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

Comparing Coro/Changes (file contents):
Revision 1.96 by root, Tue Aug 10 01:56:30 2004 UTC vs.
Revision 1.347 by root, Tue May 20 15:07:03 2008 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's
5 actor model (which is cool in a lot of important aspects (failures!),
6 but very lacking in others (higher level ipc)).
7
8 - take advantage of async name resolution of AnyEvent::Util.
9 - work around brutal inet_aton override in Coro::LWP.
10
114.7 Sun May 11 00:32:19 CEST 2008
12 - completely reworked the Coro manpage.
13 - added Coro::AnyEvent, generic event loop integration.
14 - implement cancel, ready and kill commands in Coro::Debug.
15 - document find_coro in Coro::Debug.
16 - incompatible change: rename has_stack to has_cctx.
17 - Coro::AIO and Coro::BDB no longer force event model detection,
18 use AnyEvent::AIO and AnyEvent::BDB.
19
204.6 Sat Apr 26 10:05:14 CEST 2008
21 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
22 by the Coro module. It was a mistake to have it in the first place.
23 - (experimental) support for activestate perl 5.10 (method "w").
24 - (experimental) support for strawberry perl (method "a").
25 - coro_sigelem_set did not return a value although it had to,
26 actual impact unknown.
27
284.51 Mon Apr 14 13:28:27 CEST 2008
29 - make it compile again on 5.8.
30
314.50 Thu Apr 10 09:43:17 CEST 2008
32 - I did it twice! (see 4.49).
33
344.49 Sun Apr 6 21:23:31 CEST 2008
35 - grr, instead of compiling the recent changes on 5.10 only they
36 were compiled on 5.8 only.
37
384.48 Sun Apr 6 20:36:46 CEST 2008
39 - allow coroutine switches during eval's under 5.10.x, as it seems
40 the parser is a per-interpreter option now, so this is safe
41 (this might fix the odd crashing bug).
42 - drop support for 5.9.x versions: they are dead, jim.
43
444.47 Sun Apr 6 00:37:52 CEST 2008
45 - force cctx allocation on API calls: we know we need to force one
46 and gcc actually manages to confuse our heuristic nowadays,
47 leading to crashes and worse.
48 - document force_cctx.
49
504.46 Fri Apr 4 22:05:43 CEST 2008
51 - upgrade libcoro, resulting in pthread-backend (which was only created
52 to fulfill the rules of the programming languages shootout).
53
544.45 Thu Mar 13 11:55:36 CET 2008
55 - fix a file leaking bug in eg/mhyttpd that would allow
56 downloading of any file (reported by oesi).
57 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
58 (also add testcase).
59 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
60 apparently having loads of fun with that).
61 - implement $state->swap_defsv, ->swap_defav and document ->throw.
62
634.4 Wed Feb 13 16:44:29 CET 2008
64 - only assume .cfi pseudo isns exist on GNU/Linux.
65 - add get_timed method to Coro::Channels.
66 - fixed Coro::Debug manpage.
67 - perl 5.11 compatbility improvement based on patch by
68 Andreas König.
69
704.37 Sun Jan 20 11:25:23 CET 2008
71 - soften the check for an active parser for perl 5.10 (avoiding
72 "Coro::State::transfer called while parsing" in many cases).
73
744.36 Sun Jan 13 10:53:56 CET 2008
75 - reset diehook when terminating from an async_pool as to not
76 trigger any __DIE__ handlers.
77
784.35 Sun Jan 13 04:14:13 CET 2008
79 - "bt" debug command now displays any exceptions
80 from longmess and also skips the innermost
81 stackframes, giving more useufl output.
82 - allow backtraces in coroutines blocked in Coro::EV,
83 at a <1% speed hit.
84 - handle localising of $SIG{__WARN__} and __DIE__
85 properly (with a proper amount of dirty hacking).
86
874.34 Sat Dec 22 17:49:53 CET 2007
88 - upgrade to EV version 2.0 API.
89
904.33 Mon Dec 17 08:36:12 CET 2007
91 - make Coro::AIO etc. loadable in the absence of EV.
92
934.32 Mon Dec 17 07:46:02 CET 2007
94 - majorly improved Coro::Handle's performance with Coro::EV.
95 - implemented the until now mythical Coro::BDB module.
96 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
97
984.31 Wed Dec 5 12:32:39 CET 2007
99 - remove warn statement form Coro::Util.
100
1014.3 Tue Dec 4 20:33:14 CET 2007
102 - calls to the idle function could cause stack corruption
103 when the stack changed.
104 - do no longer rely on the presence of EV::DNS if EV is used
105 (because it is gone), but instead take avdantage of EV::ADNS
106 if available.
107 - add ($) prototypes to all functions in Coro::Storable.
108 - use a conventional (safer) idle callback in Coro::EV.
109 - do accept quickly in Coro::Debug to avoid endless loops.
110
1114.22 Fri Nov 30 16:04:04 CET 2007
112 - really use optimised versions for Event and EV in Coro::Util
113 and Coro::Handle.
114
1154.21 Sun Nov 25 10:48:59 CET 2007
116 - fix a spurious memory read.
117 - Coro::EV no longer keeps the eventloop "alive".
118
1194.2 Fri Nov 9 20:47:05 CET 2007
120 - enable/disable tracing from a new coroutine, not a pooled one.
121 - fix a memleak in Coro::Event.
122 - removed killall call from fork_eval.
123 - made sure store_fd is already loaded so that fork_eval
124 does not have to parse autoload in each subprocess.
125 - only use assembly method if -O switch is in $Config{optimize}.
126 - add (optional) Coro::EV module, so far the best event loop module
127 directly supported by Coro.
128 - if the event model is EV, use EV::DNS to resolve
129 stuff in Coro::Util.
130 - don't get confused by multiple event notifications in Coro::Handle.
131 - initial support for EV (libevent interface).
132 - require Event and EV using configure_requires, to force their existance.
133
1344.13 Wed Oct 24 07:26:45 CEST 2007
135 - add Coro::Storable::blocking_thaw.
136 - use a vastly more complicated technique to localise
137 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
138 - use a coroutine for the idle callback Coro::Event,
139 instead of running Event in the current coroutine context.
140 This also catches recursive invocations.
141 - actually report fork errors in gethostbyname and inet_aton.
142
1434.11 Thu Oct 11 02:40:24 CEST 2007
144 - port to threaded perls.
145
1464.1 Thu Oct 11 02:38:16 CEST 2007
147 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
148 be local to each coro (see Coro::State).
149 - incompatible change: for very deep reasons, cede and cede_notself
150 cannot return anything, so nothing will be returned.
151 - possibly bring back 5.10 compatibility (untested).
152 - work around stupid (and wrong) warning on 5.10 :(.
153 - overlay the saved state over the context stack. This saves
154 a few hundred bytes per coroutine on average and also
155 speeds up context switching a bit.
156 - further tune default stack sizes.
157 - (more) correctly calculate stack usage in coro_rss.
158 - Coro::Storable::blocking_* did not properly lock
159 resulting in races between coroutines.
160 - added Coro::Storable::guard.
161 - stopping to trace a coroutine could destroy the cctx of
162 an unrelated coroutine.
163 - explain the relationship between Perl and C coroutines in
164 more detail in Coro::State.
165 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
166 causing a fork per resolve. This also affected Coro::Socket.
167 - switch to a separate stack in $coro->call/eval to avoid
168 invalidating pointers.
169
1704.03 Sat Oct 6 21:24:00 CEST 2007
171 - added Coro::throw method.
172 - minor code cleanups.
173
1744.02 Sat Oct 6 02:36:47 CEST 2007
175 - fix a very minor per-coroutine memleak (a single codereference).
176 - fixed a bug where the currently in-use c context would be freed
177 prematurely (can happen only when programs change the stacksize
178 or use tracing).
179 - tracing can no longer keep a coro alive after it terminated.
180 - do static branch prediction in the common path for gcc. gives
181 about 2-5% speed improvement here.
182
1834.01 Fri Oct 5 22:10:49 CEST 2007
184 - instead of recreating *a* standard output handle we simply
185 use STDOUT, which is faster and hopefully more robust.
186
1874.0 Fri Oct 5 12:56:00 CEST 2007
188 - incompatibly changed Coro::Storable::freeze.
189 - major new feature: added Coro::Debug, for interactive coroutine
190 debugging, tracing and much more.
191 - major bug fix: unbelievable, but true: $_, $/ and many other
192 "saved" variables actually weren't being saved. This has been fixed,
193 of course, while increasing performance while losing all the save
194 flags.
195 - save flags are gone, and all the api functions dealing with them.
196 - added Coro::Semaphore::adjust.
197 - added Coro::Util::fork_eval.
198 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
199 - added Coro::killall.
200 - reduce initial stack sizes to allow for "micro-coroutines".
201 - better async_pool resource management, moved parts of async_pool
202 handling to XS (major speed improvement).
203 - actually croak before modifying important data structures.
204 - refuse to transfer while compiling.
205 - possibly support eval EXPR better now.
206 - enable assembly per default on linux+bsd x86+amd64.
207 - all internal members were renamed _something for easier subclassing.
208 - many minor tweaks.
209
2103.63 Wed May 16 14:10:06 CEST 2007
211 - implement handcoded assembly for x86/amd64 SVR ABI.
212
2133.62 Fri Apr 27 21:36:06 CEST 2007
214 - upgrade libcoro (which might set unwind info correctly).
215 - change default on linux to setjmp/longjmp.
216
2173.61 Thu Apr 19 12:36:18 CEST 2007
218 - Coro::Storable caused an endless loop when thawing invalid
219 pst's sometimes.
220 - use a Semaphore in Coro::Storable, as Storable doesn't
221 seem to be reentrant (although it is documented to
222 be threadsafe...).
223 - fix Coro::Signal to bring back the original unreliable
224 but stateful semantics.
225 - fixed a lot of typos in Coro.pm (patch submitted by David
226 Steinbrunner, which applied flawlessly).
227
2283.6 Sat Apr 14 17:13:31 CEST 2007
229 - added some bugfixes to get eg/myhttpd working again.
230 - added Coro::Storable for often-cede'ing freeze/thaw.
231 - try to do a clean exit when a coroutine calls exit
232 (EXPERIMENTAL).
233 - got rid of indirect call through _coro_init.
234 - updated the partly antique examples in eg/ to
235 work again and be a bit less magic, too.
236 - fixed Coro::Signal semantics to work as documented again.
237
2383.55 Sun Mar 25 01:20:47 CET 2007
239 - add SAVE_DEFFH to save the default filehandle and enable
240 it by default.
241 - finally move socket-operations from Coro::Socket to Coro::Handle
242 to be able to unblock foreign sockets.
243 - add Coro::State::save_also and guarded_save.
244 - add count accessor to Coro::Semaphore.
245 - add Coro::State::cctx_stacksize.
246 - just for the fun of it, do not rely on implicit context,
247 which can dramatically improve performance, but people
248 using windows-process-emulation perls do not care much about
249 performance.
250
2513.51 Sun Mar 4 14:18:04 CET 2007
252 - fixed a problem when you weakened references to Coro::State's
253 (patch by Yuval Kogman).
254
2553.501 Wed Feb 28 12:44:07 CET 2007
256 - rename some global symbols as macosx from hell redefines
257 them without asking.
258
2593.5 Tue Feb 13 20:22:53 CET 2007
260 - do AnyEvent model detection earlier, avoiding problems
261 caused by first using AnyEvent and later Coro::Event.
262 - implement and document Coro::Event event objects.
263 - fix a potential problem in Coro::Event causing crashes.
264 - initialise PL_comppad when creating a new coroutine,
265 avoids crashes on early coro destruction.
266
2673.41 Mon Jan 22 19:19:49 CET 2007
268 - readline on Coro::Handle did not support undefined $/,
269 nor did it deliver partial lines on EOF or error.
270 - implement malloc fallback for stack allocation because
271 stupid broken idiotic OSX has a stupid broken
272 idiotic fits-the-whole-os mmap "implementation" and
273 my dick feels longer if Coro is portable even to
274 obsolete platforms.
275
2763.4 Fri Jan 19 21:52:54 CET 2007
277 - remove t/09_timer.t, as it isn't really testing much
278 but was rather flaky in practise.
279 - async_pool coro would keep arguments and callback alive until
280 it was reused.
281 - cancellation of a coroutine could cause spurious idle calls
282 in cede_notself.
283
2843.3 Sat Jan 6 03:45:00 CET 2007
285 - implement $coro->on_destroy.
286 - Coro::Event blocking semantics have been changed,
287 documented and - hopefully - improved.
288 - fix nice adding, not subtracting, from priority.
289 - fix ->prio and api_is_ready (patch by Mark Hinds).
290 - fixed an assert ("... == prev__cctx->idle_te")
291 that could errronously trigger.
292 - fix various large and small memleaks.
293 - use a (hopefully) more stable cancel implementation
294 that immediately frees the coroutine data.
295 - cede/cede_notself return a status now.
296 - added Coro::guard function.
297 - added a global coroutine pool for jobs (on my machine,
298 I can create and execute 48k simple coros/s with async,
299 and 128k coros with async_pool).
300 - Coro::AIO now uses the coroutine priority as io priority.
301
3023.2 Fri Dec 22 05:07:09 CET 2006
303 - improve portability to slightly older perls.
304 - use cleaner coroutine destruction.
305 - simplify configuration for users.
306 - optionally (unrecommended) prefer perl functions over
307 their coro replacements.
308
3093.11 Tue Dec 5 13:11:24 CET 2006
310 - fixed some bogus assert's, but as perl.h disables assert even
311 without NDEBUG (thank you very much), not too many people should
312 notice (that did include myself). Andreas König noticed, though :)
313 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
314 it. Also noticed by Andreas König :)
315 - save/restore tainted status.
316 - verified to pass the testsuite on my 5.9.5.
317
3183.1 Mon Dec 4 23:03:40 CET 2006
319 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
320 - incompatible change: transfer flags are now per-state.
321 - give a better error message on deadlock.
322 - document Coro::nready.
323 - enhanced testsuite.
324
3253.01 Sun Dec 3 23:47:42 CET 2006
326 - forgot to include Coro::Timer.
327
3283.0 Sun Dec 3 22:57:25 CET 2006
329 - the "FINALLY COMPLETELY STABLE" release (coming soon:
330 the "FAMOUS LAST WORDS" release).
331 - implement a new stack sharing algorithm, which uses a stack
332 pool (size currently hardcoded).
333 - make stack sharing mandatory (it no longer uses a heuristic).
334 - eval/die no longer cause weird problems under heavy use.
335 - Coro::Event could cause livelocks if it was used but
336 no Coro::Event watchers were used.
337 - Coro::Event now uses asynccheck as prepare does not
338 check for changed watchers.
339 - Coro::Event allows multiple waiting coros and will wake up one
340 per event.
341 - Coro::Event should be cleaner and more efficient now.
342 - new utility function Coro::unblock_sub.
343 - document the sad fact that Event is no longer reentrant.
344 - putting a coroutine into the ready queue twice could under
345 some circumstances lead to stack corruption.
346 - minor incompatibility: subclassing Coro::State is supported
347 directly now without going through a _coro_state member.
348 - state/coro switching is much faster now.
349 - very minor optimisations and code/documentation cleanup.
350 - avoid problems due to compiler inlining.
351 - removed timers from Coro::Timer -> use AnyEvent instead.
352 - replaced Coro::idle coroutine by (cleaner) idle handler.
353 - updated to newest libcoro.
354 - implement enhanced support for valgrind.
355 - implement is_ready and return value for ready.
356 - removed Coro::Cont, it was a misnomer (it's generators),
357 and it was somewhat annoying to get it right. Will come back
358 if somebody asks for it :)
359 - many ->wait methods and Coro::Event could return
360 spuriously without the event having happened.
361
3622.5 Tue Nov 7 12:22:33 CET 2006
363 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
364 use AnyEvent, moved them to Coro/.
365 - added Coro::LWP which contains all the uglyness required to
366 make LWP non-blocking.
367 - should work with perl 5.9.x now (Andreas König made me do it).
368 - fixed another bug in Coro::Select when the timeout was undef.
369 - reuse PL_start_env for all coros, saves some memory per coroutine.
370 - manage PL_top_env differently, hopefully to avoid panic: top_env.
371 - timeout argument was not properly used in Coro::Socket.
372 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
373 - emulate undocumented(!) functionality of IO::Socket required
374 by LWP(!!).
375 - updated eg/lwp to work with newer lwp's.
376 - remove "FATAL: uncaught exception" prefix. Coroutines that die
377 kill the whole process, just as exceptions in the main "coroutine"
378 did already.
379
3802.1 Wed Nov 1 23:01:13 CET 2006
381 - fix a long-standing bug in Coro::Select where select with
382 zero timeout would instead change the current default filehandle.
383 - use a simpler and hopefully more robust way to clone padlists
384 (uses less memory and a perl function instead of our own).
385 - coro can now create a stack guard area on many architectures.
386 - Coro::AIO properly reexports additional functions from IO::AIO.
387 - updated libcoro with a workaround for OS X,
388 pach and testing by Michael Schwern.
389
3902.0 Tue Oct 24 05:47:17 CEST 2006
391 - support additional aio requests in Coro::AIO.
392
3931.9 (never properly released due to a glitch)
394
3951.8 Thu Feb 2 00:59:06 CET 2006
396 - applied suggested patch by SAMV to avoid problems during stupid
397 mark & sweep gc run.
398 - applied patch by Scott Walters for 5.9.3 compatibility.
399
4001.7 Tue Dec 27 01:41:58 CET 2005
401 - added Coro::AIO, a thin wrapper around IO::AIO.
402 - improved Makefile.PL explanations.
403
4041.6 Mon Dec 12 21:30:05 CET 2005
405 - additionally save PM_curpm might fix as-of-yet
406 unseen problems with regex matches being attributed to the
407 wrong package.
408 - add t/10_bugs.t, which currently checks against imho broken
409 perls that use 0.26MB of stack space per Perl_magic_get
410 invocation (newer linuxes) as opposed to the 0.0002MB perl
411 normally uses.
412 - make stacksize configurable for the ultimate debian experience.
413
4141.51 Mon Dec 12 18:48:36 CET 2005
415 - remove debugging warn() accidentally left in Coro::Select.
416
4171.5 Tue Nov 29 13:32:44 CET 2005
418 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
419 - fix Coro::Select 'select'.
420 - strict'ify some modules.
421 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
422 - use XSLoader in selected modules.
423 - remove some 5.6 compatibility cruft.
424
4251.4 Tue Sep 6 00:11:05 CEST 2005
426 - libcoro did not take into account the trampoline on amd64,
427 when 'l'inux method was used.
428
4291.31 Tue Aug 30 23:31:33 CEST 2005
430 - some portability fixes/workarounds.
431
4321.3 Sat Aug 20 03:08:56 CEST 2005
433 - no code changes. module seems to work fine.
434 - improve Coro::State docs, remove reference to nonexisting
435 Coro::State::flush method.
436 - no longer autodetect windows, present it as an option instead.
437
4381.2 Mon May 16 02:00:55 CEST 2005
439 - included libcoro.c earlier in State.xs, might improve portability.
440 - use a faster and possibly more stable Coro::Cont implementation.
441 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
442 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
443 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
444
4451.11 Thu Mar 3 18:00:52 CET 2005
446 - change of contact address.
447
4481.1 Tue Feb 22 20:51:16 CET 2005
449 - support [l]inux method on amd64.
450 - allow some context switches while compiling/eval'ing.
2 451
31.0 Tue Aug 10 03:47:27 CEST 2004 4521.0 Tue Aug 10 03:47:27 CEST 2004
4 - Event 0.89 fixes the bug again. 453 - Event 0.89 fixes the bug again.
5 - fix the remaining(??) memleak. I didn't free the cache-AV for 454 - fix the remaining(??) memleak. I didn't free the cache-AV for
6 padlists, so one AV per created anon-sub leaked. 455 padlists, so one AV per created anon-sub leaked.
7 456
80.97 Fri May 14 15:23:32 CEST 2004 4570.97 Fri May 14 15:23:32 CEST 2004
9 - remove stupid gcc extension that was an experiment only anyway. 458 - remove stupid gcc extension that was an experiment only anyway.
10 459
110.96 Thu May 13 18:09:29 CEST 2004 4600.96 Thu May 13 18:09:29 CEST 2004
12 - tell people if they have yet anothe rbroken Event variant. 461 - tell people if they have yet another broken Event variant.
13 - made Coro to work _better_ with threads (it seems that, as long 462 - made Coro to work _better_ with threads (it seems that, as long
14 as you use Coros in one thread only everything will work fine. 463 as you use Coros in one thread only everything will work fine.
15 Looking at the perl source the problem might be missing locking 464 Looking at the perl source the problem might be missing locking
16 between Coro and some perl internal routines. That's not easily 465 between Coro and some perl internal routines. That's not easily
17 fixable). 466 fixable).

Diff Legend

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