| 1 | Revision history for Perl extension EV |
1 | Revision history for Perl extension EV |
| 2 | |
2 | |
|
|
3 | 3.53 Sun Feb 15 02:38:20 CET 2009 |
|
|
4 | - (libev) on win32, the event loop creation could randomly fail |
|
|
5 | due to an initialised variable having the wrong value. |
|
|
6 | - (libev) probe for CLOCK_REALTIME support at runtime as well and |
|
|
7 | fall back to gettimeofday if there is an error, to support older |
|
|
8 | operating systems with newer header files/libraries. |
|
|
9 | - prefer gettimeofday over clock_gettime by default. |
|
|
10 | |
|
|
11 | 3.52 Wed Jan 7 21:46:14 CET 2009 |
|
|
12 | - (libev) fix some issues in the select backend when in fd_set mode. |
|
|
13 | - (libev) due to a thinko, instead of disabling everything but |
|
|
14 | select on the borked OS X platform, everything but select was |
|
|
15 | allowed (reported by Emanuele Giaquinta). |
|
|
16 | - (libev) actually verify that local and remote port are matching in |
|
|
17 | libev's socketpair emulation, which makes denial-of-service |
|
|
18 | attacks harder (but not impossible - it's windows). Make sure |
|
|
19 | it even works under vista, which thinks that getpeer/sockname |
|
|
20 | should return fantasy port numbers. |
|
|
21 | |
|
|
22 | 3.51 Wed Dec 24 23:01:59 CET 2008 |
|
|
23 | - do not cache the arguments passed to callbacks if the refcount |
|
|
24 | indicates that the callback has stolen them. |
|
|
25 | - (libev) try to avoid librt on GNU/Linux. |
|
|
26 | - (libev) check port_getn return value dfferently, might potentially |
|
|
27 | avoid problems. |
|
|
28 | - (libev) fix a bug with stat watchers possibly causing freezes. |
|
|
29 | - (libev) work around OS X 10.5 breaking poll, now select is the |
|
|
30 | only viable choice left on that pile of garbage. |
|
|
31 | - play tester whore: disable some tests that typically fail only |
|
|
32 | on cpan tester machines. |
|
|
33 | |
|
|
34 | 3.49 Wed Nov 19 11:26:53 CET 2008 |
|
|
35 | - fix typos in manpage (Alex Efros). |
|
|
36 | - increase timing even further, for the ever-overloaded cpan-tester |
|
|
37 | machines, and to the detriment of everybody else who wants a fast |
|
|
38 | make test. |
|
|
39 | - possible 5.6 compatibility. |
|
|
40 | - (libev) use inotify even on buggy kernels, but don't rely on it. |
|
|
41 | - (libev) use inotify only as an added hint on network filesystems. |
|
|
42 | |
|
|
43 | 3.48 Thu Oct 30 09:09:48 CET 2008 |
| 3 | - (libev) use a generation counter to detect spurious epoll events |
44 | - (libev) use a generation counter to detect spurious epoll events |
| 4 | and recreate the kernel in such a case (sorry, it's slow, but |
45 | and recreate the kernel in such a case (sorry, it's slow, but |
| 5 | I didn't design epoll...). |
46 | I didn't design epoll...). |
| 6 | - (libev) optimise away an EPOLL_CTL_ADD/MOD combo in the epoll |
47 | - (libev) optimise away an EPOLL_CTL_ADD/MOD combo in the epoll |
| 7 | backend in some cases. |
48 | backend in some cases. |
