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

Comparing Coro/Changes (file contents):
Revision 1.213 by root, Sat Mar 17 19:51:57 2007 UTC vs.
Revision 1.310 by root, Thu Dec 27 21:52:21 2007 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: better (earlier) error messages on broken transfer's 3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO.
4TODO: destroy gets only called on cancel: fix(?) 4TODO: maybe implement a default message channel, very much like Erlang
5 (which is cool in a lot of important aspects (failures!),
6 but very lacking in others (higher level ipc)).
5 7
63.52 84.35
9 - "bt" debug command now displays any exceptions
10 from longmess.
11 - allow backtraces in coroutines blocked in Coro::EV,
12 at a <1% speed hit.
13
144.34 Sat Dec 22 17:49:53 CET 2007
15 - upgrade to EV version 2.0 API.
16
174.33 Mon Dec 17 08:36:12 CET 2007
18 - make Coro::AIO etc. loadable in the absence of EV.
19
204.32 Mon Dec 17 07:46:02 CET 2007
21 - majorly improved Coro::Handle's performance with Coro::EV.
22 - implemented the until now mythical Coro::BDB module.
23 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
24
254.31 Wed Dec 5 12:32:39 CET 2007
26 - remove warn statement form Coro::Util.
27
284.3 Tue Dec 4 20:33:14 CET 2007
29 - calls to the idle function could cause stack corruption
30 when the stack changed.
31 - do no longer rely on the presence of EV::DNS if EV is used
32 (because it is gone), but instead take avdantage of EV::ADNS
33 if available.
34 - add ($) prototypes to all functions in Coro::Storable.
35 - use a conventional (safer) idle callback in Coro::EV.
36 - do accept quickly in Coro::Debug to avoid endless loops.
37
384.22 Fri Nov 30 16:04:04 CET 2007
39 - really use optimised versions for Event and EV in Coro::Util
40 and Coro::Handle.
41
424.21 Sun Nov 25 10:48:59 CET 2007
43 - fix a spurious memory read.
44 - Coro::EV no longer keeps the eventloop "alive".
45
464.2 Fri Nov 9 20:47:05 CET 2007
47 - enable/disable tracing from a new coroutine, not a pooled one.
48 - fix a memleak in Coro::Event.
49 - removed killall call from fork_eval.
50 - made sure store_fd is already loaded so that fork_eval
51 does not have to parse autoload in each subprocess.
52 - only use assembly method if -O switch is in $Config{optimize}.
53 - add (optional) Coro::EV module, so far the best event loop module
54 directly supported by Coro.
55 - if the event model is EV, use EV::DNS to resolve
56 stuff in Coro::Util.
57 - don't get confused by multiple event notifications in Coro::Handle.
58 - initial support for EV (libevent interface).
59 - require Event and EV using configure_requires, to force their existance.
60
614.13 Wed Oct 24 07:26:45 CEST 2007
62 - add Coro::Storable::blocking_thaw.
63 - use a vastly more complicated technique to localise
64 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
65 - use a coroutine for the idle callback Coro::Event,
66 instead of running Event in the current coroutine context.
67 This also catches recursive invocations.
68 - actually report fork errors in gethostbyname and inet_aton.
69
704.11 Thu Oct 11 02:40:24 CEST 2007
71 - port to threaded perls.
72
734.1 Thu Oct 11 02:38:16 CEST 2007
74 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
75 be local to each coro (see Coro::State).
76 - incompatible change: for very deep reasons, cede and cede_notself
77 cannot return anything, so nothing will be returned.
78 - possibly bring back 5.10 compatibility (untested).
79 - work around stupid (and wrong) warning on 5.10 :(.
80 - overlay the saved state over the context stack. This saves
81 a few hundred bytes per coroutine on average and also
82 speeds up context switching a bit.
83 - further tune default stack sizes.
84 - (more) correctly calculate stack usage in coro_rss.
85 - Coro::Storable::blocking_* did not properly lock
86 resulting in races between coroutines.
87 - added Coro::Storable::guard.
88 - stopping to trace a coroutine could destroy the cctx of
89 an unrelated coroutine.
90 - explain the relationship between Perl and C coroutines in
91 more detail in Coro::State.
92 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
93 causing a fork per resolve. This also affected Coro::Socket.
94 - switch to a separate stack in $coro->call/eval to avoid
95 invalidating pointers.
96
974.03 Sat Oct 6 21:24:00 CEST 2007
98 - added Coro::throw method.
99 - minor code cleanups.
100
1014.02 Sat Oct 6 02:36:47 CEST 2007
102 - fix a very minor per-coroutine memleak (a single codereference).
103 - fixed a bug where the currently in-use c context would be freed
104 prematurely (can happen only when programs change the stacksize
105 or use tracing).
106 - tracing can no longer keep a coro alive after it terminated.
107 - do static branch prediction in the common path for gcc. gives
108 about 2-5% speed improvement here.
109
1104.01 Fri Oct 5 22:10:49 CEST 2007
111 - instead of recreating *a* standard output handle we simply
112 use STDOUT, which is faster and hopefully more robust.
113
1144.0 Fri Oct 5 12:56:00 CEST 2007
115 - incompatibly changed Coro::Storable::freeze.
116 - major new feature: added Coro::Debug, for interactive coroutine
117 debugging, tracing and much more.
118 - major bug fix: unbelievable, but true: $_, $/ and many other
119 "saved" variables actually weren't being saved. This has been fixed,
120 of course, while increasing performance while losing all the save
121 flags.
122 - save flags are gone, and all the api functions dealing with them.
123 - added Coro::Semaphore::adjust.
124 - added Coro::Util::fork_eval.
125 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
126 - added Coro::killall.
127 - reduce initial stack sizes to allow for "micro-coroutines".
128 - better async_pool resource management, moved parts of async_pool
129 handling to XS (major speed improvement).
130 - actually croak before modifying important data structures.
131 - refuse to transfer while compiling.
132 - possibly support eval EXPR better now.
133 - enable assembly per default on linux+bsd x86+amd64.
134 - all internal members were renamed _something for easier subclassing.
135 - many minor tweaks.
136
1373.63 Wed May 16 14:10:06 CEST 2007
138 - implement handcoded assembly for x86/amd64 SVR ABI.
139
1403.62 Fri Apr 27 21:36:06 CEST 2007
141 - upgrade libcoro (which might set unwind info correctly).
142 - change default on linux to setjmp/longjmp.
143
1443.61 Thu Apr 19 12:36:18 CEST 2007
145 - Coro::Storable caused an endless loop when thawing invalid
146 pst's sometimes.
147 - use a Semaphore in Coro::Storable, as Storable doesn't
148 seem to be reentrant (although it is documented to
149 be threadsafe...).
150 - fix Coro::Signal to bring back the original unreliable
151 but stateful semantics.
152 - fixed a lot of typos in Coro.pm (patch submitted by David
153 Steinbrunner, which applied flawlessly).
154
1553.6 Sat Apr 14 17:13:31 CEST 2007
156 - added some bugfixes to get eg/myhttpd working again.
157 - added Coro::Storable for often-cede'ing freeze/thaw.
158 - try to do a clean exit when a coroutine calls exit
159 (EXPERIMENTAL).
160 - got rid of indirect call through _coro_init.
161 - updated the partly antique examples in eg/ to
162 work again and be a bit less magic, too.
163 - fixed Coro::Signal semantics to work as documented again.
164
1653.55 Sun Mar 25 01:20:47 CET 2007
166 - add SAVE_DEFFH to save the default filehandle and enable
167 it by default.
168 - finally move socket-operations from Coro::Socket to Coro::Handle
169 to be able to unblock foreign sockets.
170 - add Coro::State::save_also and guarded_save.
7 - add count accessor to Coro::Semaphore. 171 - add count accessor to Coro::Semaphore.
8 - add Coro::State::cctx_stacksize. 172 - add Coro::State::cctx_stacksize.
9 - just for the fun of it, do not rely on implicit context, 173 - just for the fun of it, do not rely on implicit context,
10 which can dramatically improve performance, but people 174 which can dramatically improve performance, but people
11 using windows-process-emulation perls do not care much about 175 using windows-process-emulation perls do not care much about

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines