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

Comparing Coro/Changes (file contents):
Revision 1.259 by root, Fri Oct 5 23:38:39 2007 UTC vs.
Revision 1.335 by root, Mon Apr 14 11:28:59 2008 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO. 3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO.
4TODO: maybe implement a default message channel, very much like Erlang 4TODO: maybe implement a default message channel, very much like Erlang's
5 (which is cool in a lot of important aspects (failures!), 5 actor model (which is cool in a lot of important aspects (failures!),
6 but very lacking in others (higher level ipc)). 6 but very lacking in others (higher level ipc)).
7 7
84.02 84.51 Mon Apr 14 13:28:27 CEST 2008
9 - make it compile again on 5.8.
10
114.50 Thu Apr 10 09:43:17 CEST 2008
12 - I did it twice! (see 4.49).
13
144.49 Sun Apr 6 21:23:31 CEST 2008
15 - grr, instead of compiling the recent changes on 5.10 only they
16 were compiled on 5.8 only.
17
184.48 Sun Apr 6 20:36:46 CEST 2008
19 - allow coroutine switches during eval's under 5.10.x, as it seems
20 the parser is a per-interpreter option now, so this is safe
21 (this might fix the odd crashing bug).
22 - drop support for 5.9.x versions: they are dead, jim.
23
244.47 Sun Apr 6 00:37:52 CEST 2008
25 - force cctx allocation on API calls: we know we need to force one
26 and gcc actually manages to confuse our heuristic nowadays,
27 leading to crashes and worse.
28 - document force_cctx.
29
304.46 Fri Apr 4 22:05:43 CEST 2008
31 - upgrade libcoro, resulting in pthread-backend (which was only created
32 to fulfill the rules of the programming languages shootout).
33
344.45 Thu Mar 13 11:55:36 CET 2008
35 - fix a file leaking bug in eg/mhyttpd that would allow
36 downloading of any file (reported by oesi).
37 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
38 (also add testcase).
39 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
40 apparently having loads of fun with that).
41 - implement $state->swap_defsv, ->swap_defav and document ->throw.
42
434.4 Wed Feb 13 16:44:29 CET 2008
44 - only assume .cfi pseudo isns exist on GNU/Linux.
45 - add get_timed method to Coro::Channels.
46 - fixed Coro::Debug manpage.
47 - perl 5.11 compatbility improvement based on patch by
48 Andreas König.
49
504.37 Sun Jan 20 11:25:23 CET 2008
51 - soften the check for an active parser for perl 5.10 (avoiding
52 "Coro::State::transfer called while parsing" in many cases).
53
544.36 Sun Jan 13 10:53:56 CET 2008
55 - reset diehook when terminating from an async_pool as to not
56 trigger any __DIE__ handlers.
57
584.35 Sun Jan 13 04:14:13 CET 2008
59 - "bt" debug command now displays any exceptions
60 from longmess and also skips the innermost
61 stackframes, giving more useufl output.
62 - allow backtraces in coroutines blocked in Coro::EV,
63 at a <1% speed hit.
64 - handle localising of $SIG{__WARN__} and __DIE__
65 properly (with a proper amount of dirty hacking).
66
674.34 Sat Dec 22 17:49:53 CET 2007
68 - upgrade to EV version 2.0 API.
69
704.33 Mon Dec 17 08:36:12 CET 2007
71 - make Coro::AIO etc. loadable in the absence of EV.
72
734.32 Mon Dec 17 07:46:02 CET 2007
74 - majorly improved Coro::Handle's performance with Coro::EV.
75 - implemented the until now mythical Coro::BDB module.
76 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
77
784.31 Wed Dec 5 12:32:39 CET 2007
79 - remove warn statement form Coro::Util.
80
814.3 Tue Dec 4 20:33:14 CET 2007
82 - calls to the idle function could cause stack corruption
83 when the stack changed.
84 - do no longer rely on the presence of EV::DNS if EV is used
85 (because it is gone), but instead take avdantage of EV::ADNS
86 if available.
87 - add ($) prototypes to all functions in Coro::Storable.
88 - use a conventional (safer) idle callback in Coro::EV.
89 - do accept quickly in Coro::Debug to avoid endless loops.
90
914.22 Fri Nov 30 16:04:04 CET 2007
92 - really use optimised versions for Event and EV in Coro::Util
93 and Coro::Handle.
94
954.21 Sun Nov 25 10:48:59 CET 2007
96 - fix a spurious memory read.
97 - Coro::EV no longer keeps the eventloop "alive".
98
994.2 Fri Nov 9 20:47:05 CET 2007
100 - enable/disable tracing from a new coroutine, not a pooled one.
101 - fix a memleak in Coro::Event.
102 - removed killall call from fork_eval.
103 - made sure store_fd is already loaded so that fork_eval
104 does not have to parse autoload in each subprocess.
105 - only use assembly method if -O switch is in $Config{optimize}.
106 - add (optional) Coro::EV module, so far the best event loop module
107 directly supported by Coro.
108 - if the event model is EV, use EV::DNS to resolve
109 stuff in Coro::Util.
110 - don't get confused by multiple event notifications in Coro::Handle.
111 - initial support for EV (libevent interface).
112 - require Event and EV using configure_requires, to force their existance.
113
1144.13 Wed Oct 24 07:26:45 CEST 2007
115 - add Coro::Storable::blocking_thaw.
116 - use a vastly more complicated technique to localise
117 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
118 - use a coroutine for the idle callback Coro::Event,
119 instead of running Event in the current coroutine context.
120 This also catches recursive invocations.
121 - actually report fork errors in gethostbyname and inet_aton.
122
1234.11 Thu Oct 11 02:40:24 CEST 2007
124 - port to threaded perls.
125
1264.1 Thu Oct 11 02:38:16 CEST 2007
127 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
128 be local to each coro (see Coro::State).
129 - incompatible change: for very deep reasons, cede and cede_notself
130 cannot return anything, so nothing will be returned.
131 - possibly bring back 5.10 compatibility (untested).
132 - work around stupid (and wrong) warning on 5.10 :(.
133 - overlay the saved state over the context stack. This saves
134 a few hundred bytes per coroutine on average and also
135 speeds up context switching a bit.
136 - further tune default stack sizes.
137 - (more) correctly calculate stack usage in coro_rss.
138 - Coro::Storable::blocking_* did not properly lock
139 resulting in races between coroutines.
140 - added Coro::Storable::guard.
141 - stopping to trace a coroutine could destroy the cctx of
142 an unrelated coroutine.
143 - explain the relationship between Perl and C coroutines in
144 more detail in Coro::State.
145 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
146 causing a fork per resolve. This also affected Coro::Socket.
147 - switch to a separate stack in $coro->call/eval to avoid
148 invalidating pointers.
149
1504.03 Sat Oct 6 21:24:00 CEST 2007
151 - added Coro::throw method.
152 - minor code cleanups.
153
1544.02 Sat Oct 6 02:36:47 CEST 2007
9 - fix a very minor per-coroutine memleak (a single codereference). 155 - fix a very minor per-coroutine memleak (a single codereference).
10 - fixed a bug where the currently in-use c context would be freed 156 - fixed a bug where the currently in-use c context would be freed
11 prematurely (can happen only when programs change the stacksize 157 prematurely (can happen only when programs change the stacksize
12 or use tracing). 158 or use tracing).
13 - tracing can no longer keep a coro alive after it terminated. 159 - tracing can no longer keep a coro alive after it terminated.
160 - do static branch prediction in the common path for gcc. gives
161 about 2-5% speed improvement here.
14 162
154.01 Fri Oct 5 22:10:49 CEST 2007 1634.01 Fri Oct 5 22:10:49 CEST 2007
16 - instead of recreating *a* standard output handle we simply 164 - instead of recreating *a* standard output handle we simply
17 use STDOUT, which is faster and hopefully more robust. 165 use STDOUT, which is faster and hopefully more robust.
18 166

Diff Legend

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