| 1 | Revision history for Perl extension EV |
1 | Revision history for Perl extension EV |
| 2 | |
2 | |
| 3 | Changes marked with (libev) are changes in libev, and have more |
3 | Changes marked with (libev) are changes in libev, and have more |
| 4 | documentation in the libev Changes file. |
4 | documentation in the libev Changes file. |
|
|
5 | |
|
|
6 | 4.29 Fri Nov 22 15:34:29 CET 2019 |
|
|
7 | - (libev) add io uring autoconf and non-autoconf detection, |
|
|
8 | the latter of which should disable io_uring compilation |
|
|
9 | on old systems. |
|
|
10 | |
|
|
11 | 4.28 Tue Nov 19 13:55:39 CET 2019 |
|
|
12 | - (libev) fix ev_port backend, thanks to David H. Gutteridge for |
|
|
13 | - (libev) many bugfixes in linuxaio backend. |
|
|
14 | - (libev) experimental io uring interface. |
|
|
15 | reminding me to actually release the fix. |
|
|
16 | - try to name ev.h more explicitly, to hopefully improve portability. |
|
|
17 | - opportunistically round up wait times for poll and epoll backend, |
|
|
18 | to avoid unnecessary loop iterations. |
|
|
19 | - add build dependency on ev_linuxaio.c. |
|
|
20 | - quickly (re)-ported to minix 3.3 before minix crashed again. |
|
|
21 | |
|
|
22 | 4.27 Thu Jun 27 09:39:58 CEST 2019 |
|
|
23 | - (libev) completely rewritten linuxaio backend, maybe |
|
|
24 | usable as a general-use backend. |
|
|
25 | - (libev) use more aggressive assertions to catch |
|
|
26 | more usage errors. |
|
|
27 | - allow users to re-enable assert() in case it is disabled |
|
|
28 | by perl (which is typically the case). |
|
|
29 | |
|
|
30 | 4.26 Mon Jun 24 23:39:40 CEST 2019 |
|
|
31 | - (libev) included experimental linux aio backend. |
|
|
32 | - allow the linux aio backend to be used by default |
|
|
33 | only when explicitly configured during Makefile.PL time. |
|
|
34 | |
|
|
35 | 4.25 Fri Dec 21 08:04:26 CET 2018 |
|
|
36 | - (libev) updated to libev 4.25, minor fixes and enhancements. |
|
|
37 | - document the requirement to "use EV" when using EV::MakeMaker. |
|
|
38 | - only use __register_atfork with glibc and uclibc, as musl |
|
|
39 | defines __linux__, but doesn't implement the linux standard |
|
|
40 | base ABI, nor makes itself detectable via a macro, both apparently |
|
|
41 | by design, winning the "broken by design 2016 award" - well done. |
|
|
42 | - correct EV::periodic example 24h after midnight -> one day after |
|
|
43 | midnight (reported by Felix Ostmann). |
|
|
44 | |
|
|
45 | 4.22 Sun Dec 20 02:34:39 CET 2015 |
|
|
46 | - (libev) when epoll detects unremovable fds in the fd set, rebuild |
|
|
47 | only the epoll descriptor, not the signal pipe, to avoid |
|
|
48 | SIGPIPE in ev_async_send. This doesn't solve it on fork, |
|
|
49 | so document what needs to be done in ev_loop_fork |
|
|
50 | (analyzed by Benjamin Mahler). |
|
|
51 | - (libev) remove superfluous sys/timeb.h include on win32 |
|
|
52 | (analyzed by Jason Madden). |
|
|
53 | - updated libecb. |
|
|
54 | |
|
|
55 | 4.21 Mon Jul 13 21:47:33 CEST 2015 |
|
|
56 | - allow argument in timer->again. |
|
|
57 | - document timer->remaining. |
|
|
58 | - document default repeat value for timer->set. |
|
|
59 | |
|
|
60 | 4.20 Sat Jun 20 13:07:34 CEST 2015 |
|
|
61 | - added stability canary support. |
|
|
62 | - truly rely on automatic configuration for clock_gettime and others. |
|
|
63 | unfortunately, this doesn't help with activeperl and similar perls, |
|
|
64 | which define _POSIX_TIMERS without actually implementing any of it. |
|
|
65 | - (libev, ecb) make it compile as C++ again. |
|
|
66 | - (libev) fix a potential aliasing issue when accessing watcher cbs. |
|
|
67 | |
|
|
68 | 4.18 Sat Sep 6 20:37:23 CEST 2014 |
|
|
69 | - use slightly better weay to find includes, to support multiarch |
|
|
70 | on newer perls. |
|
|
71 | |
|
|
72 | 4.17 Fri Apr 11 06:22:38 CEST 2014 |
|
|
73 | - perl5porters broke Async::Interrupt, BDB, EV, IO::AIO, OpenCL |
|
|
74 | without warning by switching the meaning of USE_SOCKETS_AS_HANDLES |
|
|
75 | in 5.18. What's so attractive about giving a shit about backwards |
|
|
76 | compatibility - I will never understand. |
|
|
77 | |
|
|
78 | 4.16 Sat Mar 8 16:49:05 CET 2014 |
|
|
79 | - make sure EV::CHECK is available at all times. |
|
|
80 | - avoid a spurious warning when perl -w overrides "no warnings". |
|
|
81 | - use perl's signal implementation on WIN32. |
|
|
82 | - ensure extern "C" function pointers are used for externally-visible |
|
|
83 | pointers. |
|
|
84 | - (libev) mark event pipe fd as cloexec after a fork (analyzed by Sami Farin). |
|
|
85 | - (ecb) support m68k, m88k and sh (patch by Miod Vallat). |
|
|
86 | - (libev) in the absence of autoconf, do not use the clock syscall |
|
|
87 | on glibc >= 2.17 (avoids the syscall AND -lrt on systems |
|
|
88 | doing clock_gettime in userspace). |
|
|
89 | - (ecb) work around memory barriers and volatile apparently both being |
|
|
90 | broken in visual studio 2008 and later (analysed and patch by Nicolas Noble). |
|
|
91 | |
|
|
92 | 4.15 Fri Mar 1 12:15:53 CET 2013 |
|
|
93 | - (libev) upgrade to 4.15 - too many changes to list. |
|
|
94 | - EV::run now returns a boolean. |
|
|
95 | - API version 5:1. |
|
|
96 | - document that cleanup watchers are not available via perl. |
|
|
97 | - cast I32 to int in error message printf. |
|
|
98 | - remove dependencies on librt and libpthreads on GNU/Linux. |
|
|
99 | |
|
|
100 | 4.11 Sat Feb 4 19:56:26 CET 2012 |
|
|
101 | - (libev) implement memory fences for (obsolete) llvm-gcc. |
|
|
102 | |
|
|
103 | 4.10 Thu Jan 19 18:54:23 CET 2012 |
|
|
104 | - (libev) fix a race where the workaround against the epoll fork |
|
|
105 | bugs caused signals to not be handled anymore. |
|
|
106 | - (libev) correct backend_fudge for most backends, and implement |
|
|
107 | a windows specific workaround to avoid looping because we call |
|
|
108 | both select and Sleep, both with different time resolutions. |
|
|
109 | - e_new wasn't declared static (causing very minor .so bloat). |
|
|
110 | - replace more old api names by new ones. |
| 5 | |
111 | |
| 6 | 4.03 Tue Jan 11 14:51:05 CET 2011 |
112 | 4.03 Tue Jan 11 14:51:05 CET 2011 |
| 7 | - do not avoid the clock_gettime call on GNU/Linux anymore, as EV |
113 | - do not avoid the clock_gettime call on GNU/Linux anymore, as EV |
| 8 | links against -lpthread anyways - as a result, EV might now take |
114 | links against -lpthread anyways - as a result, EV might now take |
| 9 | advantage of fast userspace clock_gettime implementations, |
115 | advantage of fast userspace clock_gettime implementations, |
