ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.530
Committed: Fri Dec 11 18:32:23 2009 UTC (14 years, 5 months ago) by root
Branch: MAIN
Changes since 1.529: +6 -0 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 Revision history for Perl extension Coro.
2
3 TODO: should explore PerlIO::coroaio (perl leaks like hell).
4 TODO: 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 TODO: unready_all
8 TODO: myhttpd header parsing
9 TODO: channel->maxsize(newsize)?
10
11 TODO: better error reporting when blocking inside the event idle loop? idleblock
12 5.21
13 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
14 together.
15 - add some examples on how to combine other event loops with Coro in
16 Coro::AnyEvent.
17 - clarify the "not from signal handlers" section.
18
19 5.2 Sun Oct 4 14:54:24 CEST 2009
20 - Coro::Storable destroyed the prototypes of the functions it wrapped.
21 - export rouse_cb and rouse_wait by default now.
22 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
23 - new method $state->swap_sv.
24 - added section on "windows process emulation" to the manpage,
25 after a not-so-fruitful (nor-friendly) "discussion" with chip
26 salzenberg (discussion implies arguments, but his only arguments
27 were ad-hominems, one wonders why he started it in the first
28 place). I hope this explains it well enough for him to understand,
29 and maybe well enough for others to understand.
30 - use common::sense everywhere now.
31 - idle callbacks are no longer supported, use idle coros instead.
32 - print a thread listing when a deadlock is detected.
33
34 5.17 Sat Aug 22 23:09:31 CEST 2009
35 - work around a bug in the perl debugger causing crashes
36 when running under the debugger by marking _pool_handler as nodebug.
37 - speed up Coro::async considerably.
38 - try some hacks to get netbsd to work "more often" - their broken
39 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
40 - convert Coro to AE 5.0 API.
41
42 5.162 Tue Jul 28 04:04:03 CEST 2009
43 - perl 5.8.2 is now minimum requirement.
44 - skip t/19_handle.t on broken windows perls.
45
46 5.161 Wed Jul 22 04:47:38 CEST 2009
47 - Coro::AnyEvent::poll could have a different prototype when EV was
48 used as backend (analyzed by Tatsuhiko Miyagawa).
49 - Coro::AnyEvent errornously initialised the event loop when loaded,
50 not on demand.
51 - try to workaround rare */t/01_unblock.t failures.
52
53 5.16 Tue Jul 21 01:44:37 CEST 2009
54 - Coro::AnyEvent failed to hook into the event loop
55 when no threads had been readied between detecting
56 the event loop and actually running it.
57 - considerably speed up Coro::Select by taking avdantage
58 of AnyEvent > 4.8 and some other optimisations.
59 - implement paragraph readline mode in Coro::Handle
60 (based on patches by Zsbán Ambrus).
61 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
62 and analyzed by Yasuhiro MATSUMOTO).
63 - clarified libcoro license and copyright.
64 - someone stole my EXTRA_META!!!
65 - implement Coro::Select::patch_pp_sselect and it's brother,
66 for hardcode select overriding.
67
68 5.151 Mon Jul 6 05:41:57 CEST 2009
69 - backport to windows process emulation code again (patch by
70 Yasuhiro MATSUMOTO).
71 - slightly update Coro::MakeMaker.
72
73 5.15 Tue Jun 30 10:28:06 CEST 2009
74 - deprecate Coro::Socket, document how to get ipv6 support via
75 AnyEvent::Socket instead.
76 - implement signal->wait ($cb) interface, similar to semaphores.
77 - work around SvOK not supporting getmagic, so we have to getmagic
78 to test for undef :/ (reported by Matthias Waldorf).
79 - load Coro::AnyEvent in all modules using AnyEvent.
80 - work around perl corrupting our internal data structures,
81 reported by Tokuhiro Matsuno.
82 - enable per-coroutine real and cpu time gathering
83 (Coro::State::enable_times).
84
85 5.14 Wed Jun 24 01:37:48 CEST 2009
86 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
87 as this seems to have been a difficult concept (poll, sleep, idle,
88 idle_upto).
89 - add Coro::AnyEvent::readable/writable functions.
90 - clarify Coro::EV/Event/AnyEvent manpages.
91 - free per-thread global scalars in the thread calling ->cancel, to
92 avoid crashes when $_, $@ etc., are magical but some of those
93 had already been freed.
94 - "unexperimentalise" the callback interface for Coro::Semaphore.
95 - speed up ready queue management/context switching by using a linked
96 list instead of an array (~5-10%).
97 - implement "watch" command in Coro::Debug shells.
98 - for fun, implement time-slicing as an example in the manpage.
99 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
100 Coro::Event, use the normal AnyEvent handling instead of dieing
101 (the same is true for Coro::Handle).
102 - properly document Coro::EV::timed_io_once.
103 - avoid unneccessary ->cancel calls in Coro::Handle.
104 - maybe make it work on mingw32 with win32 backend
105 (based on patch by Yasuhiro Matsumoto).
106
107 5.132 Fri May 29 09:00:39 CEST 2009
108 - do not keep a reference to the argument itself in
109 Coro::Semaphore::guard, as it could change later.
110 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
111 should support a prepare callback.
112
113 5.131 Mon Mar 16 23:20:37 CET 2009
114 - implement and document Coro->suspend, Coro->resume.
115 - fix Coro::Select implementation to not (often) close
116 the passed file descriptors (testcase provided by pippijn).
117
118 5.13 Mon Dec 15 21:51:42 CET 2008
119 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
120 - don't set diehook to C<undef> but instead to NULL, to avoid
121 spurious warnings.
122 - fix a lot of bugs in Coro::SemaphoreSet.
123 - Coro::SemaphoreSet will less often create a semaphore needlessly.
124 - add Coro::SemaphoreSet::count and wait methods.
125 - take advantage of the new Guard module.
126 - deprecate Coro::guard.
127 - try to fix the dreaded 01_unblock tests once more. I hate it when
128 testsuites need more fixing than the code they are supposed to test.
129 - croak in more cases when a required callback isn't resolvable.
130 - fix some minor issues in Coro::State->call/eval.
131 - use current coroutine context instead of a temporary one
132 when temporarily switching to another coroutine.
133 - do not call C-level on_destroy handlers during global destruction,
134 to avoid needless segfaults.
135
136 5.12 Sun Dec 7 13:30:38 CET 2008
137 - add default config for MirOS, which seems to be a bug-to-bug
138 compatible fork of openbsd ("world domination by releasing
139 openbsd cvs before the openbsd folks do it" or so :).
140 - free_padlist did destroy the names pad, not good, but didn't
141 seem to bother perl - this could fix issues such as eval ""
142 inside a function called from multiple coroutines.
143 - use a different method to detect destruction time.
144 - be more careful when freeing padlists just before global
145 destruction.
146 - fixed and expanded the call/cc example.
147 - renamed _terminate to _coro_run.
148 - new method Coro::Channel->shutdown.
149 - try pthreads on openbsd <4.4 (broken sigaltstack, will
150 pthreads fare better?).
151 - be less picky about destroying "a" running coroutine.
152
153 5.11 Tue Nov 25 21:49:05 CET 2008
154 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
155 working code with failed assertions, introducing more bugs than
156 it fixes, requiring elaborate workarounds :(
157
158 5.1 Mon Nov 24 08:54:59 CET 2008
159 - wrote a small introductory tutorial - Coro::Intro.
160 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
161 - Coro::Select did errornously dup the file descriptors
162 and didn't work with all AnyEvent backends.
163 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
164 LWP data transfers.
165 - disassociate c contexts from coro objects - this is agruably more
166 correct, but mostly allows sharing of cctxs between coro and state
167 objects, for added memory savings and speed increases.
168 - bumped $Coro::POOL_RSS up to 32kb by default.
169 - no longer set the optype to OP_CUSTOM, as B::* understandably
170 doesn't like this very much (and we *are* a type of entersub).
171 - implement state cloning, just to prove that call/cc can be done.
172 - automatically load Coro::AnyEvent in Coro::Handle.
173 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
174 do not have this method (and don't need it either).
175 - speed up generic anyevent methods in Coro::Handle by using rouse
176 callbacks.
177 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
178 others. It also makes the debugger happier, as you can trace
179 through the idle threads now.
180 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
181 - eg/event was pretty broken.
182 - better 5.8.6 compatibility.
183
184 5.0 Thu Nov 20 10:35:05 CET 2008
185 - NEW ARCHITECTURE: use the latest 4.x version if you experience
186 stability issues.
187 - bump API version to 7 - all dependents must be recompiled.
188 - removed timed_* functions - they were not being used anyways
189 and should be replaced by a more generic mechanism -
190 and were annoying to support anyways :)
191 - removed SemaphoreSet's waiter method - use sem method instead.
192 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
193 - async_pool did free a scalar value twice
194 ("Attempt to unreference...").
195 - fix a longstanding bug where calling terminate on a coro that
196 was waiting for a semaphore that was just becoming available
197 would cause a deadlock (semaphore would get into a state where
198 it was available but waiters were still blocked).
199 - calling throw on a coroutine that is waiting for a semaphore will
200 no longer make it acquire the semaphore (and thus leak a count).
201 - perl's process emulation is now not even theoretically supported
202 anymore.
203 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
204 conversion of callback-style to blocking-style.
205 - new methods $coro->schedule_to and ->cede_to, to specifically
206 schedule or cede to a specific coroutine.
207 - new function Coro::Semaphore::wait.
208 - use named constants in Coro::Channel (Richard Hundt).
209 - directly patch the entersub opcode calling SLF functions (cede,
210 transfer and so on). this does speed up context switching, but
211 more importanly, it frees us from the hardcoded behaviour of
212 entersub, so we might actually be able to return something from
213 those functions and atcually create new ones.
214 - take advantage of __builtin_frame_address on gcc.
215 - expose THX in coroapi (not sure whether this was a wise decision,
216 as "threaded" perls are running at half speed anyways).
217 - implement execute_slf (schedule-like-function) interface that makes
218 it possible to implement schedule-like-functions in XS.
219 - use new SLF interface to massively speed up Coro::EV by roughly a
220 factor of two.
221 - used new SLF interface to massively speed up Coro::Semaphore by a
222 factor of three.
223 - used new SLF interface to speed up Coro::AIO by roughly a factor of
224 four and reduce its memory usage considerably.
225 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
226 for a nice speedup and vastly more correct behaviour. Also implement
227 a new method "sem" to get at the underlying semaphore object.
228 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
229 (in comparison) 20-40% speedup.
230 - used new SLF interface to reimplement Coro::Signal gaining
231 some unknown (because I was too lazy), but certain, speedup, and also
232 making signals reliable for the first time.
233 - used new SLF interface and other optimisations to speed up async_pool
234 by a factor of two. It also doesn't rely on perl's exception mechanism
235 to exit anymore. The overhead for terminating an async_pool, coro over
236 a normal async is now very small.
237 - sped up coroutine creation/destruction by 40%.
238 - forgot to include Coro/libcoro/README in the dist for all these years.
239 - work around a freebsd pthreads bug (manual testcancel is required as
240 pthread_cond_wait isn't a cancellation point on freebsd).
241 - use new rouse functions to speed up and simplify Coro::BDB.
242 - make "prefer perl native functions" work with threaded perls.
243 - condense Coro::Debug ps output, hint at v and w flags.
244 - (libcoro) lots of minor cleanups and portability improvements.
245
246 4.914 Wed Nov 19 12:54:18 CET 2008
247 - fix a disastrous bug in the readline optimisation
248 introduced in 4.801.
249
250 4.913 Sat Nov 15 07:58:28 CET 2008
251 - async_pool did free a scalar value twice
252 ("Attempt to unreference...").
253
254 4.912 Thu Nov 13 18:31:23 CET 2008
255 - minor cleanups.
256 - use much larger stacks on linux and perl < 5.8.8.
257 - Coro::Debug::new_unix_server did not unlink the socket
258 when destroyed.
259
260 4.911 Tue Nov 11 04:26:17 CET 2008
261 - "port" to threaded perls.
262
263 4.91 Mon Nov 10 05:36:38 CET 2008
264 - the ->throw exception object no longer leaks.
265 - creating a new cctx leaked a scopestack entry (memleak).
266 - new coroutines didn't get created with a zero flags field
267 (unknown impact).
268 - calling ->throw on a not-yet-started coroutine should now work
269 instead of being ignored.
270 - ->throw is now supported on Coro::State objects.
271 - clean up cctx creation code a bit.
272 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
273
274 4.9 Sat Nov 8 17:45:27 CET 2008
275 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
276 - (libcoro) no longer leak threads in the experimental pthread backend,
277 also speed it up considerably.
278 - (libcoro) do not rely on makecontext passing void *'s unscathed.
279 - fix compiletime dependencies on libcoro in the Makefile.
280 - cctx_count wasn't always updated properly.
281 - Coro::State::cctx_stacksize wasn't applied correctly.
282 - new function Coro::State::cctx_max_idle.
283 - the default max number of idle C contexts is now 4.
284 - (libcoro) try harder to get _setjmp/_longjmp.
285 - (libcoro) cleanup and extend the libcoro API to officially
286 allow the creation of empty source contexts.
287 - very experimental workaround for the totally broken netbsd platform.
288 - tried to hack around openbsd bugs.
289
290 4.804 Wed Nov 5 16:36:00 CET 2008
291 - Coro::Debug::new_unix_server would not create a non-blocking listening
292 socket, sometimes causing freezes.
293 - (libcoro) fix misaligned stack points for setjmp and assembly
294 methods, which can cause crashes on x86/x86_64 with a sufficiently
295 aggressive compiler.
296 - new function: Coro::Debug::new_tcp_server.
297 - move ->throw into the Coro class because it only works on coro objects.
298
299 4.803 Mon Nov 3 17:16:12 CET 2008
300 - (libcoro) use a global asm statement to become independent of gcc
301 otpimisations for CORO_ASM (thanks to pippijn for the idea).
302 - try to workaround yet another broken bsd, this time dragonfly.
303
304 4.802 Thu Oct 30 10:56:12 CET 2008
305 - support -fno-omit-frame-pointer on x86 with the assembly method.
306 - tune 01_unblock.t tests a bit.
307
308 4.801 Wed Oct 22 18:33:37 CEST 2008
309 - improve readline speed for very long "lines".
310 - backport to 5.8.8.
311
312 4.8 Thu Oct 2 13:34:40 CEST 2008
313 - new function Coro::AIO::aio_wait.
314 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
315 - greatly speed up and reduce memory usage of Coro::AIO requests.
316 - implement some other µ-optimisations.
317
318 4.749 Mon Sep 29 14:40:12 CEST 2008
319 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
320
321 4.748 Sat Sep 27 14:03:19 CEST 2008
322 - implement, but do not document, PerlIO::cede(granularity).
323 - Coro::Storable forgot to wrap Storable::pstore.
324 - work around the multitude of leaks and memory corruption
325 bugs in PerlIO::via by using our own C-level perliol. As a side
326 effect, Coro::Storable is now much, much, much faster.
327
328 4.747 Tue Sep 23 01:59:41 CEST 2008
329 - fix a per-cv memleak (one empty array was leaked per
330 code reference).
331 - avoid a crash in coro->call|rss when the coroutine was already
332 destroyed (most noticably when using Coro::Debug::ps :)
333 - also protect *Storable::FILE.
334 - push up default storable granularity to 20ms.
335
336 4.746 Sun Sep 21 03:22:20 CEST 2008
337 - be more insistent on locking Storable against reentrancy
338 in Coro::Storable.
339 - move swap_def?v and throw to Coro::State, as documented.
340
341 4.745 Thu Jul 24 00:14:38 CEST 2008
342 - remove debugging code related to MgPV_nolen_const, also try to
343 make it compile with perl 5.8.6 (yes, apple apparently loves
344 outdated software). Reported by John S.
345
346 4.744 Tue Jul 8 22:06:35 CEST 2008
347 - correctly provide default DIE/WARN handlers as documented.
348 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
349 implementations inside perl work fine for us.
350
351 4.743 Mon Jun 16 00:21:57 CEST 2008
352 - when using Coro::EV without running EV::loop it could
353 result in busy-waiting for events, this has been fixed.
354 - reduce codesize and improve performance by using EV_DEFAULT_UC.
355
356 4.742 Sat May 31 14:10:21 CEST 2008
357 - implement a workaround for (some) perl <5.8.8 versions.
358 - require EV 3.3+.
359
360 4.741 Fri May 30 23:33:09 CEST 2008
361 - tell netbsd how utterly broken their imitation of an OS is
362 and refuse to build by default if pthreads are in use.
363 - switch to "s" method on all bsd's by default, as their ucontext
364 stuff seems just too broken.
365 - fix a bug in Coro::Select.
366
367 4.74 Thu May 29 20:05:31 CEST 2008
368 - do not test Coro::LWP for lack of dependencies.
369 - do not test Coro::Select for lack of working perls.
370
371 4.73 Thu May 29 2008
372 - fix a bug in Coro::EV which would cause it to block despite
373 there being runnable coroutines.
374 - sprinkle "no warnings" freely over everything, also suppress
375 warnings for some other modules.
376 - fix typo in WSAEWOULDBLOCK.
377
378 4.72 Sun May 25 05:14:36 CEST 2008
379 - tweak META.yaml a bit, unfortunately, there is no documented way
380 to have optional dependencies with CPAN. doh :(
381 - avoid running some tests on windows because they would fail due to
382 perl bug (broken fork, broken pipes...).
383 - work around perl on windows bugs where perl returns undocumented
384 error codes for sysread, syswrite etc. by taking advantage
385 of AnyEvent's workaround for that problem.
386 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
387 a common perl implementation bug on windows.
388 - use unix domain sockets in testsuite to work around
389 common perl implementation bugs on widows (they are emulated by
390 tcp sockets on windows. ugh.)
391
392 4.71 Sat May 24 20:01:27 CEST 2008
393 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
394 - take advantage of async name resolution of AnyEvent::Util.
395 - work around brutal inet_aton override in Coro::LWP.
396 - take advantage of the readyhook in Coro::EV, for smoother
397 scheduling.
398
399 4.7 Sun May 11 00:32:19 CEST 2008
400 - completely reworked the Coro manpage.
401 - added Coro::AnyEvent, generic event loop integration.
402 - implement cancel, ready and kill commands in Coro::Debug.
403 - document find_coro in Coro::Debug.
404 - incompatible change: rename has_stack to has_cctx.
405 - Coro::AIO and Coro::BDB no longer force event model detection,
406 use AnyEvent::AIO and AnyEvent::BDB.
407
408 4.6 Sat Apr 26 10:05:14 CEST 2008
409 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
410 by the Coro module. It was a mistake to have it in the first place.
411 - (experimental) support for activestate perl 5.10 (method "w").
412 - (experimental) support for strawberry perl (method "a").
413 - coro_sigelem_set did not return a value although it had to,
414 actual impact unknown.
415
416 4.51 Mon Apr 14 13:28:27 CEST 2008
417 - make it compile again on 5.8.
418
419 4.50 Thu Apr 10 09:43:17 CEST 2008
420 - I did it twice! (see 4.49).
421
422 4.49 Sun Apr 6 21:23:31 CEST 2008
423 - grr, instead of compiling the recent changes on 5.10 only they
424 were compiled on 5.8 only.
425
426 4.48 Sun Apr 6 20:36:46 CEST 2008
427 - allow coroutine switches during eval's under 5.10.x, as it seems
428 the parser is a per-interpreter option now, so this is safe
429 (this might fix the odd crashing bug).
430 - drop support for 5.9.x versions: they are dead, jim.
431
432 4.47 Sun Apr 6 00:37:52 CEST 2008
433 - force cctx allocation on API calls: we know we need to force one
434 and gcc actually manages to confuse our heuristic nowadays,
435 leading to crashes and worse.
436 - document force_cctx.
437
438 4.46 Fri Apr 4 22:05:43 CEST 2008
439 - upgrade libcoro, resulting in pthread-backend (which was only created
440 to fulfill the rules of the programming languages shootout).
441
442 4.45 Thu Mar 13 11:55:36 CET 2008
443 - fix a file leaking bug in eg/mhyttpd that would allow
444 downloading of any file (reported by oesi).
445 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
446 (also add testcase).
447 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
448 apparently having loads of fun with that).
449 - implement $state->swap_defsv, ->swap_defav and document ->throw.
450
451 4.4 Wed Feb 13 16:44:29 CET 2008
452 - only assume .cfi pseudo isns exist on GNU/Linux.
453 - add get_timed method to Coro::Channels.
454 - fixed Coro::Debug manpage.
455 - perl 5.11 compatbility improvement based on patch by
456 Andreas König.
457
458 4.37 Sun Jan 20 11:25:23 CET 2008
459 - soften the check for an active parser for perl 5.10 (avoiding
460 "Coro::State::transfer called while parsing" in many cases).
461
462 4.36 Sun Jan 13 10:53:56 CET 2008
463 - reset diehook when terminating from an async_pool as to not
464 trigger any __DIE__ handlers.
465
466 4.35 Sun Jan 13 04:14:13 CET 2008
467 - "bt" debug command now displays any exceptions
468 from longmess and also skips the innermost
469 stackframes, giving more useufl output.
470 - allow backtraces in coroutines blocked in Coro::EV,
471 at a <1% speed hit.
472 - handle localising of $SIG{__WARN__} and __DIE__
473 properly (with a proper amount of dirty hacking).
474
475 4.34 Sat Dec 22 17:49:53 CET 2007
476 - upgrade to EV version 2.0 API.
477
478 4.33 Mon Dec 17 08:36:12 CET 2007
479 - make Coro::AIO etc. loadable in the absence of EV.
480
481 4.32 Mon Dec 17 07:46:02 CET 2007
482 - majorly improved Coro::Handle's performance with Coro::EV.
483 - implemented the until now mythical Coro::BDB module.
484 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
485
486 4.31 Wed Dec 5 12:32:39 CET 2007
487 - remove warn statement form Coro::Util.
488
489 4.3 Tue Dec 4 20:33:14 CET 2007
490 - calls to the idle function could cause stack corruption
491 when the stack changed.
492 - do no longer rely on the presence of EV::DNS if EV is used
493 (because it is gone), but instead take avdantage of EV::ADNS
494 if available.
495 - add ($) prototypes to all functions in Coro::Storable.
496 - use a conventional (safer) idle callback in Coro::EV.
497 - do accept quickly in Coro::Debug to avoid endless loops.
498
499 4.22 Fri Nov 30 16:04:04 CET 2007
500 - really use optimised versions for Event and EV in Coro::Util
501 and Coro::Handle.
502
503 4.21 Sun Nov 25 10:48:59 CET 2007
504 - fix a spurious memory read.
505 - Coro::EV no longer keeps the eventloop "alive".
506
507 4.2 Fri Nov 9 20:47:05 CET 2007
508 - enable/disable tracing from a new coroutine, not a pooled one.
509 - fix a memleak in Coro::Event.
510 - removed killall call from fork_eval.
511 - made sure store_fd is already loaded so that fork_eval
512 does not have to parse autoload in each subprocess.
513 - only use assembly method if -O switch is in $Config{optimize}.
514 - add (optional) Coro::EV module, so far the best event loop module
515 directly supported by Coro.
516 - if the event model is EV, use EV::DNS to resolve
517 stuff in Coro::Util.
518 - don't get confused by multiple event notifications in Coro::Handle.
519 - initial support for EV (libevent interface).
520 - require Event and EV using configure_requires, to force their existance.
521
522 4.13 Wed Oct 24 07:26:45 CEST 2007
523 - add Coro::Storable::blocking_thaw.
524 - use a vastly more complicated technique to localise
525 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
526 - use a coroutine for the idle callback Coro::Event,
527 instead of running Event in the current coroutine context.
528 This also catches recursive invocations.
529 - actually report fork errors in gethostbyname and inet_aton.
530
531 4.11 Thu Oct 11 02:40:24 CEST 2007
532 - port to threaded perls.
533
534 4.1 Thu Oct 11 02:38:16 CEST 2007
535 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
536 be local to each coro (see Coro::State).
537 - incompatible change: for very deep reasons, cede and cede_notself
538 cannot return anything, so nothing will be returned.
539 - possibly bring back 5.10 compatibility (untested).
540 - work around stupid (and wrong) warning on 5.10 :(.
541 - overlay the saved state over the context stack. This saves
542 a few hundred bytes per coroutine on average and also
543 speeds up context switching a bit.
544 - further tune default stack sizes.
545 - (more) correctly calculate stack usage in coro_rss.
546 - Coro::Storable::blocking_* did not properly lock
547 resulting in races between coroutines.
548 - added Coro::Storable::guard.
549 - stopping to trace a coroutine could destroy the cctx of
550 an unrelated coroutine.
551 - explain the relationship between Perl and C coroutines in
552 more detail in Coro::State.
553 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
554 causing a fork per resolve. This also affected Coro::Socket.
555 - switch to a separate stack in $coro->call/eval to avoid
556 invalidating pointers.
557
558 4.03 Sat Oct 6 21:24:00 CEST 2007
559 - added Coro::throw method.
560 - minor code cleanups.
561
562 4.02 Sat Oct 6 02:36:47 CEST 2007
563 - fix a very minor per-coroutine memleak (a single codereference).
564 - fixed a bug where the currently in-use c context would be freed
565 prematurely (can happen only when programs change the stacksize
566 or use tracing).
567 - tracing can no longer keep a coro alive after it terminated.
568 - do static branch prediction in the common path for gcc. gives
569 about 2-5% speed improvement here.
570
571 4.01 Fri Oct 5 22:10:49 CEST 2007
572 - instead of recreating *a* standard output handle we simply
573 use STDOUT, which is faster and hopefully more robust.
574
575 4.0 Fri Oct 5 12:56:00 CEST 2007
576 - incompatibly changed Coro::Storable::freeze.
577 - major new feature: added Coro::Debug, for interactive coroutine
578 debugging, tracing and much more.
579 - major bug fix: unbelievable, but true: $_, $/ and many other
580 "saved" variables actually weren't being saved. This has been fixed,
581 of course, while increasing performance while losing all the save
582 flags.
583 - save flags are gone, and all the api functions dealing with them.
584 - added Coro::Semaphore::adjust.
585 - added Coro::Util::fork_eval.
586 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
587 - added Coro::killall.
588 - reduce initial stack sizes to allow for "micro-coroutines".
589 - better async_pool resource management, moved parts of async_pool
590 handling to XS (major speed improvement).
591 - actually croak before modifying important data structures.
592 - refuse to transfer while compiling.
593 - possibly support eval EXPR better now.
594 - enable assembly per default on linux+bsd x86+amd64.
595 - all internal members were renamed _something for easier subclassing.
596 - many minor tweaks.
597
598 3.63 Wed May 16 14:10:06 CEST 2007
599 - implement handcoded assembly for x86/amd64 SVR ABI.
600
601 3.62 Fri Apr 27 21:36:06 CEST 2007
602 - upgrade libcoro (which might set unwind info correctly).
603 - change default on linux to setjmp/longjmp.
604
605 3.61 Thu Apr 19 12:36:18 CEST 2007
606 - Coro::Storable caused an endless loop when thawing invalid
607 pst's sometimes.
608 - use a Semaphore in Coro::Storable, as Storable doesn't
609 seem to be reentrant (although it is documented to
610 be threadsafe...).
611 - fix Coro::Signal to bring back the original unreliable
612 but stateful semantics.
613 - fixed a lot of typos in Coro.pm (patch submitted by David
614 Steinbrunner, which applied flawlessly).
615
616 3.6 Sat Apr 14 17:13:31 CEST 2007
617 - added some bugfixes to get eg/myhttpd working again.
618 - added Coro::Storable for often-cede'ing freeze/thaw.
619 - try to do a clean exit when a coroutine calls exit
620 (EXPERIMENTAL).
621 - got rid of indirect call through _coro_init.
622 - updated the partly antique examples in eg/ to
623 work again and be a bit less magic, too.
624 - fixed Coro::Signal semantics to work as documented again.
625
626 3.55 Sun Mar 25 01:20:47 CET 2007
627 - add SAVE_DEFFH to save the default filehandle and enable
628 it by default.
629 - finally move socket-operations from Coro::Socket to Coro::Handle
630 to be able to unblock foreign sockets.
631 - add Coro::State::save_also and guarded_save.
632 - add count accessor to Coro::Semaphore.
633 - add Coro::State::cctx_stacksize.
634 - just for the fun of it, do not rely on implicit context,
635 which can dramatically improve performance, but people
636 using windows-process-emulation perls do not care much about
637 performance.
638
639 3.51 Sun Mar 4 14:18:04 CET 2007
640 - fixed a problem when you weakened references to Coro::State's
641 (patch by Yuval Kogman).
642
643 3.501 Wed Feb 28 12:44:07 CET 2007
644 - rename some global symbols as macosx from hell redefines
645 them without asking.
646
647 3.5 Tue Feb 13 20:22:53 CET 2007
648 - do AnyEvent model detection earlier, avoiding problems
649 caused by first using AnyEvent and later Coro::Event.
650 - implement and document Coro::Event event objects.
651 - fix a potential problem in Coro::Event causing crashes.
652 - initialise PL_comppad when creating a new coroutine,
653 avoids crashes on early coro destruction.
654
655 3.41 Mon Jan 22 19:19:49 CET 2007
656 - readline on Coro::Handle did not support undefined $/,
657 nor did it deliver partial lines on EOF or error.
658 - implement malloc fallback for stack allocation because
659 stupid broken idiotic OSX has a stupid broken
660 idiotic fits-the-whole-os mmap "implementation" and
661 my dick feels longer if Coro is portable even to
662 obsolete platforms.
663
664 3.4 Fri Jan 19 21:52:54 CET 2007
665 - remove t/09_timer.t, as it isn't really testing much
666 but was rather flaky in practise.
667 - async_pool coro would keep arguments and callback alive until
668 it was reused.
669 - cancellation of a coroutine could cause spurious idle calls
670 in cede_notself.
671
672 3.3 Sat Jan 6 03:45:00 CET 2007
673 - implement $coro->on_destroy.
674 - Coro::Event blocking semantics have been changed,
675 documented and - hopefully - improved.
676 - fix nice adding, not subtracting, from priority.
677 - fix ->prio and api_is_ready (patch by Mark Hinds).
678 - fixed an assert ("... == prev__cctx->idle_te")
679 that could errronously trigger.
680 - fix various large and small memleaks.
681 - use a (hopefully) more stable cancel implementation
682 that immediately frees the coroutine data.
683 - cede/cede_notself return a status now.
684 - added Coro::guard function.
685 - added a global coroutine pool for jobs (on my machine,
686 I can create and execute 48k simple coros/s with async,
687 and 128k coros with async_pool).
688 - Coro::AIO now uses the coroutine priority as io priority.
689
690 3.2 Fri Dec 22 05:07:09 CET 2006
691 - improve portability to slightly older perls.
692 - use cleaner coroutine destruction.
693 - simplify configuration for users.
694 - optionally (unrecommended) prefer perl functions over
695 their coro replacements.
696
697 3.11 Tue Dec 5 13:11:24 CET 2006
698 - fixed some bogus assert's, but as perl.h disables assert even
699 without NDEBUG (thank you very much), not too many people should
700 notice (that did include myself). Andreas König noticed, though :)
701 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
702 it. Also noticed by Andreas König :)
703 - save/restore tainted status.
704 - verified to pass the testsuite on my 5.9.5.
705
706 3.1 Mon Dec 4 23:03:40 CET 2006
707 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
708 - incompatible change: transfer flags are now per-state.
709 - give a better error message on deadlock.
710 - document Coro::nready.
711 - enhanced testsuite.
712
713 3.01 Sun Dec 3 23:47:42 CET 2006
714 - forgot to include Coro::Timer.
715
716 3.0 Sun Dec 3 22:57:25 CET 2006
717 - the "FINALLY COMPLETELY STABLE" release (coming soon:
718 the "FAMOUS LAST WORDS" release).
719 - implement a new stack sharing algorithm, which uses a stack
720 pool (size currently hardcoded).
721 - make stack sharing mandatory (it no longer uses a heuristic).
722 - eval/die no longer cause weird problems under heavy use.
723 - Coro::Event could cause livelocks if it was used but
724 no Coro::Event watchers were used.
725 - Coro::Event now uses asynccheck as prepare does not
726 check for changed watchers.
727 - Coro::Event allows multiple waiting coros and will wake up one
728 per event.
729 - Coro::Event should be cleaner and more efficient now.
730 - new utility function Coro::unblock_sub.
731 - document the sad fact that Event is no longer reentrant.
732 - putting a coroutine into the ready queue twice could under
733 some circumstances lead to stack corruption.
734 - minor incompatibility: subclassing Coro::State is supported
735 directly now without going through a _coro_state member.
736 - state/coro switching is much faster now.
737 - very minor optimisations and code/documentation cleanup.
738 - avoid problems due to compiler inlining.
739 - removed timers from Coro::Timer -> use AnyEvent instead.
740 - replaced Coro::idle coroutine by (cleaner) idle handler.
741 - updated to newest libcoro.
742 - implement enhanced support for valgrind.
743 - implement is_ready and return value for ready.
744 - removed Coro::Cont, it was a misnomer (it's generators),
745 and it was somewhat annoying to get it right. Will come back
746 if somebody asks for it :)
747 - many ->wait methods and Coro::Event could return
748 spuriously without the event having happened.
749
750 2.5 Tue Nov 7 12:22:33 CET 2006
751 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
752 use AnyEvent, moved them to Coro/.
753 - added Coro::LWP which contains all the uglyness required to
754 make LWP non-blocking.
755 - should work with perl 5.9.x now (Andreas König made me do it).
756 - fixed another bug in Coro::Select when the timeout was undef.
757 - reuse PL_start_env for all coros, saves some memory per coroutine.
758 - manage PL_top_env differently, hopefully to avoid panic: top_env.
759 - timeout argument was not properly used in Coro::Socket.
760 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
761 - emulate undocumented(!) functionality of IO::Socket required
762 by LWP(!!).
763 - updated eg/lwp to work with newer lwp's.
764 - remove "FATAL: uncaught exception" prefix. Coroutines that die
765 kill the whole process, just as exceptions in the main "coroutine"
766 did already.
767
768 2.1 Wed Nov 1 23:01:13 CET 2006
769 - fix a long-standing bug in Coro::Select where select with
770 zero timeout would instead change the current default filehandle.
771 - use a simpler and hopefully more robust way to clone padlists
772 (uses less memory and a perl function instead of our own).
773 - coro can now create a stack guard area on many architectures.
774 - Coro::AIO properly reexports additional functions from IO::AIO.
775 - updated libcoro with a workaround for OS X,
776 pach and testing by Michael Schwern.
777
778 2.0 Tue Oct 24 05:47:17 CEST 2006
779 - support additional aio requests in Coro::AIO.
780
781 1.9 (never properly released due to a glitch)
782
783 1.8 Thu Feb 2 00:59:06 CET 2006
784 - applied suggested patch by SAMV to avoid problems during stupid
785 mark & sweep gc run.
786 - applied patch by Scott Walters for 5.9.3 compatibility.
787
788 1.7 Tue Dec 27 01:41:58 CET 2005
789 - added Coro::AIO, a thin wrapper around IO::AIO.
790 - improved Makefile.PL explanations.
791
792 1.6 Mon Dec 12 21:30:05 CET 2005
793 - additionally save PM_curpm might fix as-of-yet
794 unseen problems with regex matches being attributed to the
795 wrong package.
796 - add t/10_bugs.t, which currently checks against imho broken
797 perls that use 0.26MB of stack space per Perl_magic_get
798 invocation (newer linuxes) as opposed to the 0.0002MB perl
799 normally uses.
800 - make stacksize configurable for the ultimate debian experience.
801
802 1.51 Mon Dec 12 18:48:36 CET 2005
803 - remove debugging warn() accidentally left in Coro::Select.
804
805 1.5 Tue Nov 29 13:32:44 CET 2005
806 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
807 - fix Coro::Select 'select'.
808 - strict'ify some modules.
809 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
810 - use XSLoader in selected modules.
811 - remove some 5.6 compatibility cruft.
812
813 1.4 Tue Sep 6 00:11:05 CEST 2005
814 - libcoro did not take into account the trampoline on amd64,
815 when 'l'inux method was used.
816
817 1.31 Tue Aug 30 23:31:33 CEST 2005
818 - some portability fixes/workarounds.
819
820 1.3 Sat Aug 20 03:08:56 CEST 2005
821 - no code changes. module seems to work fine.
822 - improve Coro::State docs, remove reference to nonexisting
823 Coro::State::flush method.
824 - no longer autodetect windows, present it as an option instead.
825
826 1.2 Mon May 16 02:00:55 CEST 2005
827 - included libcoro.c earlier in State.xs, might improve portability.
828 - use a faster and possibly more stable Coro::Cont implementation.
829 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
830 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
831 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
832
833 1.11 Thu Mar 3 18:00:52 CET 2005
834 - change of contact address.
835
836 1.1 Tue Feb 22 20:51:16 CET 2005
837 - support [l]inux method on amd64.
838 - allow some context switches while compiling/eval'ing.
839
840 1.0 Tue Aug 10 03:47:27 CEST 2004
841 - Event 0.89 fixes the bug again.
842 - fix the remaining(??) memleak. I didn't free the cache-AV for
843 padlists, so one AV per created anon-sub leaked.
844
845 0.97 Fri May 14 15:23:32 CEST 2004
846 - remove stupid gcc extension that was an experiment only anyway.
847
848 0.96 Thu May 13 18:09:29 CEST 2004
849 - tell people if they have yet another broken Event variant.
850 - made Coro to work _better_ with threads (it seems that, as long
851 as you use Coros in one thread only everything will work fine.
852 Looking at the perl source the problem might be missing locking
853 between Coro and some perl internal routines. That's not easily
854 fixable).
855 - (hopefully) got rid of the static xnext inside transfer.
856
857 0.95 Sun Feb 15 03:20:28 CET 2004
858 - removed Coro::State::flush, implemented a new and much
859 faster caching mechanism by attaching magic to the CV.
860 - a memleak with (real) closures remains.
861 - dramatically reduced size of stacks to 65536 (linux/x86), after
862 observing that even myhttpd never had more than a kilobyte
863 of stack in use (stack is still only physically allocated in
864 pagesize increments for systems suporting mmap).
865 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
866 - fix for threaded perl.
867
868 0.9 Sun Nov 23 04:32:39 CET 2003
869 - fix a few very hard to track down but (of course) stupid
870 memory leaks.
871 - stringify version number to avoid locale problems :(.
872 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
873 hopefully ignoring it "just works".
874
875 0.8 Wed Nov 5 19:38:40 CET 2003
876 - port to perl5.005_03, but only 5.8.x+ is supported!
877 - honor LocalAddr even without LocalPort in Coro::Socket.
878
879 0.7 Tue May 27 03:12:38 CEST 2003
880 - the version jump indicates some level of testing, not gobs
881 of new features.
882 - uh... I found the prompt function in ExtUtils::MakeMaker.
883 Highly correct stuff, that is...
884 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
885 the fix is not well understood by the author, so beware :(.
886
887 0.652 Thu May 8 02:54:46 CEST 2003
888 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
889 cause version 4 doesn't have ucontext.
890 - Benjamin Reed reported that setjmp works fine on darwin,
891 so preselect it.
892
893 0.651 Sat Mar 29 15:00:23 CET 2003
894 - fix a bug in Coro::Handle where some bytes could get lost
895 on reading, reported by jason@nichego.net.
896
897 0.65 Sun Mar 23 00:08:26 CET 2003
898 - added a README.linux-glibc.
899 - new module Coro::Select.
900 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
901 (I never claimed that I know what I am doing ;).
902 - set default to setjmp/longjmp on non-x86-linux.
903
904 0.6 Thu Nov 21 11:09:06 CET 2002
905 - made lazy context switching the default.
906 - fixed the bug where SemaphoreSet::guard
907 was the same as timed_guard and timed_guard was missing.
908 - fix a memleak in Coro::Timer.
909 - reclassified context sharing as not experimental.
910
911 0.534 Sun Apr 14 03:05:12 CEST 2002
912 - fixed backspace => tab in header parsing (myhttpd).
913 - added eg/readline, for elmex.
914 - Coro::Event now calls ->start on first call instead of
915 ->again, which created wrong timeouts for the first ->next.
916 - fixed a bug where Coro::Socket returned a closed socket instead
917 of nothing on ECONNREFUSED and similar error conditions.
918
919 0.533 Mon Feb 18 18:49:40 CET 2002
920 - adapt to finally working Event-0.85.
921 - fixed buggy event initialization (trapped under windows, but how
922 could it possibly have worked anywhere?)
923
924 0.532 Wed Jan 16 02:45:32 CET 2002
925 - removed the reference to pp_entersub, might work on
926 (some) win32 perls now (testsuite works under cygwin,
927 without Event, which freezes).
928
929 0.531 Mon Dec 10 22:18:44 CET 2001
930 - Fixed a bug in SemaphoreSet::timed_down.
931
932 0.53 Tue Nov 27 21:11:13 CET 2001
933 - some tests for Event.
934 - slightly different internal architecture to get refcnt into the object
935 and not the reference. might fix a few bugs, certainly did introduce
936 new ones.
937 - Coro::Timer (independent of Event).
938 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
939
940 0.52 Tue Nov 6 21:36:18 CET 2001
941 - ported to cygwin (trivial).
942 patch by Gerrit P. Haase <gp@familiehaase.de>.
943 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
944
945 0.51 Thu Nov 1 20:39:01 CET 2001
946 - terminate/cancel now work properly (otherwise termination
947 could cause a "next coroutine is not and contains not..."
948 error.
949 - added Coro::Socket::shutdown.
950 - Coro::Event::loop no is the same as Event::loop.
951 - implemented terminate with args + join.
952
953 0.5 Fri Sep 28 16:15:35 CEST 2001
954 - fixed "print" on a Coro::Handle. The print method worked.
955 - small tweaks (seem to reduce memory consumption a lot)
956 in various modules.
957 - splendid use of "no warnings" scattered throughout.
958 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
959
960 0.49 Sun Sep 16 02:42:45 CEST 2001
961 - changed some method calls to function calls for speed
962 inside Coro::Handle.
963 - make Coro::Handle use an array instead of a hash for
964 speed reasons.
965 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
966 sjlj and ucontext should now work.
967 - IRIX-specific port for libcoro.
968 - swapped order of accept results to match IO::Socket.
969 - changed getsock/peername to sock/peername to match IO::Socket.
970 - fixed a bug that caused segfault when returning to main task
971 under some circumstances.
972 - other bugfixes.
973
974 0.45 Sun Sep 2 02:54:01 CEST 2001
975 - new method Coro::Handle::timeout.
976 - corrected speling of set...name to get...name in Coro::Socket.
977 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
978 - Coro::Handle now supports fileno and FILENO.
979 - added eg/myhttpd, a web server.
980 - Coro::Socket now accepts numeric ports as well as "port(number)"
981 syntax.
982 - moved some scheduling primitives into xs code. more to come.
983 - new simple priority system.
984 - implemented Coro::Event using XS, almost four times faster!
985 - small memory corruption problem fixed (boy that was difficult).
986
987 0.13 Wed Aug 8 16:53:07 CEST 2001
988 - new method Coro::Handle::readline.
989 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
990 (see eg/lwp on how to make LWP non-blocking).
991 - renamed Coro::Socket::new_inet to new.
992 - added Coro::Util, some utility functions.
993 - die/eval now works better (still get annoying (but true ;)
994 "Callback called exit" messages).
995 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
996 - much better Coro::Socket support.
997
998 0.12 Fri Jul 27 04:19:01 CEST 2001
999 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
1000 - very experimental stack sharing algorithm. not 100% safe but
1001 should work well in practise ;)
1002 - added Coro::RWLock.
1003 - Coro::Specific now works.
1004
1005 0.11 Tue Jul 24 22:49:21 CEST 2001
1006 - added specialized hack for newer and older linux versions (fast).
1007 - renamed Coro::Event::IO to Coro::Handle.
1008 - new module Coro::Socket.
1009
1010 0.1 Tue Jul 24 01:47:53 CEST 2001
1011 - release candidate 3. A rather new internal structure :(
1012 - the great renaming: Damian Conway gave me a suitable replacement
1013 for yield.
1014 - added Coro/libcoro, a portable coroutine implementation for C
1015 (not even perl-dependent!!!), which can be used by Coro::State.
1016 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
1017 perl-5.6.
1018 - new function Coro::Event::idle.
1019 - the idle process is now overriden
1020 by default in Coro::Event.
1021 - Coro::Channel now enforces the size.
1022 - canceling events now works in all cases.
1023 - Coro state now include $_ and $@.
1024 - yet another bug workaround that I do not really understand :(
1025 - new module Coro::Event::IO (very undocumented) to do
1026 non-blocking i/o.
1027 - performance tuning.
1028
1029 0.08 Thu Jul 19 06:13:25 CEST 2001
1030 - release candidate 2 (new functionality).
1031 - add Coro::State::flush function.
1032 - transfer now supports a flags value (mostly for speed ATM).
1033 - might compile and work in the presence of threads now.
1034 - continuations have a saner syntax.
1035 - no more memleaks.
1036
1037 0.07 Tue Jul 17 17:40:18 CEST 2001
1038 - release candidate 1 ;)
1039 - slightly nicer code.
1040 - fixed a scheduling bug in Coro::Event.
1041
1042 0.06 Tue Jul 17 04:23:24 CEST 2001
1043 - ok, I found the showstopper - the same sub must not be
1044 re-entered in two different coroutines, otherwise => crash. I
1045 see no easy solution to this problem, except by walking the call
1046 chaing and saving/restoring the cv's, which is what I do now.
1047 - memory leaks still latent, especially at thread termination.
1048 - Coro::Event now works (haha).
1049
1050 0.05 Sun Jul 15 17:32:20 CEST 2001
1051 - fixed a few issues in Coro::Event.
1052 - I forgot to include Coro::Event in 0.04 :(:(:(
1053
1054 0.04 Sun Jul 15 05:24:59 CEST 2001
1055 - @_ is now properly localized.
1056 - Coro::State is now easier subclassable.
1057 - Coro::Cont now coroutine-aware.
1058 - Coro::Specific is a low-overhead module to create
1059 coroutine-specific vars.
1060 - Coro::Event provides a simple interface to Event.
1061
1062 0.03 Fri Jul 13 14:51:52 CEST 2001
1063 - transfer() now implemented in XS (beware).
1064 - new module Coro::Cont for really faked continuations.
1065 - big internal architecture changes: Coro::State is now
1066 really low-level and can thus be used to implement other
1067 interesting things, While "Coro::" implements a process-like
1068 model. Still crude and subject to change.
1069 - $_ and $@ are no longer being localized.
1070
1071 0.02 Tue Jul 10 01:38:17 CEST 2001
1072 - implemented "async" attribute.
1073 - $_ and $@ are now localized.
1074 - added Coro::Channel.
1075 - more testcases, still no docs.
1076
1077 0.01 Tue Jul 3 02:18:41 CEST 2001
1078 - original version; copied from Convert::Scalar.
1079