ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.524
Committed: Thu Oct 1 23:16:27 2009 UTC (14 years, 7 months ago) by root
Branch: MAIN
Changes since 1.523: +3 -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 - Coro::Storable destroyed the prototypes of the functions it wrapped.
12 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
13 - added section on "windows process emulation" to the manpage,
14 after a not-so-fruitful (nor-friendly) "discussion" with chip
15 salzenberg (discussion implies arguments, but his only arguments
16 were ad-hominems, one wonders why he started it in the first
17 place). I hope this explains it well enough for him to understand,
18 and maybe well enough for others to understand.
19 - use common::sense everywhere now.
20
21 5.17 Sat Aug 22 23:09:31 CEST 2009
22 - work around a bug in the perl debugger causing crashes
23 when running under the debugger by marking _pool_handler as nodebug.
24 - speed up Coro::async considerably.
25 - try some hacks to get netbsd to work "more often" - their broken
26 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
27 - convert Coro to AE 5.0 API.
28
29 5.162 Tue Jul 28 04:04:03 CEST 2009
30 - perl 5.8.2 is now minimum requirement.
31 - skip t/19_handle.t on broken windows perls.
32
33 5.161 Wed Jul 22 04:47:38 CEST 2009
34 - Coro::AnyEvent::poll could have a different prototype when EV was
35 used as backend (analyzed by Tatsuhiko Miyagawa).
36 - Coro::AnyEvent errornously initialised the event loop when loaded,
37 not on demand.
38 - try to workaround rare */t/01_unblock.t failures.
39
40 5.16 Tue Jul 21 01:44:37 CEST 2009
41 - Coro::AnyEvent failed to hook into the event loop
42 when no threads had been readied between detecting
43 the event loop and actually running it.
44 - considerably speed up Coro::Select by taking avdantage
45 of AnyEvent > 4.8 and some other optimisations.
46 - implement paragraph readline mode in Coro::Handle
47 (based on patches by Zsbán Ambrus).
48 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
49 and analyzed by Yasuhiro MATSUMOTO).
50 - clarified libcoro license and copyright.
51 - someone stole my EXTRA_META!!!
52 - implement Coro::Select::patch_pp_sselect and it's brother,
53 for hardcode select overriding.
54
55 5.151 Mon Jul 6 05:41:57 CEST 2009
56 - backport to windows process emulation code again (patch by
57 Yasuhiro MATSUMOTO).
58 - slightly update Coro::MakeMaker.
59
60 5.15 Tue Jun 30 10:28:06 CEST 2009
61 - deprecate Coro::Socket, document how to get ipv6 support via
62 AnyEvent::Socket instead.
63 - implement signal->wait ($cb) interface, similar to semaphores.
64 - work around SvOK not supporting getmagic, so we have to getmagic
65 to test for undef :/ (reported by Matthias Waldorf).
66 - load Coro::AnyEvent in all modules using AnyEvent.
67 - work around perl corrupting our internal data structures,
68 reported by Tokuhiro Matsuno.
69 - enable per-coroutine real and cpu time gathering
70 (Coro::State::enable_times).
71
72 5.14 Wed Jun 24 01:37:48 CEST 2009
73 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
74 as this seems to have been a difficult concept (poll, sleep, idle,
75 idle_upto).
76 - add Coro::AnyEvent::readable/writable functions.
77 - clarify Coro::EV/Event/AnyEvent manpages.
78 - free per-thread global scalars in the thread calling ->cancel, to
79 avoid crashes when $_, $@ etc., are magical but some of those
80 had already been freed.
81 - "unexperimentalise" the callback interface for Coro::Semaphore.
82 - speed up ready queue management/context switching by using a linked
83 list instead of an array (~5-10%).
84 - implement "watch" command in Coro::Debug shells.
85 - for fun, implement time-slicing as an example in the manpage.
86 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
87 Coro::Event, use the normal AnyEvent handling instead of dieing
88 (the same is true for Coro::Handle).
89 - properly document Coro::EV::timed_io_once.
90 - avoid unneccessary ->cancel calls in Coro::Handle.
91 - maybe make it work on mingw32 with win32 backend
92 (based on patch by Yasuhiro Matsumoto).
93
94 5.132 Fri May 29 09:00:39 CEST 2009
95 - do not keep a reference to the argument itself in
96 Coro::Semaphore::guard, as it could change later.
97 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
98 should support a prepare callback.
99
100 5.131 Mon Mar 16 23:20:37 CET 2009
101 - implement and document Coro->suspend, Coro->resume.
102 - fix Coro::Select implementation to not (often) close
103 the passed file descriptors (testcase provided by pippijn).
104
105 5.13 Mon Dec 15 21:51:42 CET 2008
106 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
107 - don't set diehook to C<undef> but instead to NULL, to avoid
108 spurious warnings.
109 - fix a lot of bugs in Coro::SemaphoreSet.
110 - Coro::SemaphoreSet will less often create a semaphore needlessly.
111 - add Coro::SemaphoreSet::count and wait methods.
112 - take advantage of the new Guard module.
113 - deprecate Coro::guard.
114 - try to fix the dreaded 01_unblock tests once more. I hate it when
115 testsuites need more fixing than the code they are supposed to test.
116 - croak in more cases when a required callback isn't resolvable.
117 - fix some minor issues in Coro::State->call/eval.
118 - use current coroutine context instead of a temporary one
119 when temporarily switching to another coroutine.
120 - do not call C-level on_destroy handlers during global destruction,
121 to avoid needless segfaults.
122
123 5.12 Sun Dec 7 13:30:38 CET 2008
124 - add default config for MirOS, which seems to be a bug-to-bug
125 compatible fork of openbsd ("world domination by releasing
126 openbsd cvs before the openbsd folks do it" or so :).
127 - free_padlist did destroy the names pad, not good, but didn't
128 seem to bother perl - this could fix issues such as eval ""
129 inside a function called from multiple coroutines.
130 - use a different method to detect destruction time.
131 - be more careful when freeing padlists just before global
132 destruction.
133 - fixed and expanded the call/cc example.
134 - renamed _terminate to _coro_run.
135 - new method Coro::Channel->shutdown.
136 - try pthreads on openbsd <4.4 (broken sigaltstack, will
137 pthreads fare better?).
138 - be less picky about destroying "a" running coroutine.
139
140 5.11 Tue Nov 25 21:49:05 CET 2008
141 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
142 working code with failed assertions, introducing more bugs than
143 it fixes, requiring elaborate workarounds :(
144
145 5.1 Mon Nov 24 08:54:59 CET 2008
146 - wrote a small introductory tutorial - Coro::Intro.
147 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
148 - Coro::Select did errornously dup the file descriptors
149 and didn't work with all AnyEvent backends.
150 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
151 LWP data transfers.
152 - disassociate c contexts from coro objects - this is agruably more
153 correct, but mostly allows sharing of cctxs between coro and state
154 objects, for added memory savings and speed increases.
155 - bumped $Coro::POOL_RSS up to 32kb by default.
156 - no longer set the optype to OP_CUSTOM, as B::* understandably
157 doesn't like this very much (and we *are* a type of entersub).
158 - implement state cloning, just to prove that call/cc can be done.
159 - automatically load Coro::AnyEvent in Coro::Handle.
160 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
161 do not have this method (and don't need it either).
162 - speed up generic anyevent methods in Coro::Handle by using rouse
163 callbacks.
164 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
165 others. It also makes the debugger happier, as you can trace
166 through the idle threads now.
167 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
168 - eg/event was pretty broken.
169 - better 5.8.6 compatibility.
170
171 5.0 Thu Nov 20 10:35:05 CET 2008
172 - NEW ARCHITECTURE: use the latest 4.x version if you experience
173 stability issues.
174 - bump API version to 7 - all dependents must be recompiled.
175 - removed timed_* functions - they were not being used anyways
176 and should be replaced by a more generic mechanism -
177 and were annoying to support anyways :)
178 - removed SemaphoreSet's waiter method - use sem method instead.
179 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
180 - async_pool did free a scalar value twice
181 ("Attempt to unreference...").
182 - fix a longstanding bug where calling terminate on a coro that
183 was waiting for a semaphore that was just becoming available
184 would cause a deadlock (semaphore would get into a state where
185 it was available but waiters were still blocked).
186 - calling throw on a coroutine that is waiting for a semaphore will
187 no longer make it acquire the semaphore (and thus leak a count).
188 - perl's process emulation is now not even theoretically supported
189 anymore.
190 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
191 conversion of callback-style to blocking-style.
192 - new methods $coro->schedule_to and ->cede_to, to specifically
193 schedule or cede to a specific coroutine.
194 - new function Coro::Semaphore::wait.
195 - use named constants in Coro::Channel (Richard Hundt).
196 - directly patch the entersub opcode calling SLF functions (cede,
197 transfer and so on). this does speed up context switching, but
198 more importanly, it frees us from the hardcoded behaviour of
199 entersub, so we might actually be able to return something from
200 those functions and atcually create new ones.
201 - take advantage of __builtin_frame_address on gcc.
202 - expose THX in coroapi (not sure whether this was a wise decision,
203 as "threaded" perls are running at half speed anyways).
204 - implement execute_slf (schedule-like-function) interface that makes
205 it possible to implement schedule-like-functions in XS.
206 - use new SLF interface to massively speed up Coro::EV by roughly a
207 factor of two.
208 - used new SLF interface to massively speed up Coro::Semaphore by a
209 factor of three.
210 - used new SLF interface to speed up Coro::AIO by roughly a factor of
211 four and reduce its memory usage considerably.
212 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
213 for a nice speedup and vastly more correct behaviour. Also implement
214 a new method "sem" to get at the underlying semaphore object.
215 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
216 (in comparison) 20-40% speedup.
217 - used new SLF interface to reimplement Coro::Signal gaining
218 some unknown (because I was too lazy), but certain, speedup, and also
219 making signals reliable for the first time.
220 - used new SLF interface and other optimisations to speed up async_pool
221 by a factor of two. It also doesn't rely on perl's exception mechanism
222 to exit anymore. The overhead for terminating an async_pool, coro over
223 a normal async is now very small.
224 - sped up coroutine creation/destruction by 40%.
225 - forgot to include Coro/libcoro/README in the dist for all these years.
226 - work around a freebsd pthreads bug (manual testcancel is required as
227 pthread_cond_wait isn't a cancellation point on freebsd).
228 - use new rouse functions to speed up and simplify Coro::BDB.
229 - make "prefer perl native functions" work with threaded perls.
230 - condense Coro::Debug ps output, hint at v and w flags.
231 - (libcoro) lots of minor cleanups and portability improvements.
232
233 4.914 Wed Nov 19 12:54:18 CET 2008
234 - fix a disastrous bug in the readline optimisation
235 introduced in 4.801.
236
237 4.913 Sat Nov 15 07:58:28 CET 2008
238 - async_pool did free a scalar value twice
239 ("Attempt to unreference...").
240
241 4.912 Thu Nov 13 18:31:23 CET 2008
242 - minor cleanups.
243 - use much larger stacks on linux and perl < 5.8.8.
244 - Coro::Debug::new_unix_server did not unlink the socket
245 when destroyed.
246
247 4.911 Tue Nov 11 04:26:17 CET 2008
248 - "port" to threaded perls.
249
250 4.91 Mon Nov 10 05:36:38 CET 2008
251 - the ->throw exception object no longer leaks.
252 - creating a new cctx leaked a scopestack entry (memleak).
253 - new coroutines didn't get created with a zero flags field
254 (unknown impact).
255 - calling ->throw on a not-yet-started coroutine should now work
256 instead of being ignored.
257 - ->throw is now supported on Coro::State objects.
258 - clean up cctx creation code a bit.
259 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
260
261 4.9 Sat Nov 8 17:45:27 CET 2008
262 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
263 - (libcoro) no longer leak threads in the experimental pthread backend,
264 also speed it up considerably.
265 - (libcoro) do not rely on makecontext passing void *'s unscathed.
266 - fix compiletime dependencies on libcoro in the Makefile.
267 - cctx_count wasn't always updated properly.
268 - Coro::State::cctx_stacksize wasn't applied correctly.
269 - new function Coro::State::cctx_max_idle.
270 - the default max number of idle C contexts is now 4.
271 - (libcoro) try harder to get _setjmp/_longjmp.
272 - (libcoro) cleanup and extend the libcoro API to officially
273 allow the creation of empty source contexts.
274 - very experimental workaround for the totally broken netbsd platform.
275 - tried to hack around openbsd bugs.
276
277 4.804 Wed Nov 5 16:36:00 CET 2008
278 - Coro::Debug::new_unix_server would not create a non-blocking listening
279 socket, sometimes causing freezes.
280 - (libcoro) fix misaligned stack points for setjmp and assembly
281 methods, which can cause crashes on x86/x86_64 with a sufficiently
282 aggressive compiler.
283 - new function: Coro::Debug::new_tcp_server.
284 - move ->throw into the Coro class because it only works on coro objects.
285
286 4.803 Mon Nov 3 17:16:12 CET 2008
287 - (libcoro) use a global asm statement to become independent of gcc
288 otpimisations for CORO_ASM (thanks to pippijn for the idea).
289 - try to workaround yet another broken bsd, this time dragonfly.
290
291 4.802 Thu Oct 30 10:56:12 CET 2008
292 - support -fno-omit-frame-pointer on x86 with the assembly method.
293 - tune 01_unblock.t tests a bit.
294
295 4.801 Wed Oct 22 18:33:37 CEST 2008
296 - improve readline speed for very long "lines".
297 - backport to 5.8.8.
298
299 4.8 Thu Oct 2 13:34:40 CEST 2008
300 - new function Coro::AIO::aio_wait.
301 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
302 - greatly speed up and reduce memory usage of Coro::AIO requests.
303 - implement some other µ-optimisations.
304
305 4.749 Mon Sep 29 14:40:12 CEST 2008
306 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
307
308 4.748 Sat Sep 27 14:03:19 CEST 2008
309 - implement, but do not document, PerlIO::cede(granularity).
310 - Coro::Storable forgot to wrap Storable::pstore.
311 - work around the multitude of leaks and memory corruption
312 bugs in PerlIO::via by using our own C-level perliol. As a side
313 effect, Coro::Storable is now much, much, much faster.
314
315 4.747 Tue Sep 23 01:59:41 CEST 2008
316 - fix a per-cv memleak (one empty array was leaked per
317 code reference).
318 - avoid a crash in coro->call|rss when the coroutine was already
319 destroyed (most noticably when using Coro::Debug::ps :)
320 - also protect *Storable::FILE.
321 - push up default storable granularity to 20ms.
322
323 4.746 Sun Sep 21 03:22:20 CEST 2008
324 - be more insistent on locking Storable against reentrancy
325 in Coro::Storable.
326 - move swap_def?v and throw to Coro::State, as documented.
327
328 4.745 Thu Jul 24 00:14:38 CEST 2008
329 - remove debugging code related to MgPV_nolen_const, also try to
330 make it compile with perl 5.8.6 (yes, apple apparently loves
331 outdated software). Reported by John S.
332
333 4.744 Tue Jul 8 22:06:35 CEST 2008
334 - correctly provide default DIE/WARN handlers as documented.
335 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
336 implementations inside perl work fine for us.
337
338 4.743 Mon Jun 16 00:21:57 CEST 2008
339 - when using Coro::EV without running EV::loop it could
340 result in busy-waiting for events, this has been fixed.
341 - reduce codesize and improve performance by using EV_DEFAULT_UC.
342
343 4.742 Sat May 31 14:10:21 CEST 2008
344 - implement a workaround for (some) perl <5.8.8 versions.
345 - require EV 3.3+.
346
347 4.741 Fri May 30 23:33:09 CEST 2008
348 - tell netbsd how utterly broken their imitation of an OS is
349 and refuse to build by default if pthreads are in use.
350 - switch to "s" method on all bsd's by default, as their ucontext
351 stuff seems just too broken.
352 - fix a bug in Coro::Select.
353
354 4.74 Thu May 29 20:05:31 CEST 2008
355 - do not test Coro::LWP for lack of dependencies.
356 - do not test Coro::Select for lack of working perls.
357
358 4.73 Thu May 29 2008
359 - fix a bug in Coro::EV which would cause it to block despite
360 there being runnable coroutines.
361 - sprinkle "no warnings" freely over everything, also suppress
362 warnings for some other modules.
363 - fix typo in WSAEWOULDBLOCK.
364
365 4.72 Sun May 25 05:14:36 CEST 2008
366 - tweak META.yaml a bit, unfortunately, there is no documented way
367 to have optional dependencies with CPAN. doh :(
368 - avoid running some tests on windows because they would fail due to
369 perl bug (broken fork, broken pipes...).
370 - work around perl on windows bugs where perl returns undocumented
371 error codes for sysread, syswrite etc. by taking advantage
372 of AnyEvent's workaround for that problem.
373 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
374 a common perl implementation bug on windows.
375 - use unix domain sockets in testsuite to work around
376 common perl implementation bugs on widows (they are emulated by
377 tcp sockets on windows. ugh.)
378
379 4.71 Sat May 24 20:01:27 CEST 2008
380 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
381 - take advantage of async name resolution of AnyEvent::Util.
382 - work around brutal inet_aton override in Coro::LWP.
383 - take advantage of the readyhook in Coro::EV, for smoother
384 scheduling.
385
386 4.7 Sun May 11 00:32:19 CEST 2008
387 - completely reworked the Coro manpage.
388 - added Coro::AnyEvent, generic event loop integration.
389 - implement cancel, ready and kill commands in Coro::Debug.
390 - document find_coro in Coro::Debug.
391 - incompatible change: rename has_stack to has_cctx.
392 - Coro::AIO and Coro::BDB no longer force event model detection,
393 use AnyEvent::AIO and AnyEvent::BDB.
394
395 4.6 Sat Apr 26 10:05:14 CEST 2008
396 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
397 by the Coro module. It was a mistake to have it in the first place.
398 - (experimental) support for activestate perl 5.10 (method "w").
399 - (experimental) support for strawberry perl (method "a").
400 - coro_sigelem_set did not return a value although it had to,
401 actual impact unknown.
402
403 4.51 Mon Apr 14 13:28:27 CEST 2008
404 - make it compile again on 5.8.
405
406 4.50 Thu Apr 10 09:43:17 CEST 2008
407 - I did it twice! (see 4.49).
408
409 4.49 Sun Apr 6 21:23:31 CEST 2008
410 - grr, instead of compiling the recent changes on 5.10 only they
411 were compiled on 5.8 only.
412
413 4.48 Sun Apr 6 20:36:46 CEST 2008
414 - allow coroutine switches during eval's under 5.10.x, as it seems
415 the parser is a per-interpreter option now, so this is safe
416 (this might fix the odd crashing bug).
417 - drop support for 5.9.x versions: they are dead, jim.
418
419 4.47 Sun Apr 6 00:37:52 CEST 2008
420 - force cctx allocation on API calls: we know we need to force one
421 and gcc actually manages to confuse our heuristic nowadays,
422 leading to crashes and worse.
423 - document force_cctx.
424
425 4.46 Fri Apr 4 22:05:43 CEST 2008
426 - upgrade libcoro, resulting in pthread-backend (which was only created
427 to fulfill the rules of the programming languages shootout).
428
429 4.45 Thu Mar 13 11:55:36 CET 2008
430 - fix a file leaking bug in eg/mhyttpd that would allow
431 downloading of any file (reported by oesi).
432 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
433 (also add testcase).
434 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
435 apparently having loads of fun with that).
436 - implement $state->swap_defsv, ->swap_defav and document ->throw.
437
438 4.4 Wed Feb 13 16:44:29 CET 2008
439 - only assume .cfi pseudo isns exist on GNU/Linux.
440 - add get_timed method to Coro::Channels.
441 - fixed Coro::Debug manpage.
442 - perl 5.11 compatbility improvement based on patch by
443 Andreas König.
444
445 4.37 Sun Jan 20 11:25:23 CET 2008
446 - soften the check for an active parser for perl 5.10 (avoiding
447 "Coro::State::transfer called while parsing" in many cases).
448
449 4.36 Sun Jan 13 10:53:56 CET 2008
450 - reset diehook when terminating from an async_pool as to not
451 trigger any __DIE__ handlers.
452
453 4.35 Sun Jan 13 04:14:13 CET 2008
454 - "bt" debug command now displays any exceptions
455 from longmess and also skips the innermost
456 stackframes, giving more useufl output.
457 - allow backtraces in coroutines blocked in Coro::EV,
458 at a <1% speed hit.
459 - handle localising of $SIG{__WARN__} and __DIE__
460 properly (with a proper amount of dirty hacking).
461
462 4.34 Sat Dec 22 17:49:53 CET 2007
463 - upgrade to EV version 2.0 API.
464
465 4.33 Mon Dec 17 08:36:12 CET 2007
466 - make Coro::AIO etc. loadable in the absence of EV.
467
468 4.32 Mon Dec 17 07:46:02 CET 2007
469 - majorly improved Coro::Handle's performance with Coro::EV.
470 - implemented the until now mythical Coro::BDB module.
471 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
472
473 4.31 Wed Dec 5 12:32:39 CET 2007
474 - remove warn statement form Coro::Util.
475
476 4.3 Tue Dec 4 20:33:14 CET 2007
477 - calls to the idle function could cause stack corruption
478 when the stack changed.
479 - do no longer rely on the presence of EV::DNS if EV is used
480 (because it is gone), but instead take avdantage of EV::ADNS
481 if available.
482 - add ($) prototypes to all functions in Coro::Storable.
483 - use a conventional (safer) idle callback in Coro::EV.
484 - do accept quickly in Coro::Debug to avoid endless loops.
485
486 4.22 Fri Nov 30 16:04:04 CET 2007
487 - really use optimised versions for Event and EV in Coro::Util
488 and Coro::Handle.
489
490 4.21 Sun Nov 25 10:48:59 CET 2007
491 - fix a spurious memory read.
492 - Coro::EV no longer keeps the eventloop "alive".
493
494 4.2 Fri Nov 9 20:47:05 CET 2007
495 - enable/disable tracing from a new coroutine, not a pooled one.
496 - fix a memleak in Coro::Event.
497 - removed killall call from fork_eval.
498 - made sure store_fd is already loaded so that fork_eval
499 does not have to parse autoload in each subprocess.
500 - only use assembly method if -O switch is in $Config{optimize}.
501 - add (optional) Coro::EV module, so far the best event loop module
502 directly supported by Coro.
503 - if the event model is EV, use EV::DNS to resolve
504 stuff in Coro::Util.
505 - don't get confused by multiple event notifications in Coro::Handle.
506 - initial support for EV (libevent interface).
507 - require Event and EV using configure_requires, to force their existance.
508
509 4.13 Wed Oct 24 07:26:45 CEST 2007
510 - add Coro::Storable::blocking_thaw.
511 - use a vastly more complicated technique to localise
512 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
513 - use a coroutine for the idle callback Coro::Event,
514 instead of running Event in the current coroutine context.
515 This also catches recursive invocations.
516 - actually report fork errors in gethostbyname and inet_aton.
517
518 4.11 Thu Oct 11 02:40:24 CEST 2007
519 - port to threaded perls.
520
521 4.1 Thu Oct 11 02:38:16 CEST 2007
522 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
523 be local to each coro (see Coro::State).
524 - incompatible change: for very deep reasons, cede and cede_notself
525 cannot return anything, so nothing will be returned.
526 - possibly bring back 5.10 compatibility (untested).
527 - work around stupid (and wrong) warning on 5.10 :(.
528 - overlay the saved state over the context stack. This saves
529 a few hundred bytes per coroutine on average and also
530 speeds up context switching a bit.
531 - further tune default stack sizes.
532 - (more) correctly calculate stack usage in coro_rss.
533 - Coro::Storable::blocking_* did not properly lock
534 resulting in races between coroutines.
535 - added Coro::Storable::guard.
536 - stopping to trace a coroutine could destroy the cctx of
537 an unrelated coroutine.
538 - explain the relationship between Perl and C coroutines in
539 more detail in Coro::State.
540 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
541 causing a fork per resolve. This also affected Coro::Socket.
542 - switch to a separate stack in $coro->call/eval to avoid
543 invalidating pointers.
544
545 4.03 Sat Oct 6 21:24:00 CEST 2007
546 - added Coro::throw method.
547 - minor code cleanups.
548
549 4.02 Sat Oct 6 02:36:47 CEST 2007
550 - fix a very minor per-coroutine memleak (a single codereference).
551 - fixed a bug where the currently in-use c context would be freed
552 prematurely (can happen only when programs change the stacksize
553 or use tracing).
554 - tracing can no longer keep a coro alive after it terminated.
555 - do static branch prediction in the common path for gcc. gives
556 about 2-5% speed improvement here.
557
558 4.01 Fri Oct 5 22:10:49 CEST 2007
559 - instead of recreating *a* standard output handle we simply
560 use STDOUT, which is faster and hopefully more robust.
561
562 4.0 Fri Oct 5 12:56:00 CEST 2007
563 - incompatibly changed Coro::Storable::freeze.
564 - major new feature: added Coro::Debug, for interactive coroutine
565 debugging, tracing and much more.
566 - major bug fix: unbelievable, but true: $_, $/ and many other
567 "saved" variables actually weren't being saved. This has been fixed,
568 of course, while increasing performance while losing all the save
569 flags.
570 - save flags are gone, and all the api functions dealing with them.
571 - added Coro::Semaphore::adjust.
572 - added Coro::Util::fork_eval.
573 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
574 - added Coro::killall.
575 - reduce initial stack sizes to allow for "micro-coroutines".
576 - better async_pool resource management, moved parts of async_pool
577 handling to XS (major speed improvement).
578 - actually croak before modifying important data structures.
579 - refuse to transfer while compiling.
580 - possibly support eval EXPR better now.
581 - enable assembly per default on linux+bsd x86+amd64.
582 - all internal members were renamed _something for easier subclassing.
583 - many minor tweaks.
584
585 3.63 Wed May 16 14:10:06 CEST 2007
586 - implement handcoded assembly for x86/amd64 SVR ABI.
587
588 3.62 Fri Apr 27 21:36:06 CEST 2007
589 - upgrade libcoro (which might set unwind info correctly).
590 - change default on linux to setjmp/longjmp.
591
592 3.61 Thu Apr 19 12:36:18 CEST 2007
593 - Coro::Storable caused an endless loop when thawing invalid
594 pst's sometimes.
595 - use a Semaphore in Coro::Storable, as Storable doesn't
596 seem to be reentrant (although it is documented to
597 be threadsafe...).
598 - fix Coro::Signal to bring back the original unreliable
599 but stateful semantics.
600 - fixed a lot of typos in Coro.pm (patch submitted by David
601 Steinbrunner, which applied flawlessly).
602
603 3.6 Sat Apr 14 17:13:31 CEST 2007
604 - added some bugfixes to get eg/myhttpd working again.
605 - added Coro::Storable for often-cede'ing freeze/thaw.
606 - try to do a clean exit when a coroutine calls exit
607 (EXPERIMENTAL).
608 - got rid of indirect call through _coro_init.
609 - updated the partly antique examples in eg/ to
610 work again and be a bit less magic, too.
611 - fixed Coro::Signal semantics to work as documented again.
612
613 3.55 Sun Mar 25 01:20:47 CET 2007
614 - add SAVE_DEFFH to save the default filehandle and enable
615 it by default.
616 - finally move socket-operations from Coro::Socket to Coro::Handle
617 to be able to unblock foreign sockets.
618 - add Coro::State::save_also and guarded_save.
619 - add count accessor to Coro::Semaphore.
620 - add Coro::State::cctx_stacksize.
621 - just for the fun of it, do not rely on implicit context,
622 which can dramatically improve performance, but people
623 using windows-process-emulation perls do not care much about
624 performance.
625
626 3.51 Sun Mar 4 14:18:04 CET 2007
627 - fixed a problem when you weakened references to Coro::State's
628 (patch by Yuval Kogman).
629
630 3.501 Wed Feb 28 12:44:07 CET 2007
631 - rename some global symbols as macosx from hell redefines
632 them without asking.
633
634 3.5 Tue Feb 13 20:22:53 CET 2007
635 - do AnyEvent model detection earlier, avoiding problems
636 caused by first using AnyEvent and later Coro::Event.
637 - implement and document Coro::Event event objects.
638 - fix a potential problem in Coro::Event causing crashes.
639 - initialise PL_comppad when creating a new coroutine,
640 avoids crashes on early coro destruction.
641
642 3.41 Mon Jan 22 19:19:49 CET 2007
643 - readline on Coro::Handle did not support undefined $/,
644 nor did it deliver partial lines on EOF or error.
645 - implement malloc fallback for stack allocation because
646 stupid broken idiotic OSX has a stupid broken
647 idiotic fits-the-whole-os mmap "implementation" and
648 my dick feels longer if Coro is portable even to
649 obsolete platforms.
650
651 3.4 Fri Jan 19 21:52:54 CET 2007
652 - remove t/09_timer.t, as it isn't really testing much
653 but was rather flaky in practise.
654 - async_pool coro would keep arguments and callback alive until
655 it was reused.
656 - cancellation of a coroutine could cause spurious idle calls
657 in cede_notself.
658
659 3.3 Sat Jan 6 03:45:00 CET 2007
660 - implement $coro->on_destroy.
661 - Coro::Event blocking semantics have been changed,
662 documented and - hopefully - improved.
663 - fix nice adding, not subtracting, from priority.
664 - fix ->prio and api_is_ready (patch by Mark Hinds).
665 - fixed an assert ("... == prev__cctx->idle_te")
666 that could errronously trigger.
667 - fix various large and small memleaks.
668 - use a (hopefully) more stable cancel implementation
669 that immediately frees the coroutine data.
670 - cede/cede_notself return a status now.
671 - added Coro::guard function.
672 - added a global coroutine pool for jobs (on my machine,
673 I can create and execute 48k simple coros/s with async,
674 and 128k coros with async_pool).
675 - Coro::AIO now uses the coroutine priority as io priority.
676
677 3.2 Fri Dec 22 05:07:09 CET 2006
678 - improve portability to slightly older perls.
679 - use cleaner coroutine destruction.
680 - simplify configuration for users.
681 - optionally (unrecommended) prefer perl functions over
682 their coro replacements.
683
684 3.11 Tue Dec 5 13:11:24 CET 2006
685 - fixed some bogus assert's, but as perl.h disables assert even
686 without NDEBUG (thank you very much), not too many people should
687 notice (that did include myself). Andreas König noticed, though :)
688 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
689 it. Also noticed by Andreas König :)
690 - save/restore tainted status.
691 - verified to pass the testsuite on my 5.9.5.
692
693 3.1 Mon Dec 4 23:03:40 CET 2006
694 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
695 - incompatible change: transfer flags are now per-state.
696 - give a better error message on deadlock.
697 - document Coro::nready.
698 - enhanced testsuite.
699
700 3.01 Sun Dec 3 23:47:42 CET 2006
701 - forgot to include Coro::Timer.
702
703 3.0 Sun Dec 3 22:57:25 CET 2006
704 - the "FINALLY COMPLETELY STABLE" release (coming soon:
705 the "FAMOUS LAST WORDS" release).
706 - implement a new stack sharing algorithm, which uses a stack
707 pool (size currently hardcoded).
708 - make stack sharing mandatory (it no longer uses a heuristic).
709 - eval/die no longer cause weird problems under heavy use.
710 - Coro::Event could cause livelocks if it was used but
711 no Coro::Event watchers were used.
712 - Coro::Event now uses asynccheck as prepare does not
713 check for changed watchers.
714 - Coro::Event allows multiple waiting coros and will wake up one
715 per event.
716 - Coro::Event should be cleaner and more efficient now.
717 - new utility function Coro::unblock_sub.
718 - document the sad fact that Event is no longer reentrant.
719 - putting a coroutine into the ready queue twice could under
720 some circumstances lead to stack corruption.
721 - minor incompatibility: subclassing Coro::State is supported
722 directly now without going through a _coro_state member.
723 - state/coro switching is much faster now.
724 - very minor optimisations and code/documentation cleanup.
725 - avoid problems due to compiler inlining.
726 - removed timers from Coro::Timer -> use AnyEvent instead.
727 - replaced Coro::idle coroutine by (cleaner) idle handler.
728 - updated to newest libcoro.
729 - implement enhanced support for valgrind.
730 - implement is_ready and return value for ready.
731 - removed Coro::Cont, it was a misnomer (it's generators),
732 and it was somewhat annoying to get it right. Will come back
733 if somebody asks for it :)
734 - many ->wait methods and Coro::Event could return
735 spuriously without the event having happened.
736
737 2.5 Tue Nov 7 12:22:33 CET 2006
738 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
739 use AnyEvent, moved them to Coro/.
740 - added Coro::LWP which contains all the uglyness required to
741 make LWP non-blocking.
742 - should work with perl 5.9.x now (Andreas König made me do it).
743 - fixed another bug in Coro::Select when the timeout was undef.
744 - reuse PL_start_env for all coros, saves some memory per coroutine.
745 - manage PL_top_env differently, hopefully to avoid panic: top_env.
746 - timeout argument was not properly used in Coro::Socket.
747 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
748 - emulate undocumented(!) functionality of IO::Socket required
749 by LWP(!!).
750 - updated eg/lwp to work with newer lwp's.
751 - remove "FATAL: uncaught exception" prefix. Coroutines that die
752 kill the whole process, just as exceptions in the main "coroutine"
753 did already.
754
755 2.1 Wed Nov 1 23:01:13 CET 2006
756 - fix a long-standing bug in Coro::Select where select with
757 zero timeout would instead change the current default filehandle.
758 - use a simpler and hopefully more robust way to clone padlists
759 (uses less memory and a perl function instead of our own).
760 - coro can now create a stack guard area on many architectures.
761 - Coro::AIO properly reexports additional functions from IO::AIO.
762 - updated libcoro with a workaround for OS X,
763 pach and testing by Michael Schwern.
764
765 2.0 Tue Oct 24 05:47:17 CEST 2006
766 - support additional aio requests in Coro::AIO.
767
768 1.9 (never properly released due to a glitch)
769
770 1.8 Thu Feb 2 00:59:06 CET 2006
771 - applied suggested patch by SAMV to avoid problems during stupid
772 mark & sweep gc run.
773 - applied patch by Scott Walters for 5.9.3 compatibility.
774
775 1.7 Tue Dec 27 01:41:58 CET 2005
776 - added Coro::AIO, a thin wrapper around IO::AIO.
777 - improved Makefile.PL explanations.
778
779 1.6 Mon Dec 12 21:30:05 CET 2005
780 - additionally save PM_curpm might fix as-of-yet
781 unseen problems with regex matches being attributed to the
782 wrong package.
783 - add t/10_bugs.t, which currently checks against imho broken
784 perls that use 0.26MB of stack space per Perl_magic_get
785 invocation (newer linuxes) as opposed to the 0.0002MB perl
786 normally uses.
787 - make stacksize configurable for the ultimate debian experience.
788
789 1.51 Mon Dec 12 18:48:36 CET 2005
790 - remove debugging warn() accidentally left in Coro::Select.
791
792 1.5 Tue Nov 29 13:32:44 CET 2005
793 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
794 - fix Coro::Select 'select'.
795 - strict'ify some modules.
796 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
797 - use XSLoader in selected modules.
798 - remove some 5.6 compatibility cruft.
799
800 1.4 Tue Sep 6 00:11:05 CEST 2005
801 - libcoro did not take into account the trampoline on amd64,
802 when 'l'inux method was used.
803
804 1.31 Tue Aug 30 23:31:33 CEST 2005
805 - some portability fixes/workarounds.
806
807 1.3 Sat Aug 20 03:08:56 CEST 2005
808 - no code changes. module seems to work fine.
809 - improve Coro::State docs, remove reference to nonexisting
810 Coro::State::flush method.
811 - no longer autodetect windows, present it as an option instead.
812
813 1.2 Mon May 16 02:00:55 CEST 2005
814 - included libcoro.c earlier in State.xs, might improve portability.
815 - use a faster and possibly more stable Coro::Cont implementation.
816 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
817 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
818 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
819
820 1.11 Thu Mar 3 18:00:52 CET 2005
821 - change of contact address.
822
823 1.1 Tue Feb 22 20:51:16 CET 2005
824 - support [l]inux method on amd64.
825 - allow some context switches while compiling/eval'ing.
826
827 1.0 Tue Aug 10 03:47:27 CEST 2004
828 - Event 0.89 fixes the bug again.
829 - fix the remaining(??) memleak. I didn't free the cache-AV for
830 padlists, so one AV per created anon-sub leaked.
831
832 0.97 Fri May 14 15:23:32 CEST 2004
833 - remove stupid gcc extension that was an experiment only anyway.
834
835 0.96 Thu May 13 18:09:29 CEST 2004
836 - tell people if they have yet another broken Event variant.
837 - made Coro to work _better_ with threads (it seems that, as long
838 as you use Coros in one thread only everything will work fine.
839 Looking at the perl source the problem might be missing locking
840 between Coro and some perl internal routines. That's not easily
841 fixable).
842 - (hopefully) got rid of the static xnext inside transfer.
843
844 0.95 Sun Feb 15 03:20:28 CET 2004
845 - removed Coro::State::flush, implemented a new and much
846 faster caching mechanism by attaching magic to the CV.
847 - a memleak with (real) closures remains.
848 - dramatically reduced size of stacks to 65536 (linux/x86), after
849 observing that even myhttpd never had more than a kilobyte
850 of stack in use (stack is still only physically allocated in
851 pagesize increments for systems suporting mmap).
852 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
853 - fix for threaded perl.
854
855 0.9 Sun Nov 23 04:32:39 CET 2003
856 - fix a few very hard to track down but (of course) stupid
857 memory leaks.
858 - stringify version number to avoid locale problems :(.
859 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
860 hopefully ignoring it "just works".
861
862 0.8 Wed Nov 5 19:38:40 CET 2003
863 - port to perl5.005_03, but only 5.8.x+ is supported!
864 - honor LocalAddr even without LocalPort in Coro::Socket.
865
866 0.7 Tue May 27 03:12:38 CEST 2003
867 - the version jump indicates some level of testing, not gobs
868 of new features.
869 - uh... I found the prompt function in ExtUtils::MakeMaker.
870 Highly correct stuff, that is...
871 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
872 the fix is not well understood by the author, so beware :(.
873
874 0.652 Thu May 8 02:54:46 CEST 2003
875 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
876 cause version 4 doesn't have ucontext.
877 - Benjamin Reed reported that setjmp works fine on darwin,
878 so preselect it.
879
880 0.651 Sat Mar 29 15:00:23 CET 2003
881 - fix a bug in Coro::Handle where some bytes could get lost
882 on reading, reported by jason@nichego.net.
883
884 0.65 Sun Mar 23 00:08:26 CET 2003
885 - added a README.linux-glibc.
886 - new module Coro::Select.
887 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
888 (I never claimed that I know what I am doing ;).
889 - set default to setjmp/longjmp on non-x86-linux.
890
891 0.6 Thu Nov 21 11:09:06 CET 2002
892 - made lazy context switching the default.
893 - fixed the bug where SemaphoreSet::guard
894 was the same as timed_guard and timed_guard was missing.
895 - fix a memleak in Coro::Timer.
896 - reclassified context sharing as not experimental.
897
898 0.534 Sun Apr 14 03:05:12 CEST 2002
899 - fixed backspace => tab in header parsing (myhttpd).
900 - added eg/readline, for elmex.
901 - Coro::Event now calls ->start on first call instead of
902 ->again, which created wrong timeouts for the first ->next.
903 - fixed a bug where Coro::Socket returned a closed socket instead
904 of nothing on ECONNREFUSED and similar error conditions.
905
906 0.533 Mon Feb 18 18:49:40 CET 2002
907 - adapt to finally working Event-0.85.
908 - fixed buggy event initialization (trapped under windows, but how
909 could it possibly have worked anywhere?)
910
911 0.532 Wed Jan 16 02:45:32 CET 2002
912 - removed the reference to pp_entersub, might work on
913 (some) win32 perls now (testsuite works under cygwin,
914 without Event, which freezes).
915
916 0.531 Mon Dec 10 22:18:44 CET 2001
917 - Fixed a bug in SemaphoreSet::timed_down.
918
919 0.53 Tue Nov 27 21:11:13 CET 2001
920 - some tests for Event.
921 - slightly different internal architecture to get refcnt into the object
922 and not the reference. might fix a few bugs, certainly did introduce
923 new ones.
924 - Coro::Timer (independent of Event).
925 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
926
927 0.52 Tue Nov 6 21:36:18 CET 2001
928 - ported to cygwin (trivial).
929 patch by Gerrit P. Haase <gp@familiehaase.de>.
930 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
931
932 0.51 Thu Nov 1 20:39:01 CET 2001
933 - terminate/cancel now work properly (otherwise termination
934 could cause a "next coroutine is not and contains not..."
935 error.
936 - added Coro::Socket::shutdown.
937 - Coro::Event::loop no is the same as Event::loop.
938 - implemented terminate with args + join.
939
940 0.5 Fri Sep 28 16:15:35 CEST 2001
941 - fixed "print" on a Coro::Handle. The print method worked.
942 - small tweaks (seem to reduce memory consumption a lot)
943 in various modules.
944 - splendid use of "no warnings" scattered throughout.
945 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
946
947 0.49 Sun Sep 16 02:42:45 CEST 2001
948 - changed some method calls to function calls for speed
949 inside Coro::Handle.
950 - make Coro::Handle use an array instead of a hash for
951 speed reasons.
952 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
953 sjlj and ucontext should now work.
954 - IRIX-specific port for libcoro.
955 - swapped order of accept results to match IO::Socket.
956 - changed getsock/peername to sock/peername to match IO::Socket.
957 - fixed a bug that caused segfault when returning to main task
958 under some circumstances.
959 - other bugfixes.
960
961 0.45 Sun Sep 2 02:54:01 CEST 2001
962 - new method Coro::Handle::timeout.
963 - corrected speling of set...name to get...name in Coro::Socket.
964 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
965 - Coro::Handle now supports fileno and FILENO.
966 - added eg/myhttpd, a web server.
967 - Coro::Socket now accepts numeric ports as well as "port(number)"
968 syntax.
969 - moved some scheduling primitives into xs code. more to come.
970 - new simple priority system.
971 - implemented Coro::Event using XS, almost four times faster!
972 - small memory corruption problem fixed (boy that was difficult).
973
974 0.13 Wed Aug 8 16:53:07 CEST 2001
975 - new method Coro::Handle::readline.
976 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
977 (see eg/lwp on how to make LWP non-blocking).
978 - renamed Coro::Socket::new_inet to new.
979 - added Coro::Util, some utility functions.
980 - die/eval now works better (still get annoying (but true ;)
981 "Callback called exit" messages).
982 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
983 - much better Coro::Socket support.
984
985 0.12 Fri Jul 27 04:19:01 CEST 2001
986 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
987 - very experimental stack sharing algorithm. not 100% safe but
988 should work well in practise ;)
989 - added Coro::RWLock.
990 - Coro::Specific now works.
991
992 0.11 Tue Jul 24 22:49:21 CEST 2001
993 - added specialized hack for newer and older linux versions (fast).
994 - renamed Coro::Event::IO to Coro::Handle.
995 - new module Coro::Socket.
996
997 0.1 Tue Jul 24 01:47:53 CEST 2001
998 - release candidate 3. A rather new internal structure :(
999 - the great renaming: Damian Conway gave me a suitable replacement
1000 for yield.
1001 - added Coro/libcoro, a portable coroutine implementation for C
1002 (not even perl-dependent!!!), which can be used by Coro::State.
1003 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
1004 perl-5.6.
1005 - new function Coro::Event::idle.
1006 - the idle process is now overriden
1007 by default in Coro::Event.
1008 - Coro::Channel now enforces the size.
1009 - canceling events now works in all cases.
1010 - Coro state now include $_ and $@.
1011 - yet another bug workaround that I do not really understand :(
1012 - new module Coro::Event::IO (very undocumented) to do
1013 non-blocking i/o.
1014 - performance tuning.
1015
1016 0.08 Thu Jul 19 06:13:25 CEST 2001
1017 - release candidate 2 (new functionality).
1018 - add Coro::State::flush function.
1019 - transfer now supports a flags value (mostly for speed ATM).
1020 - might compile and work in the presence of threads now.
1021 - continuations have a saner syntax.
1022 - no more memleaks.
1023
1024 0.07 Tue Jul 17 17:40:18 CEST 2001
1025 - release candidate 1 ;)
1026 - slightly nicer code.
1027 - fixed a scheduling bug in Coro::Event.
1028
1029 0.06 Tue Jul 17 04:23:24 CEST 2001
1030 - ok, I found the showstopper - the same sub must not be
1031 re-entered in two different coroutines, otherwise => crash. I
1032 see no easy solution to this problem, except by walking the call
1033 chaing and saving/restoring the cv's, which is what I do now.
1034 - memory leaks still latent, especially at thread termination.
1035 - Coro::Event now works (haha).
1036
1037 0.05 Sun Jul 15 17:32:20 CEST 2001
1038 - fixed a few issues in Coro::Event.
1039 - I forgot to include Coro::Event in 0.04 :(:(:(
1040
1041 0.04 Sun Jul 15 05:24:59 CEST 2001
1042 - @_ is now properly localized.
1043 - Coro::State is now easier subclassable.
1044 - Coro::Cont now coroutine-aware.
1045 - Coro::Specific is a low-overhead module to create
1046 coroutine-specific vars.
1047 - Coro::Event provides a simple interface to Event.
1048
1049 0.03 Fri Jul 13 14:51:52 CEST 2001
1050 - transfer() now implemented in XS (beware).
1051 - new module Coro::Cont for really faked continuations.
1052 - big internal architecture changes: Coro::State is now
1053 really low-level and can thus be used to implement other
1054 interesting things, While "Coro::" implements a process-like
1055 model. Still crude and subject to change.
1056 - $_ and $@ are no longer being localized.
1057
1058 0.02 Tue Jul 10 01:38:17 CEST 2001
1059 - implemented "async" attribute.
1060 - $_ and $@ are now localized.
1061 - added Coro::Channel.
1062 - more testcases, still no docs.
1063
1064 0.01 Tue Jul 3 02:18:41 CEST 2001
1065 - original version; copied from Convert::Scalar.
1066