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

Comparing Coro/Changes (file contents):
Revision 1.49 by root, Mon Sep 24 00:16:30 2001 UTC vs.
Revision 1.381 by root, Mon Oct 6 10:05:28 2008 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
30.5 3TODO: should explore PerlIO::coroaio.
4 - small tweaks. 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 - improve readline speed for very long "lines".
9
104.8 Thu Oct 2 13:34:40 CEST 2008
11 - new function Coro::AIO::aio_wait.
12 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
13 - greatly speed up and reduce memory usage of Coro::AIO requests.
14 - implement some other µ-optimisations.
15
164.749 Mon Sep 29 14:40:12 CEST 2008
17 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
18
194.748 Sat Sep 27 14:03:19 CEST 2008
20 - implement, but do not document, PerlIO::cede(granularity).
21 - Coro::Storable forgot to wrap Storable::pstore.
22 - work around the multitude of leaks and memory corruption
23 bugs in PerlIO::via by using our own C-level perliol. As a side
24 effect, Coro::Storable is now much, much, much faster.
25
264.747 Tue Sep 23 01:59:41 CEST 2008
27 - fix a per-cv memleak (one empty array was leaked per
28 code reference).
29 - avoid a crash in coro->call|rss when the coroutine was already
30 destroyed (most noticably when using Coro::Debug::ps :)
31 - also protect *Storable::FILE.
32 - push up default storable granularity to 20ms.
33
344.746 Sun Sep 21 03:22:20 CEST 2008
35 - be more insistent on locking Storable against reentrancy
36 in Coro::Storable.
37 - move swap_def?v and throw to Coro::State, as documented.
38
394.745 Thu Jul 24 00:14:38 CEST 2008
40 - remove debugging code related to MgPV_nolen_const, also try to
41 make it compile with perl 5.8.6 (yes, apple apparently loves
42 outdated software). Reported by John S.
43
444.744 Tue Jul 8 22:06:35 CEST 2008
45 - correctly provide default DIE/WARN handlers as documented.
46 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
47 implementations inside perl work fine for us.
48
494.743 Mon Jun 16 00:21:57 CEST 2008
50 - when using Coro::EV without running EV::loop it could
51 result in busy-waiting for events, this has been fixed.
52 - reduce codesize and improve performance by using EV_DEFAULT_UC.
53
544.742 Sat May 31 14:10:21 CEST 2008
55 - implement a workaround for (some) perl <5.8.8 versions.
56 - require EV 3.3+.
57
584.741 Fri May 30 23:33:09 CEST 2008
59 - tell netbsd how utterly broken their imitation of an OS is
60 and refuse to build by default if pthreads are in use.
61 - switch to "s" method on all bsd's by default, as their ucontext
62 stuff seems just too broken.
63 - fix a bug in Coro::Select.
64
654.74 Thu May 29 20:05:31 CEST 2008
66 - do not test Coro::LWP for lack of dependencies.
67 - do not test Coro::Select for lack of working perls.
68
694.73 Thu May 29 2008
70 - fix a bug in Coro::EV which would cause it to block despite
71 there being runnable coroutines.
72 - sprinkle "no warnings" freely over everything, also suppress
73 warnings for some other modules.
74 - fix typo in WSAEWOULDBLOCK.
75
764.72 Sun May 25 05:14:36 CEST 2008
77 - tweak META.yaml a bit, unfortunately, there is no documented way
78 to have optional dependencies with CPAN. doh :(
79 - avoid running some tests on windows because they would fail due to
80 perl bug (broken fork, broken pipes...).
81 - work around perl on windows bugs where perl returns undocumented
82 error codes for sysread, syswrite etc. by taking advantage
83 of AnyEvent's workaround for that problem.
84 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
85 a common perl implementation bug on windows.
86 - use unix domain sockets in testsuite to work around
87 common perl implementation bugs on widows (they are emulated by
88 tcp sockets on windows. ugh.)
89
904.71 Sat May 24 20:01:27 CEST 2008
91 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
92 - take advantage of async name resolution of AnyEvent::Util.
93 - work around brutal inet_aton override in Coro::LWP.
94 - take advantage of the readyhook in Coro::EV, for smoother
95 scheduling.
96
974.7 Sun May 11 00:32:19 CEST 2008
98 - completely reworked the Coro manpage.
99 - added Coro::AnyEvent, generic event loop integration.
100 - implement cancel, ready and kill commands in Coro::Debug.
101 - document find_coro in Coro::Debug.
102 - incompatible change: rename has_stack to has_cctx.
103 - Coro::AIO and Coro::BDB no longer force event model detection,
104 use AnyEvent::AIO and AnyEvent::BDB.
105
1064.6 Sat Apr 26 10:05:14 CEST 2008
107 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
108 by the Coro module. It was a mistake to have it in the first place.
109 - (experimental) support for activestate perl 5.10 (method "w").
110 - (experimental) support for strawberry perl (method "a").
111 - coro_sigelem_set did not return a value although it had to,
112 actual impact unknown.
113
1144.51 Mon Apr 14 13:28:27 CEST 2008
115 - make it compile again on 5.8.
116
1174.50 Thu Apr 10 09:43:17 CEST 2008
118 - I did it twice! (see 4.49).
119
1204.49 Sun Apr 6 21:23:31 CEST 2008
121 - grr, instead of compiling the recent changes on 5.10 only they
122 were compiled on 5.8 only.
123
1244.48 Sun Apr 6 20:36:46 CEST 2008
125 - allow coroutine switches during eval's under 5.10.x, as it seems
126 the parser is a per-interpreter option now, so this is safe
127 (this might fix the odd crashing bug).
128 - drop support for 5.9.x versions: they are dead, jim.
129
1304.47 Sun Apr 6 00:37:52 CEST 2008
131 - force cctx allocation on API calls: we know we need to force one
132 and gcc actually manages to confuse our heuristic nowadays,
133 leading to crashes and worse.
134 - document force_cctx.
135
1364.46 Fri Apr 4 22:05:43 CEST 2008
137 - upgrade libcoro, resulting in pthread-backend (which was only created
138 to fulfill the rules of the programming languages shootout).
139
1404.45 Thu Mar 13 11:55:36 CET 2008
141 - fix a file leaking bug in eg/mhyttpd that would allow
142 downloading of any file (reported by oesi).
143 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
144 (also add testcase).
145 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
146 apparently having loads of fun with that).
147 - implement $state->swap_defsv, ->swap_defav and document ->throw.
148
1494.4 Wed Feb 13 16:44:29 CET 2008
150 - only assume .cfi pseudo isns exist on GNU/Linux.
151 - add get_timed method to Coro::Channels.
152 - fixed Coro::Debug manpage.
153 - perl 5.11 compatbility improvement based on patch by
154 Andreas König.
155
1564.37 Sun Jan 20 11:25:23 CET 2008
157 - soften the check for an active parser for perl 5.10 (avoiding
158 "Coro::State::transfer called while parsing" in many cases).
159
1604.36 Sun Jan 13 10:53:56 CET 2008
161 - reset diehook when terminating from an async_pool as to not
162 trigger any __DIE__ handlers.
163
1644.35 Sun Jan 13 04:14:13 CET 2008
165 - "bt" debug command now displays any exceptions
166 from longmess and also skips the innermost
167 stackframes, giving more useufl output.
168 - allow backtraces in coroutines blocked in Coro::EV,
169 at a <1% speed hit.
170 - handle localising of $SIG{__WARN__} and __DIE__
171 properly (with a proper amount of dirty hacking).
172
1734.34 Sat Dec 22 17:49:53 CET 2007
174 - upgrade to EV version 2.0 API.
175
1764.33 Mon Dec 17 08:36:12 CET 2007
177 - make Coro::AIO etc. loadable in the absence of EV.
178
1794.32 Mon Dec 17 07:46:02 CET 2007
180 - majorly improved Coro::Handle's performance with Coro::EV.
181 - implemented the until now mythical Coro::BDB module.
182 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
183
1844.31 Wed Dec 5 12:32:39 CET 2007
185 - remove warn statement form Coro::Util.
186
1874.3 Tue Dec 4 20:33:14 CET 2007
188 - calls to the idle function could cause stack corruption
189 when the stack changed.
190 - do no longer rely on the presence of EV::DNS if EV is used
191 (because it is gone), but instead take avdantage of EV::ADNS
192 if available.
193 - add ($) prototypes to all functions in Coro::Storable.
194 - use a conventional (safer) idle callback in Coro::EV.
195 - do accept quickly in Coro::Debug to avoid endless loops.
196
1974.22 Fri Nov 30 16:04:04 CET 2007
198 - really use optimised versions for Event and EV in Coro::Util
199 and Coro::Handle.
200
2014.21 Sun Nov 25 10:48:59 CET 2007
202 - fix a spurious memory read.
203 - Coro::EV no longer keeps the eventloop "alive".
204
2054.2 Fri Nov 9 20:47:05 CET 2007
206 - enable/disable tracing from a new coroutine, not a pooled one.
207 - fix a memleak in Coro::Event.
208 - removed killall call from fork_eval.
209 - made sure store_fd is already loaded so that fork_eval
210 does not have to parse autoload in each subprocess.
211 - only use assembly method if -O switch is in $Config{optimize}.
212 - add (optional) Coro::EV module, so far the best event loop module
213 directly supported by Coro.
214 - if the event model is EV, use EV::DNS to resolve
215 stuff in Coro::Util.
216 - don't get confused by multiple event notifications in Coro::Handle.
217 - initial support for EV (libevent interface).
218 - require Event and EV using configure_requires, to force their existance.
219
2204.13 Wed Oct 24 07:26:45 CEST 2007
221 - add Coro::Storable::blocking_thaw.
222 - use a vastly more complicated technique to localise
223 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
224 - use a coroutine for the idle callback Coro::Event,
225 instead of running Event in the current coroutine context.
226 This also catches recursive invocations.
227 - actually report fork errors in gethostbyname and inet_aton.
228
2294.11 Thu Oct 11 02:40:24 CEST 2007
230 - port to threaded perls.
231
2324.1 Thu Oct 11 02:38:16 CEST 2007
233 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
234 be local to each coro (see Coro::State).
235 - incompatible change: for very deep reasons, cede and cede_notself
236 cannot return anything, so nothing will be returned.
237 - possibly bring back 5.10 compatibility (untested).
238 - work around stupid (and wrong) warning on 5.10 :(.
239 - overlay the saved state over the context stack. This saves
240 a few hundred bytes per coroutine on average and also
241 speeds up context switching a bit.
242 - further tune default stack sizes.
243 - (more) correctly calculate stack usage in coro_rss.
244 - Coro::Storable::blocking_* did not properly lock
245 resulting in races between coroutines.
246 - added Coro::Storable::guard.
247 - stopping to trace a coroutine could destroy the cctx of
248 an unrelated coroutine.
249 - explain the relationship between Perl and C coroutines in
250 more detail in Coro::State.
251 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
252 causing a fork per resolve. This also affected Coro::Socket.
253 - switch to a separate stack in $coro->call/eval to avoid
254 invalidating pointers.
255
2564.03 Sat Oct 6 21:24:00 CEST 2007
257 - added Coro::throw method.
258 - minor code cleanups.
259
2604.02 Sat Oct 6 02:36:47 CEST 2007
261 - fix a very minor per-coroutine memleak (a single codereference).
262 - fixed a bug where the currently in-use c context would be freed
263 prematurely (can happen only when programs change the stacksize
264 or use tracing).
265 - tracing can no longer keep a coro alive after it terminated.
266 - do static branch prediction in the common path for gcc. gives
267 about 2-5% speed improvement here.
268
2694.01 Fri Oct 5 22:10:49 CEST 2007
270 - instead of recreating *a* standard output handle we simply
271 use STDOUT, which is faster and hopefully more robust.
272
2734.0 Fri Oct 5 12:56:00 CEST 2007
274 - incompatibly changed Coro::Storable::freeze.
275 - major new feature: added Coro::Debug, for interactive coroutine
276 debugging, tracing and much more.
277 - major bug fix: unbelievable, but true: $_, $/ and many other
278 "saved" variables actually weren't being saved. This has been fixed,
279 of course, while increasing performance while losing all the save
280 flags.
281 - save flags are gone, and all the api functions dealing with them.
282 - added Coro::Semaphore::adjust.
283 - added Coro::Util::fork_eval.
284 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
285 - added Coro::killall.
286 - reduce initial stack sizes to allow for "micro-coroutines".
287 - better async_pool resource management, moved parts of async_pool
288 handling to XS (major speed improvement).
289 - actually croak before modifying important data structures.
290 - refuse to transfer while compiling.
291 - possibly support eval EXPR better now.
292 - enable assembly per default on linux+bsd x86+amd64.
293 - all internal members were renamed _something for easier subclassing.
294 - many minor tweaks.
295
2963.63 Wed May 16 14:10:06 CEST 2007
297 - implement handcoded assembly for x86/amd64 SVR ABI.
298
2993.62 Fri Apr 27 21:36:06 CEST 2007
300 - upgrade libcoro (which might set unwind info correctly).
301 - change default on linux to setjmp/longjmp.
302
3033.61 Thu Apr 19 12:36:18 CEST 2007
304 - Coro::Storable caused an endless loop when thawing invalid
305 pst's sometimes.
306 - use a Semaphore in Coro::Storable, as Storable doesn't
307 seem to be reentrant (although it is documented to
308 be threadsafe...).
309 - fix Coro::Signal to bring back the original unreliable
310 but stateful semantics.
311 - fixed a lot of typos in Coro.pm (patch submitted by David
312 Steinbrunner, which applied flawlessly).
313
3143.6 Sat Apr 14 17:13:31 CEST 2007
315 - added some bugfixes to get eg/myhttpd working again.
316 - added Coro::Storable for often-cede'ing freeze/thaw.
317 - try to do a clean exit when a coroutine calls exit
318 (EXPERIMENTAL).
319 - got rid of indirect call through _coro_init.
320 - updated the partly antique examples in eg/ to
321 work again and be a bit less magic, too.
322 - fixed Coro::Signal semantics to work as documented again.
323
3243.55 Sun Mar 25 01:20:47 CET 2007
325 - add SAVE_DEFFH to save the default filehandle and enable
326 it by default.
327 - finally move socket-operations from Coro::Socket to Coro::Handle
328 to be able to unblock foreign sockets.
329 - add Coro::State::save_also and guarded_save.
330 - add count accessor to Coro::Semaphore.
331 - add Coro::State::cctx_stacksize.
332 - just for the fun of it, do not rely on implicit context,
333 which can dramatically improve performance, but people
334 using windows-process-emulation perls do not care much about
335 performance.
336
3373.51 Sun Mar 4 14:18:04 CET 2007
338 - fixed a problem when you weakened references to Coro::State's
339 (patch by Yuval Kogman).
340
3413.501 Wed Feb 28 12:44:07 CET 2007
342 - rename some global symbols as macosx from hell redefines
343 them without asking.
344
3453.5 Tue Feb 13 20:22:53 CET 2007
346 - do AnyEvent model detection earlier, avoiding problems
347 caused by first using AnyEvent and later Coro::Event.
348 - implement and document Coro::Event event objects.
349 - fix a potential problem in Coro::Event causing crashes.
350 - initialise PL_comppad when creating a new coroutine,
351 avoids crashes on early coro destruction.
352
3533.41 Mon Jan 22 19:19:49 CET 2007
354 - readline on Coro::Handle did not support undefined $/,
355 nor did it deliver partial lines on EOF or error.
356 - implement malloc fallback for stack allocation because
357 stupid broken idiotic OSX has a stupid broken
358 idiotic fits-the-whole-os mmap "implementation" and
359 my dick feels longer if Coro is portable even to
360 obsolete platforms.
361
3623.4 Fri Jan 19 21:52:54 CET 2007
363 - remove t/09_timer.t, as it isn't really testing much
364 but was rather flaky in practise.
365 - async_pool coro would keep arguments and callback alive until
366 it was reused.
367 - cancellation of a coroutine could cause spurious idle calls
368 in cede_notself.
369
3703.3 Sat Jan 6 03:45:00 CET 2007
371 - implement $coro->on_destroy.
372 - Coro::Event blocking semantics have been changed,
373 documented and - hopefully - improved.
374 - fix nice adding, not subtracting, from priority.
375 - fix ->prio and api_is_ready (patch by Mark Hinds).
376 - fixed an assert ("... == prev__cctx->idle_te")
377 that could errronously trigger.
378 - fix various large and small memleaks.
379 - use a (hopefully) more stable cancel implementation
380 that immediately frees the coroutine data.
381 - cede/cede_notself return a status now.
382 - added Coro::guard function.
383 - added a global coroutine pool for jobs (on my machine,
384 I can create and execute 48k simple coros/s with async,
385 and 128k coros with async_pool).
386 - Coro::AIO now uses the coroutine priority as io priority.
387
3883.2 Fri Dec 22 05:07:09 CET 2006
389 - improve portability to slightly older perls.
390 - use cleaner coroutine destruction.
391 - simplify configuration for users.
392 - optionally (unrecommended) prefer perl functions over
393 their coro replacements.
394
3953.11 Tue Dec 5 13:11:24 CET 2006
396 - fixed some bogus assert's, but as perl.h disables assert even
397 without NDEBUG (thank you very much), not too many people should
398 notice (that did include myself). Andreas König noticed, though :)
399 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
400 it. Also noticed by Andreas König :)
401 - save/restore tainted status.
402 - verified to pass the testsuite on my 5.9.5.
403
4043.1 Mon Dec 4 23:03:40 CET 2006
405 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
406 - incompatible change: transfer flags are now per-state.
407 - give a better error message on deadlock.
408 - document Coro::nready.
409 - enhanced testsuite.
410
4113.01 Sun Dec 3 23:47:42 CET 2006
412 - forgot to include Coro::Timer.
413
4143.0 Sun Dec 3 22:57:25 CET 2006
415 - the "FINALLY COMPLETELY STABLE" release (coming soon:
416 the "FAMOUS LAST WORDS" release).
417 - implement a new stack sharing algorithm, which uses a stack
418 pool (size currently hardcoded).
419 - make stack sharing mandatory (it no longer uses a heuristic).
420 - eval/die no longer cause weird problems under heavy use.
421 - Coro::Event could cause livelocks if it was used but
422 no Coro::Event watchers were used.
423 - Coro::Event now uses asynccheck as prepare does not
424 check for changed watchers.
425 - Coro::Event allows multiple waiting coros and will wake up one
426 per event.
427 - Coro::Event should be cleaner and more efficient now.
428 - new utility function Coro::unblock_sub.
429 - document the sad fact that Event is no longer reentrant.
430 - putting a coroutine into the ready queue twice could under
431 some circumstances lead to stack corruption.
432 - minor incompatibility: subclassing Coro::State is supported
433 directly now without going through a _coro_state member.
434 - state/coro switching is much faster now.
435 - very minor optimisations and code/documentation cleanup.
436 - avoid problems due to compiler inlining.
437 - removed timers from Coro::Timer -> use AnyEvent instead.
438 - replaced Coro::idle coroutine by (cleaner) idle handler.
439 - updated to newest libcoro.
440 - implement enhanced support for valgrind.
441 - implement is_ready and return value for ready.
442 - removed Coro::Cont, it was a misnomer (it's generators),
443 and it was somewhat annoying to get it right. Will come back
444 if somebody asks for it :)
445 - many ->wait methods and Coro::Event could return
446 spuriously without the event having happened.
447
4482.5 Tue Nov 7 12:22:33 CET 2006
449 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
450 use AnyEvent, moved them to Coro/.
451 - added Coro::LWP which contains all the uglyness required to
452 make LWP non-blocking.
453 - should work with perl 5.9.x now (Andreas König made me do it).
454 - fixed another bug in Coro::Select when the timeout was undef.
455 - reuse PL_start_env for all coros, saves some memory per coroutine.
456 - manage PL_top_env differently, hopefully to avoid panic: top_env.
457 - timeout argument was not properly used in Coro::Socket.
458 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
459 - emulate undocumented(!) functionality of IO::Socket required
460 by LWP(!!).
461 - updated eg/lwp to work with newer lwp's.
462 - remove "FATAL: uncaught exception" prefix. Coroutines that die
463 kill the whole process, just as exceptions in the main "coroutine"
464 did already.
465
4662.1 Wed Nov 1 23:01:13 CET 2006
467 - fix a long-standing bug in Coro::Select where select with
468 zero timeout would instead change the current default filehandle.
469 - use a simpler and hopefully more robust way to clone padlists
470 (uses less memory and a perl function instead of our own).
471 - coro can now create a stack guard area on many architectures.
472 - Coro::AIO properly reexports additional functions from IO::AIO.
473 - updated libcoro with a workaround for OS X,
474 pach and testing by Michael Schwern.
475
4762.0 Tue Oct 24 05:47:17 CEST 2006
477 - support additional aio requests in Coro::AIO.
478
4791.9 (never properly released due to a glitch)
480
4811.8 Thu Feb 2 00:59:06 CET 2006
482 - applied suggested patch by SAMV to avoid problems during stupid
483 mark & sweep gc run.
484 - applied patch by Scott Walters for 5.9.3 compatibility.
485
4861.7 Tue Dec 27 01:41:58 CET 2005
487 - added Coro::AIO, a thin wrapper around IO::AIO.
488 - improved Makefile.PL explanations.
489
4901.6 Mon Dec 12 21:30:05 CET 2005
491 - additionally save PM_curpm might fix as-of-yet
492 unseen problems with regex matches being attributed to the
493 wrong package.
494 - add t/10_bugs.t, which currently checks against imho broken
495 perls that use 0.26MB of stack space per Perl_magic_get
496 invocation (newer linuxes) as opposed to the 0.0002MB perl
497 normally uses.
498 - make stacksize configurable for the ultimate debian experience.
499
5001.51 Mon Dec 12 18:48:36 CET 2005
501 - remove debugging warn() accidentally left in Coro::Select.
502
5031.5 Tue Nov 29 13:32:44 CET 2005
504 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
505 - fix Coro::Select 'select'.
506 - strict'ify some modules.
507 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
508 - use XSLoader in selected modules.
509 - remove some 5.6 compatibility cruft.
510
5111.4 Tue Sep 6 00:11:05 CEST 2005
512 - libcoro did not take into account the trampoline on amd64,
513 when 'l'inux method was used.
514
5151.31 Tue Aug 30 23:31:33 CEST 2005
516 - some portability fixes/workarounds.
517
5181.3 Sat Aug 20 03:08:56 CEST 2005
519 - no code changes. module seems to work fine.
520 - improve Coro::State docs, remove reference to nonexisting
521 Coro::State::flush method.
522 - no longer autodetect windows, present it as an option instead.
523
5241.2 Mon May 16 02:00:55 CEST 2005
525 - included libcoro.c earlier in State.xs, might improve portability.
526 - use a faster and possibly more stable Coro::Cont implementation.
527 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
528 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
529 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
530
5311.11 Thu Mar 3 18:00:52 CET 2005
532 - change of contact address.
533
5341.1 Tue Feb 22 20:51:16 CET 2005
535 - support [l]inux method on amd64.
536 - allow some context switches while compiling/eval'ing.
537
5381.0 Tue Aug 10 03:47:27 CEST 2004
539 - Event 0.89 fixes the bug again.
540 - fix the remaining(??) memleak. I didn't free the cache-AV for
541 padlists, so one AV per created anon-sub leaked.
542
5430.97 Fri May 14 15:23:32 CEST 2004
544 - remove stupid gcc extension that was an experiment only anyway.
545
5460.96 Thu May 13 18:09:29 CEST 2004
547 - tell people if they have yet another broken Event variant.
548 - made Coro to work _better_ with threads (it seems that, as long
549 as you use Coros in one thread only everything will work fine.
550 Looking at the perl source the problem might be missing locking
551 between Coro and some perl internal routines. That's not easily
552 fixable).
553 - (hopefully) got rid of the static xnext inside transfer.
554
5550.95 Sun Feb 15 03:20:28 CET 2004
556 - removed Coro::State::flush, implemented a new and much
557 faster caching mechanism by attaching magic to the CV.
558 - a memleak with (real) closures remains.
559 - dramatically reduced size of stacks to 65536 (linux/x86), after
560 observing that even myhttpd never had more than a kilobyte
561 of stack in use (stack is still only physically allocated in
562 pagesize increments for systems suporting mmap).
563 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
564 - fix for threaded perl.
565
5660.9 Sun Nov 23 04:32:39 CET 2003
567 - fix a few very hard to track down but (of course) stupid
568 memory leaks.
569 - stringify version number to avoid locale problems :(.
570 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
571 hopefully ignoring it "just works".
572
5730.8 Wed Nov 5 19:38:40 CET 2003
574 - port to perl5.005_03, but only 5.8.x+ is supported!
575 - honor LocalAddr even without LocalPort in Coro::Socket.
576
5770.7 Tue May 27 03:12:38 CEST 2003
578 - the version jump indicates some level of testing, not gobs
579 of new features.
580 - uh... I found the prompt function in ExtUtils::MakeMaker.
581 Highly correct stuff, that is...
582 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
583 the fix is not well understood by the author, so beware :(.
584
5850.652 Thu May 8 02:54:46 CEST 2003
586 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
587 cause version 4 doesn't have ucontext.
588 - Benjamin Reed reported that setjmp works fine on darwin,
589 so preselect it.
590
5910.651 Sat Mar 29 15:00:23 CET 2003
592 - fix a bug in Coro::Handle where some bytes could get lost
593 on reading, reported by jason@nichego.net.
594
5950.65 Sun Mar 23 00:08:26 CET 2003
596 - added a README.linux-glibc.
597 - new module Coro::Select.
598 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
599 (I never claimed that I know what I am doing ;).
600 - set default to setjmp/longjmp on non-x86-linux.
601
6020.6 Thu Nov 21 11:09:06 CET 2002
603 - made lazy context switching the default.
604 - fixed the bug where SemaphoreSet::guard
605 was the same as timed_guard and timed_guard was missing.
606 - fix a memleak in Coro::Timer.
607 - reclassified context sharing as not experimental.
608
6090.534 Sun Apr 14 03:05:12 CEST 2002
610 - fixed backspace => tab in header parsing (myhttpd).
611 - added eg/readline, for elmex.
612 - Coro::Event now calls ->start on first call instead of
613 ->again, which created wrong timeouts for the first ->next.
614 - fixed a bug where Coro::Socket returned a closed socket instead
615 of nothing on ECONNREFUSED and similar error conditions.
616
6170.533 Mon Feb 18 18:49:40 CET 2002
618 - adapt to finally working Event-0.85.
619 - fixed buggy event initialization (trapped under windows, but how
620 could it possibly have worked anywhere?)
621
6220.532 Wed Jan 16 02:45:32 CET 2002
623 - removed the reference to pp_entersub, might work on
624 (some) win32 perls now (testsuite works under cygwin,
625 without Event, which freezes).
626
6270.531 Mon Dec 10 22:18:44 CET 2001
628 - Fixed a bug in SemaphoreSet::timed_down.
629
6300.53 Tue Nov 27 21:11:13 CET 2001
631 - some tests for Event.
632 - slightly different internal architecture to get refcnt into the object
633 and not the reference. might fix a few bugs, certainly did introduce
634 new ones.
635 - Coro::Timer (independent of Event).
636 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
637
6380.52 Tue Nov 6 21:36:18 CET 2001
639 - ported to cygwin (trivial).
640 patch by Gerrit P. Haase <gp@familiehaase.de>.
641 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
642
6430.51 Thu Nov 1 20:39:01 CET 2001
644 - terminate/cancel now work properly (otherwise termination
645 could cause a "next coroutine is not and contains not..."
646 error.
647 - added Coro::Socket::shutdown.
648 - Coro::Event::loop no is the same as Event::loop.
649 - implemented terminate with args + join.
650
6510.5 Fri Sep 28 16:15:35 CEST 2001
652 - fixed "print" on a Coro::Handle. The print method worked.
653 - small tweaks (seem to reduce memory consumption a lot)
654 in various modules.
655 - splendid use of "no warnings" scattered throughout.
656 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
5 657
60.49 Sun Sep 16 02:42:45 CEST 2001 6580.49 Sun Sep 16 02:42:45 CEST 2001
7 - changed some method calls to function calls for speed 659 - changed some method calls to function calls for speed
8 inside Coro::Handle. 660 inside Coro::Handle.
9 - make Coro::Handle use an array instead of a hash for 661 - make Coro::Handle use an array instead of a hash for

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines