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

Comparing Coro/Changes (file contents):
Revision 1.124 by root, Tue Dec 27 00:42:24 2005 UTC vs.
Revision 1.300 by root, Wed Dec 5 11:32:58 2007 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2
3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO.
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)).
7
84.31 Wed Dec 5 12:32:39 CET 2007
9 - remove warn statement form Coro::Util.
10
114.3 Tue Dec 4 20:33:14 CET 2007
12 - calls to the idle function could cause stack corruption
13 when the stack changed.
14 - do no longer rely on the presence of EV::DNS if EV is used
15 (because it is gone), but instead take avdantage of EV::ADNS
16 if available.
17 - add ($) prototypes to all functions in Coro::Storable.
18 - use a conventional (safer) idle callback in Coro::EV.
19 - do accept quickly in Coro::Debug to avoid endless loops.
20
214.22 Fri Nov 30 16:04:04 CET 2007
22 - really use optimised versions for Event and EV in Coro::Util
23 and Coro::Handle.
24
254.21 Sun Nov 25 10:48:59 CET 2007
26 - fix a spurious memory read.
27 - Coro::EV no longer keeps the eventloop "alive".
28
294.2 Fri Nov 9 20:47:05 CET 2007
30 - enable/disable tracing from a new coroutine, not a pooled one.
31 - fix a memleak in Coro::Event.
32 - removed killall call from fork_eval.
33 - made sure store_fd is already loaded so that fork_eval
34 does not have to parse autoload in each subprocess.
35 - only use assembly method if -O switch is in $Config{optimize}.
36 - add (optional) Coro::EV module, so far the best event loop module
37 directly supported by Coro.
38 - if the event model is EV, use EV::DNS to resolve
39 stuff in Coro::Util.
40 - don't get confused by multiple event notifications in Coro::Handle.
41 - initial support for EV (libevent interface).
42 - require Event and EV using configure_requires, to force their existance.
43
444.13 Wed Oct 24 07:26:45 CEST 2007
45 - add Coro::Storable::blocking_thaw.
46 - use a vastly more complicated technique to localise
47 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
48 - use a coroutine for the idle callback Coro::Event,
49 instead of running Event in the current coroutine context.
50 This also catches recursive invocations.
51 - actually report fork errors in gethostbyname and inet_aton.
52
534.11 Thu Oct 11 02:40:24 CEST 2007
54 - port to threaded perls.
55
564.1 Thu Oct 11 02:38:16 CEST 2007
57 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
58 be local to each coro (see Coro::State).
59 - incompatible change: for very deep reasons, cede and cede_notself
60 cannot return anything, so nothing will be returned.
61 - possibly bring back 5.10 compatibility (untested).
62 - work around stupid (and wrong) warning on 5.10 :(.
63 - overlay the saved state over the context stack. This saves
64 a few hundred bytes per coroutine on average and also
65 speeds up context switching a bit.
66 - further tune default stack sizes.
67 - (more) correctly calculate stack usage in coro_rss.
68 - Coro::Storable::blocking_* did not properly lock
69 resulting in races between coroutines.
70 - added Coro::Storable::guard.
71 - stopping to trace a coroutine could destroy the cctx of
72 an unrelated coroutine.
73 - explain the relationship between Perl and C coroutines in
74 more detail in Coro::State.
75 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
76 causing a fork per resolve. This also affected Coro::Socket.
77 - switch to a separate stack in $coro->call/eval to avoid
78 invalidating pointers.
79
804.03 Sat Oct 6 21:24:00 CEST 2007
81 - added Coro::throw method.
82 - minor code cleanups.
83
844.02 Sat Oct 6 02:36:47 CEST 2007
85 - fix a very minor per-coroutine memleak (a single codereference).
86 - fixed a bug where the currently in-use c context would be freed
87 prematurely (can happen only when programs change the stacksize
88 or use tracing).
89 - tracing can no longer keep a coro alive after it terminated.
90 - do static branch prediction in the common path for gcc. gives
91 about 2-5% speed improvement here.
92
934.01 Fri Oct 5 22:10:49 CEST 2007
94 - instead of recreating *a* standard output handle we simply
95 use STDOUT, which is faster and hopefully more robust.
96
974.0 Fri Oct 5 12:56:00 CEST 2007
98 - incompatibly changed Coro::Storable::freeze.
99 - major new feature: added Coro::Debug, for interactive coroutine
100 debugging, tracing and much more.
101 - major bug fix: unbelievable, but true: $_, $/ and many other
102 "saved" variables actually weren't being saved. This has been fixed,
103 of course, while increasing performance while losing all the save
104 flags.
105 - save flags are gone, and all the api functions dealing with them.
106 - added Coro::Semaphore::adjust.
107 - added Coro::Util::fork_eval.
108 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
109 - added Coro::killall.
110 - reduce initial stack sizes to allow for "micro-coroutines".
111 - better async_pool resource management, moved parts of async_pool
112 handling to XS (major speed improvement).
113 - actually croak before modifying important data structures.
114 - refuse to transfer while compiling.
115 - possibly support eval EXPR better now.
116 - enable assembly per default on linux+bsd x86+amd64.
117 - all internal members were renamed _something for easier subclassing.
118 - many minor tweaks.
119
1203.63 Wed May 16 14:10:06 CEST 2007
121 - implement handcoded assembly for x86/amd64 SVR ABI.
122
1233.62 Fri Apr 27 21:36:06 CEST 2007
124 - upgrade libcoro (which might set unwind info correctly).
125 - change default on linux to setjmp/longjmp.
126
1273.61 Thu Apr 19 12:36:18 CEST 2007
128 - Coro::Storable caused an endless loop when thawing invalid
129 pst's sometimes.
130 - use a Semaphore in Coro::Storable, as Storable doesn't
131 seem to be reentrant (although it is documented to
132 be threadsafe...).
133 - fix Coro::Signal to bring back the original unreliable
134 but stateful semantics.
135 - fixed a lot of typos in Coro.pm (patch submitted by David
136 Steinbrunner, which applied flawlessly).
137
1383.6 Sat Apr 14 17:13:31 CEST 2007
139 - added some bugfixes to get eg/myhttpd working again.
140 - added Coro::Storable for often-cede'ing freeze/thaw.
141 - try to do a clean exit when a coroutine calls exit
142 (EXPERIMENTAL).
143 - got rid of indirect call through _coro_init.
144 - updated the partly antique examples in eg/ to
145 work again and be a bit less magic, too.
146 - fixed Coro::Signal semantics to work as documented again.
147
1483.55 Sun Mar 25 01:20:47 CET 2007
149 - add SAVE_DEFFH to save the default filehandle and enable
150 it by default.
151 - finally move socket-operations from Coro::Socket to Coro::Handle
152 to be able to unblock foreign sockets.
153 - add Coro::State::save_also and guarded_save.
154 - add count accessor to Coro::Semaphore.
155 - add Coro::State::cctx_stacksize.
156 - just for the fun of it, do not rely on implicit context,
157 which can dramatically improve performance, but people
158 using windows-process-emulation perls do not care much about
159 performance.
160
1613.51 Sun Mar 4 14:18:04 CET 2007
162 - fixed a problem when you weakened references to Coro::State's
163 (patch by Yuval Kogman).
164
1653.501 Wed Feb 28 12:44:07 CET 2007
166 - rename some global symbols as macosx from hell redefines
167 them without asking.
168
1693.5 Tue Feb 13 20:22:53 CET 2007
170 - do AnyEvent model detection earlier, avoiding problems
171 caused by first using AnyEvent and later Coro::Event.
172 - implement and document Coro::Event event objects.
173 - fix a potential problem in Coro::Event causing crashes.
174 - initialise PL_comppad when creating a new coroutine,
175 avoids crashes on early coro destruction.
176
1773.41 Mon Jan 22 19:19:49 CET 2007
178 - readline on Coro::Handle did not support undefined $/,
179 nor did it deliver partial lines on EOF or error.
180 - implement malloc fallback for stack allocation because
181 stupid broken idiotic OSX has a stupid broken
182 idiotic fits-the-whole-os mmap "implementation" and
183 my dick feels longer if Coro is portable even to
184 obsolete platforms.
185
1863.4 Fri Jan 19 21:52:54 CET 2007
187 - remove t/09_timer.t, as it isn't really testing much
188 but was rather flaky in practise.
189 - async_pool coro would keep arguments and callback alive until
190 it was reused.
191 - cancellation of a coroutine could cause spurious idle calls
192 in cede_notself.
193
1943.3 Sat Jan 6 03:45:00 CET 2007
195 - implement $coro->on_destroy.
196 - Coro::Event blocking semantics have been changed,
197 documented and - hopefully - improved.
198 - fix nice adding, not subtracting, from priority.
199 - fix ->prio and api_is_ready (patch by Mark Hinds).
200 - fixed an assert ("... == prev__cctx->idle_te")
201 that could errronously trigger.
202 - fix various large and small memleaks.
203 - use a (hopefully) more stable cancel implementation
204 that immediately frees the coroutine data.
205 - cede/cede_notself return a status now.
206 - added Coro::guard function.
207 - added a global coroutine pool for jobs (on my machine,
208 I can create and execute 48k simple coros/s with async,
209 and 128k coros with async_pool).
210 - Coro::AIO now uses the coroutine priority as io priority.
211
2123.2 Fri Dec 22 05:07:09 CET 2006
213 - improve portability to slightly older perls.
214 - use cleaner coroutine destruction.
215 - simplify configuration for users.
216 - optionally (unrecommended) prefer perl functions over
217 their coro replacements.
218
2193.11 Tue Dec 5 13:11:24 CET 2006
220 - fixed some bogus assert's, but as perl.h disables assert even
221 without NDEBUG (thank you very much), not too many people should
222 notice (that did include myself). Andreas König noticed, though :)
223 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
224 it. Also noticed by Andreas König :)
225 - save/restore tainted status.
226 - verified to pass the testsuite on my 5.9.5.
227
2283.1 Mon Dec 4 23:03:40 CET 2006
229 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
230 - incompatible change: transfer flags are now per-state.
231 - give a better error message on deadlock.
232 - document Coro::nready.
233 - enhanced testsuite.
234
2353.01 Sun Dec 3 23:47:42 CET 2006
236 - forgot to include Coro::Timer.
237
2383.0 Sun Dec 3 22:57:25 CET 2006
239 - the "FINALLY COMPLETELY STABLE" release (coming soon:
240 the "FAMOUS LAST WORDS" release).
241 - implement a new stack sharing algorithm, which uses a stack
242 pool (size currently hardcoded).
243 - make stack sharing mandatory (it no longer uses a heuristic).
244 - eval/die no longer cause weird problems under heavy use.
245 - Coro::Event could cause livelocks if it was used but
246 no Coro::Event watchers were used.
247 - Coro::Event now uses asynccheck as prepare does not
248 check for changed watchers.
249 - Coro::Event allows multiple waiting coros and will wake up one
250 per event.
251 - Coro::Event should be cleaner and more efficient now.
252 - new utility function Coro::unblock_sub.
253 - document the sad fact that Event is no longer reentrant.
254 - putting a coroutine into the ready queue twice could under
255 some circumstances lead to stack corruption.
256 - minor incompatibility: subclassing Coro::State is supported
257 directly now without going through a _coro_state member.
258 - state/coro switching is much faster now.
259 - very minor optimisations and code/documentation cleanup.
260 - avoid problems due to compiler inlining.
261 - removed timers from Coro::Timer -> use AnyEvent instead.
262 - replaced Coro::idle coroutine by (cleaner) idle handler.
263 - updated to newest libcoro.
264 - implement enhanced support for valgrind.
265 - implement is_ready and return value for ready.
266 - removed Coro::Cont, it was a misnomer (it's generators),
267 and it was somewhat annoying to get it right. Will come back
268 if somebody asks for it :)
269 - many ->wait methods and Coro::Event could return
270 spuriously without the event having happened.
271
2722.5 Tue Nov 7 12:22:33 CET 2006
273 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
274 use AnyEvent, moved them to Coro/.
275 - added Coro::LWP which contains all the uglyness required to
276 make LWP non-blocking.
277 - should work with perl 5.9.x now (Andreas König made me do it).
278 - fixed another bug in Coro::Select when the timeout was undef.
279 - reuse PL_start_env for all coros, saves some memory per coroutine.
280 - manage PL_top_env differently, hopefully to avoid panic: top_env.
281 - timeout argument was not properly used in Coro::Socket.
282 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
283 - emulate undocumented(!) functionality of IO::Socket required
284 by LWP(!!).
285 - updated eg/lwp to work with newer lwp's.
286 - remove "FATAL: uncaught exception" prefix. Coroutines that die
287 kill the whole process, just as exceptions in the main "coroutine"
288 did already.
289
2902.1 Wed Nov 1 23:01:13 CET 2006
291 - fix a long-standing bug in Coro::Select where select with
292 zero timeout would instead change the current default filehandle.
293 - use a simpler and hopefully more robust way to clone padlists
294 (uses less memory and a perl function instead of our own).
295 - coro can now create a stack guard area on many architectures.
296 - Coro::AIO properly reexports additional functions from IO::AIO.
297 - updated libcoro with a workaround for OS X,
298 pach and testing by Michael Schwern.
299
3002.0 Tue Oct 24 05:47:17 CEST 2006
301 - support additional aio requests in Coro::AIO.
302
3031.9 (never properly released due to a glitch)
304
3051.8 Thu Feb 2 00:59:06 CET 2006
306 - applied suggested patch by SAMV to avoid problems during stupid
307 mark & sweep gc run.
308 - applied patch by Scott Walters for 5.9.3 compatibility.
2 309
31.7 Tue Dec 27 01:41:58 CET 2005 3101.7 Tue Dec 27 01:41:58 CET 2005
4 - added Coro::AIO, a thin wrapper around IO::AIO. 311 - added Coro::AIO, a thin wrapper around IO::AIO.
5 - improved Makefile.PL explanations. 312 - improved Makefile.PL explanations.
6 313

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines