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

Comparing Coro/Changes (file contents):
Revision 1.154 by root, Fri Nov 24 15:34:33 2006 UTC vs.
Revision 1.293 by root, Sun Nov 25 13:53:48 2007 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: check wether stack-sharing is still effective 3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO.
4TODO: more effective stack sharing by having one "default stack" for "simple" coroutines? 4TODO: maybe implement a default message channel, very much like Erlang
5TODO: verify exception frame changes and force stack copies? 5 (which is cool in a lot of important aspects (failures!),
6 but very lacking in others (higher level ipc)).
6 7
84.22
9 - really use optimised versions for Event and EV in Coro::Util
10 and Coro::Handle.
11
124.21 Sun Nov 25 10:48:59 CET 2007
13 - fix a spurious memory read.
14 - Coro::EV no longer keeps the eventloop "alive".
15
164.2 Fri Nov 9 20:47:05 CET 2007
17 - enable/disable tracing from a new coroutine, not a pooled one.
18 - fix a memleak in Coro::Event.
19 - removed killall call from fork_eval.
20 - made sure store_fd is already loaded so that fork_eval
21 does not have to parse autoload in each subprocess.
22 - only use assembly method if -O switch is in $Config{optimize}.
23 - add (optional) Coro::EV module, so far the best event loop module
24 directly supported by Coro.
25 - if the event model is EV, use EV::DNS to resolve
26 stuff in Coro::Util.
27 - don't get confused by multiple event notifications in Coro::Handle.
28 - initial support for EV (libevent interface).
29 - require Event and EV using configure_requires, to force their existance.
30
314.13 Wed Oct 24 07:26:45 CEST 2007
32 - add Coro::Storable::blocking_thaw.
33 - use a vastly more complicated technique to localise
34 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
35 - use a coroutine for the idle callback Coro::Event,
36 instead of running Event in the current coroutine context.
37 This also catches recursive invocations.
38 - actually report fork errors in gethostbyname and inet_aton.
39
404.11 Thu Oct 11 02:40:24 CEST 2007
41 - port to threaded perls.
42
434.1 Thu Oct 11 02:38:16 CEST 2007
44 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
45 be local to each coro (see Coro::State).
46 - incompatible change: for very deep reasons, cede and cede_notself
47 cannot return anything, so nothing will be returned.
48 - possibly bring back 5.10 compatibility (untested).
49 - work around stupid (and wrong) warning on 5.10 :(.
50 - overlay the saved state over the context stack. This saves
51 a few hundred bytes per coroutine on average and also
52 speeds up context switching a bit.
53 - further tune default stack sizes.
54 - (more) correctly calculate stack usage in coro_rss.
55 - Coro::Storable::blocking_* did not properly lock
56 resulting in races between coroutines.
57 - added Coro::Storable::guard.
58 - stopping to trace a coroutine could destroy the cctx of
59 an unrelated coroutine.
60 - explain the relationship between Perl and C coroutines in
61 more detail in Coro::State.
62 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
63 causing a fork per resolve. This also affected Coro::Socket.
64 - switch to a separate stack in $coro->call/eval to avoid
65 invalidating pointers.
66
674.03 Sat Oct 6 21:24:00 CEST 2007
68 - added Coro::throw method.
69 - minor code cleanups.
70
714.02 Sat Oct 6 02:36:47 CEST 2007
72 - fix a very minor per-coroutine memleak (a single codereference).
73 - fixed a bug where the currently in-use c context would be freed
74 prematurely (can happen only when programs change the stacksize
75 or use tracing).
76 - tracing can no longer keep a coro alive after it terminated.
77 - do static branch prediction in the common path for gcc. gives
78 about 2-5% speed improvement here.
79
804.01 Fri Oct 5 22:10:49 CEST 2007
81 - instead of recreating *a* standard output handle we simply
82 use STDOUT, which is faster and hopefully more robust.
83
844.0 Fri Oct 5 12:56:00 CEST 2007
85 - incompatibly changed Coro::Storable::freeze.
86 - major new feature: added Coro::Debug, for interactive coroutine
87 debugging, tracing and much more.
88 - major bug fix: unbelievable, but true: $_, $/ and many other
89 "saved" variables actually weren't being saved. This has been fixed,
90 of course, while increasing performance while losing all the save
91 flags.
92 - save flags are gone, and all the api functions dealing with them.
93 - added Coro::Semaphore::adjust.
94 - added Coro::Util::fork_eval.
95 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
96 - added Coro::killall.
97 - reduce initial stack sizes to allow for "micro-coroutines".
98 - better async_pool resource management, moved parts of async_pool
99 handling to XS (major speed improvement).
100 - actually croak before modifying important data structures.
101 - refuse to transfer while compiling.
102 - possibly support eval EXPR better now.
103 - enable assembly per default on linux+bsd x86+amd64.
104 - all internal members were renamed _something for easier subclassing.
105 - many minor tweaks.
106
1073.63 Wed May 16 14:10:06 CEST 2007
108 - implement handcoded assembly for x86/amd64 SVR ABI.
109
1103.62 Fri Apr 27 21:36:06 CEST 2007
111 - upgrade libcoro (which might set unwind info correctly).
112 - change default on linux to setjmp/longjmp.
113
1143.61 Thu Apr 19 12:36:18 CEST 2007
115 - Coro::Storable caused an endless loop when thawing invalid
116 pst's sometimes.
117 - use a Semaphore in Coro::Storable, as Storable doesn't
118 seem to be reentrant (although it is documented to
119 be threadsafe...).
120 - fix Coro::Signal to bring back the original unreliable
121 but stateful semantics.
122 - fixed a lot of typos in Coro.pm (patch submitted by David
123 Steinbrunner, which applied flawlessly).
124
1253.6 Sat Apr 14 17:13:31 CEST 2007
126 - added some bugfixes to get eg/myhttpd working again.
127 - added Coro::Storable for often-cede'ing freeze/thaw.
128 - try to do a clean exit when a coroutine calls exit
129 (EXPERIMENTAL).
130 - got rid of indirect call through _coro_init.
131 - updated the partly antique examples in eg/ to
132 work again and be a bit less magic, too.
133 - fixed Coro::Signal semantics to work as documented again.
134
1353.55 Sun Mar 25 01:20:47 CET 2007
136 - add SAVE_DEFFH to save the default filehandle and enable
137 it by default.
138 - finally move socket-operations from Coro::Socket to Coro::Handle
139 to be able to unblock foreign sockets.
140 - add Coro::State::save_also and guarded_save.
141 - add count accessor to Coro::Semaphore.
142 - add Coro::State::cctx_stacksize.
143 - just for the fun of it, do not rely on implicit context,
144 which can dramatically improve performance, but people
145 using windows-process-emulation perls do not care much about
146 performance.
147
1483.51 Sun Mar 4 14:18:04 CET 2007
149 - fixed a problem when you weakened references to Coro::State's
150 (patch by Yuval Kogman).
151
1523.501 Wed Feb 28 12:44:07 CET 2007
153 - rename some global symbols as macosx from hell redefines
154 them without asking.
155
1563.5 Tue Feb 13 20:22:53 CET 2007
157 - do AnyEvent model detection earlier, avoiding problems
158 caused by first using AnyEvent and later Coro::Event.
159 - implement and document Coro::Event event objects.
160 - fix a potential problem in Coro::Event causing crashes.
161 - initialise PL_comppad when creating a new coroutine,
162 avoids crashes on early coro destruction.
163
1643.41 Mon Jan 22 19:19:49 CET 2007
165 - readline on Coro::Handle did not support undefined $/,
166 nor did it deliver partial lines on EOF or error.
167 - implement malloc fallback for stack allocation because
168 stupid broken idiotic OSX has a stupid broken
169 idiotic fits-the-whole-os mmap "implementation" and
170 my dick feels longer if Coro is portable even to
171 obsolete platforms.
172
1733.4 Fri Jan 19 21:52:54 CET 2007
174 - remove t/09_timer.t, as it isn't really testing much
175 but was rather flaky in practise.
176 - async_pool coro would keep arguments and callback alive until
177 it was reused.
178 - cancellation of a coroutine could cause spurious idle calls
179 in cede_notself.
180
1813.3 Sat Jan 6 03:45:00 CET 2007
182 - implement $coro->on_destroy.
183 - Coro::Event blocking semantics have been changed,
184 documented and - hopefully - improved.
185 - fix nice adding, not subtracting, from priority.
186 - fix ->prio and api_is_ready (patch by Mark Hinds).
187 - fixed an assert ("... == prev__cctx->idle_te")
188 that could errronously trigger.
189 - fix various large and small memleaks.
190 - use a (hopefully) more stable cancel implementation
191 that immediately frees the coroutine data.
192 - cede/cede_notself return a status now.
193 - added Coro::guard function.
194 - added a global coroutine pool for jobs (on my machine,
195 I can create and execute 48k simple coros/s with async,
196 and 128k coros with async_pool).
197 - Coro::AIO now uses the coroutine priority as io priority.
198
1993.2 Fri Dec 22 05:07:09 CET 2006
200 - improve portability to slightly older perls.
201 - use cleaner coroutine destruction.
202 - simplify configuration for users.
203 - optionally (unrecommended) prefer perl functions over
204 their coro replacements.
205
2063.11 Tue Dec 5 13:11:24 CET 2006
207 - fixed some bogus assert's, but as perl.h disables assert even
208 without NDEBUG (thank you very much), not too many people should
209 notice (that did include myself). Andreas König noticed, though :)
210 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
211 it. Also noticed by Andreas König :)
212 - save/restore tainted status.
213 - verified to pass the testsuite on my 5.9.5.
214
2153.1 Mon Dec 4 23:03:40 CET 2006
216 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
217 - incompatible change: transfer flags are now per-state.
218 - give a better error message on deadlock.
219 - document Coro::nready.
220 - enhanced testsuite.
221
2223.01 Sun Dec 3 23:47:42 CET 2006
223 - forgot to include Coro::Timer.
224
2253.0 Sun Dec 3 22:57:25 CET 2006
226 - the "FINALLY COMPLETELY STABLE" release (coming soon:
227 the "FAMOUS LAST WORDS" release).
228 - implement a new stack sharing algorithm, which uses a stack
229 pool (size currently hardcoded).
230 - make stack sharing mandatory (it no longer uses a heuristic).
231 - eval/die no longer cause weird problems under heavy use.
7 - Coro::Event could cause livelocks if it was used but 232 - Coro::Event could cause livelocks if it was used but
8 Event watchers were used directly. 233 no Coro::Event watchers were used.
234 - Coro::Event now uses asynccheck as prepare does not
235 check for changed watchers.
236 - Coro::Event allows multiple waiting coros and will wake up one
237 per event.
9 - Coro::Event should be cleaner and more efficient now. 238 - Coro::Event should be cleaner and more efficient now.
239 - new utility function Coro::unblock_sub.
240 - document the sad fact that Event is no longer reentrant.
241 - putting a coroutine into the ready queue twice could under
242 some circumstances lead to stack corruption.
10 - minor incompatibility: subclassing is supported directly now 243 - minor incompatibility: subclassing Coro::State is supported
11 without going through a _coro_state member. 244 directly now without going through a _coro_state member.
12 - process swicthing is much faster now. 245 - state/coro switching is much faster now.
13 - very minor optimisations and code/documentation cleanup. 246 - very minor optimisations and code/documentation cleanup.
14 - avoid problems due to compiler inlining. 247 - avoid problems due to compiler inlining.
15 - possibly better support for threads.
16 - removed timers from Coro::Timer -> use AnyEvent instead. 248 - removed timers from Coro::Timer -> use AnyEvent instead.
17 - replaced Coro::idle coroutine by (cleaner) idle handler. 249 - replaced Coro::idle coroutine by (cleaner) idle handler.
250 - updated to newest libcoro.
251 - implement enhanced support for valgrind.
252 - implement is_ready and return value for ready.
253 - removed Coro::Cont, it was a misnomer (it's generators),
254 and it was somewhat annoying to get it right. Will come back
255 if somebody asks for it :)
256 - many ->wait methods and Coro::Event could return
257 spuriously without the event having happened.
18 258
192.5 Tue Nov 7 12:22:33 CET 2006 2592.5 Tue Nov 7 12:22:33 CET 2006
20 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket 260 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
21 use AnyEvent, moved them to Coro/. 261 use AnyEvent, moved them to Coro/.
22 - added Coro::LWP which contains all the uglyness required to 262 - added Coro::LWP which contains all the uglyness required to

Diff Legend

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