argument of watcher.set (loop) by a direct
14 ev_loop * in ev++.h, to avoid clashes with functor patch.
15 - do not try to watch the empty string via inotify.
16 - inotify watchers could be leaked under certain circumstances.
17 - OS X 10.5 is actually even more broken than earlier versions,
18 so fall back to select on that piece of garbage.
19
203.49 Wed Nov 19 11:26:53 CET 2008
21 - ev_stat watchers will now use inotify as a mere hint on
22 kernels <2.6.25, or if the filesystem is not in the
23 "known to be good" list.
24 - better mingw32 compatibility (it's not as borked as native win32)
25 (analysed by Roger Pack).
26 - include stdio.h in the example program, as too many people are
27 confused by the weird C language otherwise. I guess the next thing
28 I get told is that the "..." ellipses in the examples don't compile
29 with their C compiler.
30
313.48 Thu Oct 30 09:02:37 CET 2008
32 - further optimise away the EPOLL_CTL_ADD/MOD combo in the epoll
33 backend by assuming the kernel event mask hasn't changed if
34 ADD fails with EEXIST.
35 - work around spurious event notification bugs in epoll by using
36 a 32-bit generation counter. recreate kernel state if we receive
37 spurious notifications or unwanted events. this is very costly,
38 but I didn't come up with this horrible design.
39 - use memset to initialise most arrays now and do away with the
40 init functions.
41 - expand time-out strategies into a "Be smart about timeouts" section.
42 - drop the "struct" from all ev_watcher declarations in the
43 documentation and did other clarifications (yeah, it was a mistake
44 to have a struct AND a function called ev_loop).
45 - fix a bug where ev_default would not initialise the default
46 loop again after it was destroyed with ev_default_destroy.
47 - rename syserr to ev_syserr to avoid name clashes when embedding,
48 do similar changes for event.c.
49
503.45 Tue Oct 21 21:59:26 CEST 2008
51 - disable inotify usage on linux <2.6.25, as it is broken
52 (reported by Yoann Vandoorselaere).
53 - ev_stat errornously would try to add inotify watchers
54 even when inotify wasn't available (this should only
55 have a performance impact).
56 - ev_once now passes both timeout and io to the callback if both
57 occur concurrently, instead of giving timeouts precedence.
58 - disable EV_USE_INOTIFY when sys/inotify.h is too old.
59
603.44 Mon Sep 29 05:18:39 CEST 2008
61 - embed watchers now automatically invoke ev_loop_fork on the
62 embedded loop when the parent loop forks.
63 - new function: ev_now_update (loop).
64 - verify_watcher was not marked static.
65 - improve the "associating..." manpage section.
66 - documentation tweaks here and there.
67
683.43 Sun Jul 6 05:34:41 CEST 2008
69 - include more include files on windows to get struct _stati64
70 (reported by Chris Hulbert, but doesn't quite fix his issue).
71 - add missing #include <io.h> in ev.c on windows (reported by
72 Matt Tolton).
73
743.42 Tue Jun 17 12:12:07 CEST 2008
75 - work around yet another windows bug: FD_SET actually adds fd's
76 multiple times to the fd_*SET*, despite official MSN docs claiming
77 otherwise. Reported and well-analysed by Matt Tolton.
78 - define NFDBITS to 0 when EV_SELECT_IS_WINSOCKET to make it compile
79 (reported any analysed by Chris Hulbert).
80 - fix a bug in ev_ebadf (this function is only used to catch
81 programming errors in the libev user). reported by Matt Tolton.
82 - fix a bug in fd_intern on win32 (could lead to compile errors
83 under some circumstances, but would work correctly if it compiles).
84 reported by Matt Tolton.
85 - (try to) work around missing lstat on windows.
3 - pass in the write fd set as except fd set under windows. windows 86 - pass in the write fd set as except fd set under windows. windows
4 is so uncontrollably lame that it requires this. this means that 87 is so uncontrollably lame that it requires this. this means that
5 switching off oobinline is not supported (but tcp/ip doesn't 88 switching off oobinline is not supported (but tcp/ip doesn't
6 have oob, so that would be stupid anyways. 89 have oob, so that would be stupid anyways.
90 - use posix module symbol to auto-detect monotonic clock presence
91 and some other default values.
7 92
83.41 Fri May 23 18:42:54 CEST 2008 933.41 Fri May 23 18:42:54 CEST 2008
9 - work around an obscure bug in winsocket select: if you 94 - work around an obscure bug in winsocket select: if you
10 provide only empty fd sets then select returns WSAEINVAL. how sucky. 95 provide only empty fd sets then select returns WSAEINVAL. how sucky.
11 - improve timer scheduling stability and reduce use of time_epsilon. 96 - improve timer scheduling stability and reduce use of time_epsilon.

Diff Legend

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