1 | Revision history for Perl extension EV |
1 | Revision history for Perl extension EV |
2 | |
2 | |
|
|
3 | - (libev) probe for CLOCK_REALTIME support at runtime as well and |
|
|
4 | fall back to gettimeofday if there is an error, to support older |
|
|
5 | operating systems with newer header files/libraries. |
|
|
6 | - prefer gettimeofday over clock_gettime by default. |
|
|
7 | |
|
|
8 | 3.52 Wed Jan 7 21:46:14 CET 2009 |
|
|
9 | - (libev) fix some issues in the select backend when in fd_set mode. |
|
|
10 | - (libev) due to a thinko, instead of disabling everything but |
|
|
11 | select on the borked OS X platform, everything but select was |
|
|
12 | allowed (reported by Emanuele Giaquinta). |
|
|
13 | - (libev) actually verify that local and remote port are matching in |
|
|
14 | libev's socketpair emulation, which makes denial-of-service |
|
|
15 | attacks harder (but not impossible - it's windows). Make sure |
|
|
16 | it even works under vista, which thinks that getpeer/sockname |
|
|
17 | should return fantasy port numbers. |
|
|
18 | |
|
|
19 | 3.51 Wed Dec 24 23:01:59 CET 2008 |
|
|
20 | - do not cache the arguments passed to callbacks if the refcount |
|
|
21 | indicates that the callback has stolen them. |
|
|
22 | - (libev) try to avoid librt on GNU/Linux. |
|
|
23 | - (libev) check port_getn return value dfferently, might potentially |
|
|
24 | avoid problems. |
|
|
25 | - (libev) fix a bug with stat watchers possibly causing freezes. |
|
|
26 | - (libev) work around OS X 10.5 breaking poll, now select is the |
|
|
27 | only viable choice left on that pile of garbage. |
|
|
28 | - play tester whore: disable some tests that typically fail only |
|
|
29 | on cpan tester machines. |
|
|
30 | |
|
|
31 | 3.49 Wed Nov 19 11:26:53 CET 2008 |
|
|
32 | - fix typos in manpage (Alex Efros). |
|
|
33 | - increase timing even further, for the ever-overloaded cpan-tester |
|
|
34 | machines, and to the detriment of everybody else who wants a fast |
|
|
35 | make test. |
|
|
36 | - possible 5.6 compatibility. |
|
|
37 | - (libev) use inotify even on buggy kernels, but don't rely on it. |
|
|
38 | - (libev) use inotify only as an added hint on network filesystems. |
|
|
39 | |
|
|
40 | 3.48 Thu Oct 30 09:09:48 CET 2008 |
|
|
41 | - (libev) use a generation counter to detect spurious epoll events |
|
|
42 | and recreate the kernel in such a case (sorry, it's slow, but |
|
|
43 | I didn't design epoll...). |
|
|
44 | - (libev) optimise away an EPOLL_CTL_ADD/MOD combo in the epoll |
|
|
45 | backend in some cases. |
|
|
46 | - (libev) use memset to initialise most arrays now and do away |
|
|
47 | with the init functions. |
|
|
48 | - speed up event callback invocation further (~15%), by |
|
|
49 | not creating the object reference each time. |
|
|
50 | - EV::sleep was documented, but not implemented. fun. |
|
|
51 | |
|
|
52 | 3.45 Tue Oct 21 22:20:39 CEST 2008 |
|
|
53 | - (libev) disable inotify usage on linux <2.6.25 (kernel bug). |
3 | - (libev) ev_embed will now automatically follow fork. |
54 | - (libev) ev_embed will now automatically follow fork. |
4 | - (libev) ev_once will now pass both io and timeout events to |
55 | - (libev) ev_once will now pass both io and timeout events to |
5 | the callback when both happen concurrently, instead of giving |
56 | the callback when both happen concurrently, instead of giving |
6 | one precedence. |
57 | one precedence. |
|
|
58 | - install C library documentation as EV::libev manpage (sorry, oesi). |
|
|
59 | - (libev) fix a minor performance bug in ev_stat handling. |
7 | |
60 | |
8 | 3.44 Mon Sep 29 05:16:31 CEST 2008 |
61 | 3.44 Mon Sep 29 05:16:31 CEST 2008 |
9 | - (libev) add EV::now_update and $loop->now_update. |
62 | - (libev) add EV::now_update and $loop->now_update. |
10 | |
63 | |
11 | 3.431 Sun Jul 13 00:19:02 CEST 2008 |
64 | 3.431 Sun Jul 13 00:19:02 CEST 2008 |