ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.543
Committed: Fri Oct 29 15:25:07 2010 UTC (13 years, 6 months ago) by root
Branch: MAIN
Changes since 1.542: +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: unready_all
5 TODO: myhttpd header parsing
6 TODO: channel->maxsize(newsize)?
7 TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
8
9 - try a different approach on netbsd - netbsd 5 finally has marginally
10 working pthreads, but still broken ucontext/sigaltstack.
11
12 5.24 Sat Oct 23 11:27:12 CEST 2010
13 - port to the EV 4.0 API.
14 - work around bugs in mingw32, making strawberry perl work
15 out of the box.
16 - correctly modify Coro::AIO function prototypes
17 so that they reflect the "no optional parameters" rule.
18 - "ported" libcoro to C++.
19
20 5.23 Mon May 17 18:50:42 CEST 2010
21 - be more resistant to ordering changes when initialising
22 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
23 Waldorf).
24 - document that perl 5.12 deliberately removed support for cloning.
25
26 5.22 Wed Apr 14 03:55:35 CEST 2010
27 - correctly return udnef on errors in Coro::Handle::read/write
28 (testcase by Marc Mims).
29 - convert Coro::Util into a "perl compatibility wrapper" - the functions
30 are less useful now, but are drop-in replacements for existing
31 functions, listing better alternatives in the documentation. This also
32 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
33 with Coro::Util::inet_aton.
34 - do not override $Coro::idle unconditionally in Coro.pm, as other
35 modules could have provided their own idle coro already
36 (for exmaple, Coro::AnyEvent).
37 - fix Coro::Util::gethost* functions.
38 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
39
40 5.21 Wed Dec 16 07:19:51 CET 2009
41 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
42 together.
43 - add some examples on how to combine other event loops with Coro in
44 Coro::AnyEvent, and how to run it (and not to block). Seems to be
45 the most common source of confusion.
46 - try to catch people naively blocking in an event callback.
47 - work around the perl filehandle bug issue in conjunction with
48 older common::sense (as indirectly pointed out by ZSystem).
49 - clarify the "not from signal handlers" section.
50
51 5.2 Sun Oct 4 14:54:24 CEST 2009
52 - Coro::Storable destroyed the prototypes of the functions it wrapped.
53 - export rouse_cb and rouse_wait by default now.
54 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
55 - new method $state->swap_sv.
56 - added section on "windows process emulation" to the manpage,
57 after a not-so-fruitful (nor-friendly) "discussion" with chip
58 salzenberg (discussion implies arguments, but his only arguments
59 were ad-hominems, one wonders why he started it in the first
60 place). I hope this explains it well enough for him to understand,
61 and maybe well enough for others to understand.
62 - use common::sense everywhere now.
63 - idle callbacks are no longer supported, use idle coros instead.
64 - print a thread listing when a deadlock is detected.
65
66 5.17 Sat Aug 22 23:09:31 CEST 2009
67 - work around a bug in the perl debugger causing crashes
68 when running under the debugger by marking _pool_handler as nodebug.
69 - speed up Coro::async considerably.
70 - try some hacks to get netbsd to work "more often" - their broken
71 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
72 - convert Coro to AE 5.0 API.
73
74 5.162 Tue Jul 28 04:04:03 CEST 2009
75 - perl 5.8.2 is now minimum requirement.
76 - skip t/19_handle.t on broken windows perls.
77
78 5.161 Wed Jul 22 04:47:38 CEST 2009
79 - Coro::AnyEvent::poll could have a different prototype when EV was
80 used as backend (analyzed by Tatsuhiko Miyagawa).
81 - Coro::AnyEvent errornously initialised the event loop when loaded,
82 not on demand.
83 - try to workaround rare */t/01_unblock.t failures.
84
85 5.16 Tue Jul 21 01:44:37 CEST 2009
86 - Coro::AnyEvent failed to hook into the event loop
87 when no threads had been readied between detecting
88 the event loop and actually running it.
89 - considerably speed up Coro::Select by taking avdantage
90 of AnyEvent > 4.8 and some other optimisations.
91 - implement paragraph readline mode in Coro::Handle
92 (based on patches by Zsbán Ambrus).
93 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
94 and analyzed by Yasuhiro MATSUMOTO).
95 - clarified libcoro license and copyright.
96 - someone stole my EXTRA_META!!!
97 - implement Coro::Select::patch_pp_sselect and it's brother,
98 for hardcode select overriding.
99
100 5.151 Mon Jul 6 05:41:57 CEST 2009
101 - backport to windows process emulation code again (patch by
102 Yasuhiro MATSUMOTO).
103 - slightly update Coro::MakeMaker.
104
105 5.15 Tue Jun 30 10:28:06 CEST 2009
106 - deprecate Coro::Socket, document how to get ipv6 support via
107 AnyEvent::Socket instead.
108 - implement signal->wait ($cb) interface, similar to semaphores.
109 - work around SvOK not supporting getmagic, so we have to getmagic
110 to test for undef :/ (reported by Matthias Waldorf).
111 - load Coro::AnyEvent in all modules using AnyEvent.
112 - work around perl corrupting our internal data structures,
113 reported by Tokuhiro Matsuno.
114 - enable per-coroutine real and cpu time gathering
115 (Coro::State::enable_times).
116
117 5.14 Wed Jun 24 01:37:48 CEST 2009
118 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
119 as this seems to have been a difficult concept (poll, sleep, idle,
120 idle_upto).
121 - add Coro::AnyEvent::readable/writable functions.
122 - clarify Coro::EV/Event/AnyEvent manpages.
123 - free per-thread global scalars in the thread calling ->cancel, to
124 avoid crashes when $_, $@ etc., are magical but some of those
125 had already been freed.
126 - "unexperimentalise" the callback interface for Coro::Semaphore.
127 - speed up ready queue management/context switching by using a linked
128 list instead of an array (~5-10%).
129 - implement "watch" command in Coro::Debug shells.
130 - for fun, implement time-slicing as an example in the manpage.
131 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
132 Coro::Event, use the normal AnyEvent handling instead of dieing
133 (the same is true for Coro::Handle).
134 - properly document Coro::EV::timed_io_once.
135 - avoid unneccessary ->cancel calls in Coro::Handle.
136 - maybe make it work on mingw32 with win32 backend
137 (based on patch by Yasuhiro Matsumoto).
138
139 5.132 Fri May 29 09:00:39 CEST 2009
140 - do not keep a reference to the argument itself in
141 Coro::Semaphore::guard, as it could change later.
142 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
143 should support a prepare callback.
144
145 5.131 Mon Mar 16 23:20:37 CET 2009
146 - implement and document Coro->suspend, Coro->resume.
147 - fix Coro::Select implementation to not (often) close
148 the passed file descriptors (testcase provided by pippijn).
149
150 5.13 Mon Dec 15 21:51:42 CET 2008
151 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
152 - don't set diehook to C<undef> but instead to NULL, to avoid
153 spurious warnings.
154 - fix a lot of bugs in Coro::SemaphoreSet.
155 - Coro::SemaphoreSet will less often create a semaphore needlessly.
156 - add Coro::SemaphoreSet::count and wait methods.
157 - take advantage of the new Guard module.
158 - deprecate Coro::guard.
159 - try to fix the dreaded 01_unblock tests once more. I hate it when
160 testsuites need more fixing than the code they are supposed to test.
161 - croak in more cases when a required callback isn't resolvable.
162 - fix some minor issues in Coro::State->call/eval.
163 - use current coroutine context instead of a temporary one
164 when temporarily switching to another coroutine.
165 - do not call C-level on_destroy handlers during global destruction,
166 to avoid needless segfaults.
167
168 5.12 Sun Dec 7 13:30:38 CET 2008
169 - add default config for MirOS, which seems to be a bug-to-bug
170 compatible fork of openbsd ("world domination by releasing
171 openbsd cvs before the openbsd folks do it" or so :).
172 - free_padlist did destroy the names pad, not good, but didn't
173 seem to bother perl - this could fix issues such as eval ""
174 inside a function called from multiple coroutines.
175 - use a different method to detect destruction time.
176 - be more careful when freeing padlists just before global
177 destruction.
178 - fixed and expanded the call/cc example.
179 - renamed _terminate to _coro_run.
180 - new method Coro::Channel->shutdown.
181 - try pthreads on openbsd <4.4 (broken sigaltstack, will
182 pthreads fare better?).
183 - be less picky about destroying "a" running coroutine.
184
185 5.11 Tue Nov 25 21:49:05 CET 2008
186 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
187 working code with failed assertions, introducing more bugs than
188 it fixes, requiring elaborate workarounds :(
189
190 5.1 Mon Nov 24 08:54:59 CET 2008
191 - wrote a small introductory tutorial - Coro::Intro.
192 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
193 - Coro::Select did errornously dup the file descriptors
194 and didn't work with all AnyEvent backends.
195 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
196 LWP data transfers.
197 - disassociate c contexts from coro objects - this is agruably more
198 correct, but mostly allows sharing of cctxs between coro and state
199 objects, for added memory savings and speed increases.
200 - bumped $Coro::POOL_RSS up to 32kb by default.
201 - no longer set the optype to OP_CUSTOM, as B::* understandably
202 doesn't like this very much (and we *are* a type of entersub).
203 - implement state cloning, just to prove that call/cc can be done.
204 - automatically load Coro::AnyEvent in Coro::Handle.
205 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
206 do not have this method (and don't need it either).
207 - speed up generic anyevent methods in Coro::Handle by using rouse
208 callbacks.
209 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
210 others. It also makes the debugger happier, as you can trace
211 through the idle threads now.
212 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
213 - eg/event was pretty broken.
214 - better 5.8.6 compatibility.
215
216 5.0 Thu Nov 20 10:35:05 CET 2008
217 - NEW ARCHITECTURE: use the latest 4.x version if you experience
218 stability issues.
219 - bump API version to 7 - all dependents must be recompiled.
220 - removed timed_* functions - they were not being used anyways
221 and should be replaced by a more generic mechanism -
222 and were annoying to support anyways :)
223 - removed SemaphoreSet's waiter method - use sem method instead.
224 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
225 - async_pool did free a scalar value twice
226 ("Attempt to unreference...").
227 - fix a longstanding bug where calling terminate on a coro that
228 was waiting for a semaphore that was just becoming available
229 would cause a deadlock (semaphore would get into a state where
230 it was available but waiters were still blocked).
231 - calling throw on a coroutine that is waiting for a semaphore will
232 no longer make it acquire the semaphore (and thus leak a count).
233 - perl's process emulation is now not even theoretically supported
234 anymore.
235 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
236 conversion of callback-style to blocking-style.
237 - new methods $coro->schedule_to and ->cede_to, to specifically
238 schedule or cede to a specific coroutine.
239 - new function Coro::Semaphore::wait.
240 - use named constants in Coro::Channel (Richard Hundt).
241 - directly patch the entersub opcode calling SLF functions (cede,
242 transfer and so on). this does speed up context switching, but
243 more importanly, it frees us from the hardcoded behaviour of
244 entersub, so we might actually be able to return something from
245 those functions and atcually create new ones.
246 - take advantage of __builtin_frame_address on gcc.
247 - expose THX in coroapi (not sure whether this was a wise decision,
248 as "threaded" perls are running at half speed anyways).
249 - implement execute_slf (schedule-like-function) interface that makes
250 it possible to implement schedule-like-functions in XS.
251 - use new SLF interface to massively speed up Coro::EV by roughly a
252 factor of two.
253 - used new SLF interface to massively speed up Coro::Semaphore by a
254 factor of three.
255 - used new SLF interface to speed up Coro::AIO by roughly a factor of
256 four and reduce its memory usage considerably.
257 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
258 for a nice speedup and vastly more correct behaviour. Also implement
259 a new method "sem" to get at the underlying semaphore object.
260 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
261 (in comparison) 20-40% speedup.
262 - used new SLF interface to reimplement Coro::Signal gaining
263 some unknown (because I was too lazy), but certain, speedup, and also
264 making signals reliable for the first time.
265 - used new SLF interface and other optimisations to speed up async_pool
266 by a factor of two. It also doesn't rely on perl's exception mechanism
267 to exit anymore. The overhead for terminating an async_pool, coro over
268 a normal async is now very small.
269 - sped up coroutine creation/destruction by 40%.
270 - forgot to include Coro/libcoro/README in the dist for all these years.
271 - work around a freebsd pthreads bug (manual testcancel is required as
272 pthread_cond_wait isn't a cancellation point on freebsd).
273 - use new rouse functions to speed up and simplify Coro::BDB.
274 - make "prefer perl native functions" work with threaded perls.
275 - condense Coro::Debug ps output, hint at v and w flags.
276 - (libcoro) lots of minor cleanups and portability improvements.
277
278 4.914 Wed Nov 19 12:54:18 CET 2008
279 - fix a disastrous bug in the readline optimisation
280 introduced in 4.801.
281
282 4.913 Sat Nov 15 07:58:28 CET 2008
283 - async_pool did free a scalar value twice
284 ("Attempt to unreference...").
285
286 4.912 Thu Nov 13 18:31:23 CET 2008
287 - minor cleanups.
288 - use much larger stacks on linux and perl < 5.8.8.
289 - Coro::Debug::new_unix_server did not unlink the socket
290 when destroyed.
291
292 4.911 Tue Nov 11 04:26:17 CET 2008
293 - "port" to threaded perls.
294
295 4.91 Mon Nov 10 05:36:38 CET 2008
296 - the ->throw exception object no longer leaks.
297 - creating a new cctx leaked a scopestack entry (memleak).
298 - new coroutines didn't get created with a zero flags field
299 (unknown impact).
300 - calling ->throw on a not-yet-started coroutine should now work
301 instead of being ignored.
302 - ->throw is now supported on Coro::State objects.
303 - clean up cctx creation code a bit.
304 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
305
306 4.9 Sat Nov 8 17:45:27 CET 2008
307 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
308 - (libcoro) no longer leak threads in the experimental pthread backend,
309 also speed it up considerably.
310 - (libcoro) do not rely on makecontext passing void *'s unscathed.
311 - fix compiletime dependencies on libcoro in the Makefile.
312 - cctx_count wasn't always updated properly.
313 - Coro::State::cctx_stacksize wasn't applied correctly.
314 - new function Coro::State::cctx_max_idle.
315 - the default max number of idle C contexts is now 4.
316 - (libcoro) try harder to get _setjmp/_longjmp.
317 - (libcoro) cleanup and extend the libcoro API to officially
318 allow the creation of empty source contexts.
319 - very experimental workaround for the totally broken netbsd platform.
320 - tried to hack around openbsd bugs.
321
322 4.804 Wed Nov 5 16:36:00 CET 2008
323 - Coro::Debug::new_unix_server would not create a non-blocking listening
324 socket, sometimes causing freezes.
325 - (libcoro) fix misaligned stack points for setjmp and assembly
326 methods, which can cause crashes on x86/x86_64 with a sufficiently
327 aggressive compiler.
328 - new function: Coro::Debug::new_tcp_server.
329 - move ->throw into the Coro class because it only works on coro objects.
330
331 4.803 Mon Nov 3 17:16:12 CET 2008
332 - (libcoro) use a global asm statement to become independent of gcc
333 otpimisations for CORO_ASM (thanks to pippijn for the idea).
334 - try to workaround yet another broken bsd, this time dragonfly.
335
336 4.802 Thu Oct 30 10:56:12 CET 2008
337 - support -fno-omit-frame-pointer on x86 with the assembly method.
338 - tune 01_unblock.t tests a bit.
339
340 4.801 Wed Oct 22 18:33:37 CEST 2008
341 - improve readline speed for very long "lines".
342 - backport to 5.8.8.
343
344 4.8 Thu Oct 2 13:34:40 CEST 2008
345 - new function Coro::AIO::aio_wait.
346 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
347 - greatly speed up and reduce memory usage of Coro::AIO requests.
348 - implement some other µ-optimisations.
349
350 4.749 Mon Sep 29 14:40:12 CEST 2008
351 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
352
353 4.748 Sat Sep 27 14:03:19 CEST 2008
354 - implement, but do not document, PerlIO::cede(granularity).
355 - Coro::Storable forgot to wrap Storable::pstore.
356 - work around the multitude of leaks and memory corruption
357 bugs in PerlIO::via by using our own C-level perliol. As a side
358 effect, Coro::Storable is now much, much, much faster.
359
360 4.747 Tue Sep 23 01:59:41 CEST 2008
361 - fix a per-cv memleak (one empty array was leaked per
362 code reference).
363 - avoid a crash in coro->call|rss when the coroutine was already
364 destroyed (most noticably when using Coro::Debug::ps :)
365 - also protect *Storable::FILE.
366 - push up default storable granularity to 20ms.
367
368 4.746 Sun Sep 21 03:22:20 CEST 2008
369 - be more insistent on locking Storable against reentrancy
370 in Coro::Storable.
371 - move swap_def?v and throw to Coro::State, as documented.
372
373 4.745 Thu Jul 24 00:14:38 CEST 2008
374 - remove debugging code related to MgPV_nolen_const, also try to
375 make it compile with perl 5.8.6 (yes, apple apparently loves
376 outdated software). Reported by John S.
377
378 4.744 Tue Jul 8 22:06:35 CEST 2008
379 - correctly provide default DIE/WARN handlers as documented.
380 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
381 implementations inside perl work fine for us.
382
383 4.743 Mon Jun 16 00:21:57 CEST 2008
384 - when using Coro::EV without running EV::loop it could
385 result in busy-waiting for events, this has been fixed.
386 - reduce codesize and improve performance by using EV_DEFAULT_UC.
387
388 4.742 Sat May 31 14:10:21 CEST 2008
389 - implement a workaround for (some) perl <5.8.8 versions.
390 - require EV 3.3+.
391
392 4.741 Fri May 30 23:33:09 CEST 2008
393 - tell netbsd how utterly broken their imitation of an OS is
394 and refuse to build by default if pthreads are in use.
395 - switch to "s" method on all bsd's by default, as their ucontext
396 stuff seems just too broken.
397 - fix a bug in Coro::Select.
398
399 4.74 Thu May 29 20:05:31 CEST 2008
400 - do not test Coro::LWP for lack of dependencies.
401 - do not test Coro::Select for lack of working perls.
402
403 4.73 Thu May 29 2008
404 - fix a bug in Coro::EV which would cause it to block despite
405 there being runnable coroutines.
406 - sprinkle "no warnings" freely over everything, also suppress
407 warnings for some other modules.
408 - fix typo in WSAEWOULDBLOCK.
409
410 4.72 Sun May 25 05:14:36 CEST 2008
411 - tweak META.yaml a bit, unfortunately, there is no documented way
412 to have optional dependencies with CPAN. doh :(
413 - avoid running some tests on windows because they would fail due to
414 perl bug (broken fork, broken pipes...).
415 - work around perl on windows bugs where perl returns undocumented
416 error codes for sysread, syswrite etc. by taking advantage
417 of AnyEvent's workaround for that problem.
418 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
419 a common perl implementation bug on windows.
420 - use unix domain sockets in testsuite to work around
421 common perl implementation bugs on widows (they are emulated by
422 tcp sockets on windows. ugh.)
423
424 4.71 Sat May 24 20:01:27 CEST 2008
425 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
426 - take advantage of async name resolution of AnyEvent::Util.
427 - work around brutal inet_aton override in Coro::LWP.
428 - take advantage of the readyhook in Coro::EV, for smoother
429 scheduling.
430
431 4.7 Sun May 11 00:32:19 CEST 2008
432 - completely reworked the Coro manpage.
433 - added Coro::AnyEvent, generic event loop integration.
434 - implement cancel, ready and kill commands in Coro::Debug.
435 - document find_coro in Coro::Debug.
436 - incompatible change: rename has_stack to has_cctx.
437 - Coro::AIO and Coro::BDB no longer force event model detection,
438 use AnyEvent::AIO and AnyEvent::BDB.
439
440 4.6 Sat Apr 26 10:05:14 CEST 2008
441 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
442 by the Coro module. It was a mistake to have it in the first place.
443 - (experimental) support for activestate perl 5.10 (method "w").
444 - (experimental) support for strawberry perl (method "a").
445 - coro_sigelem_set did not return a value although it had to,
446 actual impact unknown.
447
448 4.51 Mon Apr 14 13:28:27 CEST 2008
449 - make it compile again on 5.8.
450
451 4.50 Thu Apr 10 09:43:17 CEST 2008
452 - I did it twice! (see 4.49).
453
454 4.49 Sun Apr 6 21:23:31 CEST 2008
455 - grr, instead of compiling the recent changes on 5.10 only they
456 were compiled on 5.8 only.
457
458 4.48 Sun Apr 6 20:36:46 CEST 2008
459 - allow coroutine switches during eval's under 5.10.x, as it seems
460 the parser is a per-interpreter option now, so this is safe
461 (this might fix the odd crashing bug).
462 - drop support for 5.9.x versions: they are dead, jim.
463
464 4.47 Sun Apr 6 00:37:52 CEST 2008
465 - force cctx allocation on API calls: we know we need to force one
466 and gcc actually manages to confuse our heuristic nowadays,
467 leading to crashes and worse.
468 - document force_cctx.
469
470 4.46 Fri Apr 4 22:05:43 CEST 2008
471 - upgrade libcoro, resulting in pthread-backend (which was only created
472 to fulfill the rules of the programming languages shootout).
473
474 4.45 Thu Mar 13 11:55:36 CET 2008
475 - fix a file leaking bug in eg/mhyttpd that would allow
476 downloading of any file (reported by oesi).
477 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
478 (also add testcase).
479 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
480 apparently having loads of fun with that).
481 - implement $state->swap_defsv, ->swap_defav and document ->throw.
482
483 4.4 Wed Feb 13 16:44:29 CET 2008
484 - only assume .cfi pseudo isns exist on GNU/Linux.
485 - add get_timed method to Coro::Channels.
486 - fixed Coro::Debug manpage.
487 - perl 5.11 compatbility improvement based on patch by
488 Andreas König.
489
490 4.37 Sun Jan 20 11:25:23 CET 2008
491 - soften the check for an active parser for perl 5.10 (avoiding
492 "Coro::State::transfer called while parsing" in many cases).
493
494 4.36 Sun Jan 13 10:53:56 CET 2008
495 - reset diehook when terminating from an async_pool as to not
496 trigger any __DIE__ handlers.
497
498 4.35 Sun Jan 13 04:14:13 CET 2008
499 - "bt" debug command now displays any exceptions
500 from longmess and also skips the innermost
501 stackframes, giving more useufl output.
502 - allow backtraces in coroutines blocked in Coro::EV,
503 at a <1% speed hit.
504 - handle localising of $SIG{__WARN__} and __DIE__
505 properly (with a proper amount of dirty hacking).
506
507 4.34 Sat Dec 22 17:49:53 CET 2007
508 - upgrade to EV version 2.0 API.
509
510 4.33 Mon Dec 17 08:36:12 CET 2007
511 - make Coro::AIO etc. loadable in the absence of EV.
512
513 4.32 Mon Dec 17 07:46:02 CET 2007
514 - majorly improved Coro::Handle's performance with Coro::EV.
515 - implemented the until now mythical Coro::BDB module.
516 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
517
518 4.31 Wed Dec 5 12:32:39 CET 2007
519 - remove warn statement form Coro::Util.
520
521 4.3 Tue Dec 4 20:33:14 CET 2007
522 - calls to the idle function could cause stack corruption
523 when the stack changed.
524 - do no longer rely on the presence of EV::DNS if EV is used
525 (because it is gone), but instead take avdantage of EV::ADNS
526 if available.
527 - add ($) prototypes to all functions in Coro::Storable.
528 - use a conventional (safer) idle callback in Coro::EV.
529 - do accept quickly in Coro::Debug to avoid endless loops.
530
531 4.22 Fri Nov 30 16:04:04 CET 2007
532 - really use optimised versions for Event and EV in Coro::Util
533 and Coro::Handle.
534
535 4.21 Sun Nov 25 10:48:59 CET 2007
536 - fix a spurious memory read.
537 - Coro::EV no longer keeps the eventloop "alive".
538
539 4.2 Fri Nov 9 20:47:05 CET 2007
540 - enable/disable tracing from a new coroutine, not a pooled one.
541 - fix a memleak in Coro::Event.
542 - removed killall call from fork_eval.
543 - made sure store_fd is already loaded so that fork_eval
544 does not have to parse autoload in each subprocess.
545 - only use assembly method if -O switch is in $Config{optimize}.
546 - add (optional) Coro::EV module, so far the best event loop module
547 directly supported by Coro.
548 - if the event model is EV, use EV::DNS to resolve
549 stuff in Coro::Util.
550 - don't get confused by multiple event notifications in Coro::Handle.
551 - initial support for EV (libevent interface).
552 - require Event and EV using configure_requires, to force their existance.
553
554 4.13 Wed Oct 24 07:26:45 CEST 2007
555 - add Coro::Storable::blocking_thaw.
556 - use a vastly more complicated technique to localise
557 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
558 - use a coroutine for the idle callback Coro::Event,
559 instead of running Event in the current coroutine context.
560 This also catches recursive invocations.
561 - actually report fork errors in gethostbyname and inet_aton.
562
563 4.11 Thu Oct 11 02:40:24 CEST 2007
564 - port to threaded perls.
565
566 4.1 Thu Oct 11 02:38:16 CEST 2007
567 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
568 be local to each coro (see Coro::State).
569 - incompatible change: for very deep reasons, cede and cede_notself
570 cannot return anything, so nothing will be returned.
571 - possibly bring back 5.10 compatibility (untested).
572 - work around stupid (and wrong) warning on 5.10 :(.
573 - overlay the saved state over the context stack. This saves
574 a few hundred bytes per coroutine on average and also
575 speeds up context switching a bit.
576 - further tune default stack sizes.
577 - (more) correctly calculate stack usage in coro_rss.
578 - Coro::Storable::blocking_* did not properly lock
579 resulting in races between coroutines.
580 - added Coro::Storable::guard.
581 - stopping to trace a coroutine could destroy the cctx of
582 an unrelated coroutine.
583 - explain the relationship between Perl and C coroutines in
584 more detail in Coro::State.
585 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
586 causing a fork per resolve. This also affected Coro::Socket.
587 - switch to a separate stack in $coro->call/eval to avoid
588 invalidating pointers.
589
590 4.03 Sat Oct 6 21:24:00 CEST 2007
591 - added Coro::throw method.
592 - minor code cleanups.
593
594 4.02 Sat Oct 6 02:36:47 CEST 2007
595 - fix a very minor per-coroutine memleak (a single codereference).
596 - fixed a bug where the currently in-use c context would be freed
597 prematurely (can happen only when programs change the stacksize
598 or use tracing).
599 - tracing can no longer keep a coro alive after it terminated.
600 - do static branch prediction in the common path for gcc. gives
601 about 2-5% speed improvement here.
602
603 4.01 Fri Oct 5 22:10:49 CEST 2007
604 - instead of recreating *a* standard output handle we simply
605 use STDOUT, which is faster and hopefully more robust.
606
607 4.0 Fri Oct 5 12:56:00 CEST 2007
608 - incompatibly changed Coro::Storable::freeze.
609 - major new feature: added Coro::Debug, for interactive coroutine
610 debugging, tracing and much more.
611 - major bug fix: unbelievable, but true: $_, $/ and many other
612 "saved" variables actually weren't being saved. This has been fixed,
613 of course, while increasing performance while losing all the save
614 flags.
615 - save flags are gone, and all the api functions dealing with them.
616 - added Coro::Semaphore::adjust.
617 - added Coro::Util::fork_eval.
618 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
619 - added Coro::killall.
620 - reduce initial stack sizes to allow for "micro-coroutines".
621 - better async_pool resource management, moved parts of async_pool
622 handling to XS (major speed improvement).
623 - actually croak before modifying important data structures.
624 - refuse to transfer while compiling.
625 - possibly support eval EXPR better now.
626 - enable assembly per default on linux+bsd x86+amd64.
627 - all internal members were renamed _something for easier subclassing.
628 - many minor tweaks.
629
630 3.63 Wed May 16 14:10:06 CEST 2007
631 - implement handcoded assembly for x86/amd64 SVR ABI.
632
633 3.62 Fri Apr 27 21:36:06 CEST 2007
634 - upgrade libcoro (which might set unwind info correctly).
635 - change default on linux to setjmp/longjmp.
636
637 3.61 Thu Apr 19 12:36:18 CEST 2007
638 - Coro::Storable caused an endless loop when thawing invalid
639 pst's sometimes.
640 - use a Semaphore in Coro::Storable, as Storable doesn't
641 seem to be reentrant (although it is documented to
642 be threadsafe...).
643 - fix Coro::Signal to bring back the original unreliable
644 but stateful semantics.
645 - fixed a lot of typos in Coro.pm (patch submitted by David
646 Steinbrunner, which applied flawlessly).
647
648 3.6 Sat Apr 14 17:13:31 CEST 2007
649 - added some bugfixes to get eg/myhttpd working again.
650 - added Coro::Storable for often-cede'ing freeze/thaw.
651 - try to do a clean exit when a coroutine calls exit
652 (EXPERIMENTAL).
653 - got rid of indirect call through _coro_init.
654 - updated the partly antique examples in eg/ to
655 work again and be a bit less magic, too.
656 - fixed Coro::Signal semantics to work as documented again.
657
658 3.55 Sun Mar 25 01:20:47 CET 2007
659 - add SAVE_DEFFH to save the default filehandle and enable
660 it by default.
661 - finally move socket-operations from Coro::Socket to Coro::Handle
662 to be able to unblock foreign sockets.
663 - add Coro::State::save_also and guarded_save.
664 - add count accessor to Coro::Semaphore.
665 - add Coro::State::cctx_stacksize.
666 - just for the fun of it, do not rely on implicit context,
667 which can dramatically improve performance, but people
668 using windows-process-emulation perls do not care much about
669 performance.
670
671 3.51 Sun Mar 4 14:18:04 CET 2007
672 - fixed a problem when you weakened references to Coro::State's
673 (patch by Yuval Kogman).
674
675 3.501 Wed Feb 28 12:44:07 CET 2007
676 - rename some global symbols as macosx from hell redefines
677 them without asking.
678
679 3.5 Tue Feb 13 20:22:53 CET 2007
680 - do AnyEvent model detection earlier, avoiding problems
681 caused by first using AnyEvent and later Coro::Event.
682 - implement and document Coro::Event event objects.
683 - fix a potential problem in Coro::Event causing crashes.
684 - initialise PL_comppad when creating a new coroutine,
685 avoids crashes on early coro destruction.
686
687 3.41 Mon Jan 22 19:19:49 CET 2007
688 - readline on Coro::Handle did not support undefined $/,
689 nor did it deliver partial lines on EOF or error.
690 - implement malloc fallback for stack allocation because
691 stupid broken idiotic OSX has a stupid broken
692 idiotic fits-the-whole-os mmap "implementation" and
693 my dick feels longer if Coro is portable even to
694 obsolete platforms.
695
696 3.4 Fri Jan 19 21:52:54 CET 2007
697 - remove t/09_timer.t, as it isn't really testing much
698 but was rather flaky in practise.
699 - async_pool coro would keep arguments and callback alive until
700 it was reused.
701 - cancellation of a coroutine could cause spurious idle calls
702 in cede_notself.
703
704 3.3 Sat Jan 6 03:45:00 CET 2007
705 - implement $coro->on_destroy.
706 - Coro::Event blocking semantics have been changed,
707 documented and - hopefully - improved.
708 - fix nice adding, not subtracting, from priority.
709 - fix ->prio and api_is_ready (patch by Mark Hinds).
710 - fixed an assert ("... == prev__cctx->idle_te")
711 that could errronously trigger.
712 - fix various large and small memleaks.
713 - use a (hopefully) more stable cancel implementation
714 that immediately frees the coroutine data.
715 - cede/cede_notself return a status now.
716 - added Coro::guard function.
717 - added a global coroutine pool for jobs (on my machine,
718 I can create and execute 48k simple coros/s with async,
719 and 128k coros with async_pool).
720 - Coro::AIO now uses the coroutine priority as io priority.
721
722 3.2 Fri Dec 22 05:07:09 CET 2006
723 - improve portability to slightly older perls.
724 - use cleaner coroutine destruction.
725 - simplify configuration for users.
726 - optionally (unrecommended) prefer perl functions over
727 their coro replacements.
728
729 3.11 Tue Dec 5 13:11:24 CET 2006
730 - fixed some bogus assert's, but as perl.h disables assert even
731 without NDEBUG (thank you very much), not too many people should
732 notice (that did include myself). Andreas König noticed, though :)
733 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
734 it. Also noticed by Andreas König :)
735 - save/restore tainted status.
736 - verified to pass the testsuite on my 5.9.5.
737
738 3.1 Mon Dec 4 23:03:40 CET 2006
739 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
740 - incompatible change: transfer flags are now per-state.
741 - give a better error message on deadlock.
742 - document Coro::nready.
743 - enhanced testsuite.
744
745 3.01 Sun Dec 3 23:47:42 CET 2006
746 - forgot to include Coro::Timer.
747
748 3.0 Sun Dec 3 22:57:25 CET 2006
749 - the "FINALLY COMPLETELY STABLE" release (coming soon:
750 the "FAMOUS LAST WORDS" release).
751 - implement a new stack sharing algorithm, which uses a stack
752 pool (size currently hardcoded).
753 - make stack sharing mandatory (it no longer uses a heuristic).
754 - eval/die no longer cause weird problems under heavy use.
755 - Coro::Event could cause livelocks if it was used but
756 no Coro::Event watchers were used.
757 - Coro::Event now uses asynccheck as prepare does not
758 check for changed watchers.
759 - Coro::Event allows multiple waiting coros and will wake up one
760 per event.
761 - Coro::Event should be cleaner and more efficient now.
762 - new utility function Coro::unblock_sub.
763 - document the sad fact that Event is no longer reentrant.
764 - putting a coroutine into the ready queue twice could under
765 some circumstances lead to stack corruption.
766 - minor incompatibility: subclassing Coro::State is supported
767 directly now without going through a _coro_state member.
768 - state/coro switching is much faster now.
769 - very minor optimisations and code/documentation cleanup.
770 - avoid problems due to compiler inlining.
771 - removed timers from Coro::Timer -> use AnyEvent instead.
772 - replaced Coro::idle coroutine by (cleaner) idle handler.
773 - updated to newest libcoro.
774 - implement enhanced support for valgrind.
775 - implement is_ready and return value for ready.
776 - removed Coro::Cont, it was a misnomer (it's generators),
777 and it was somewhat annoying to get it right. Will come back
778 if somebody asks for it :)
779 - many ->wait methods and Coro::Event could return
780 spuriously without the event having happened.
781
782 2.5 Tue Nov 7 12:22:33 CET 2006
783 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
784 use AnyEvent, moved them to Coro/.
785 - added Coro::LWP which contains all the uglyness required to
786 make LWP non-blocking.
787 - should work with perl 5.9.x now (Andreas König made me do it).
788 - fixed another bug in Coro::Select when the timeout was undef.
789 - reuse PL_start_env for all coros, saves some memory per coroutine.
790 - manage PL_top_env differently, hopefully to avoid panic: top_env.
791 - timeout argument was not properly used in Coro::Socket.
792 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
793 - emulate undocumented(!) functionality of IO::Socket required
794 by LWP(!!).
795 - updated eg/lwp to work with newer lwp's.
796 - remove "FATAL: uncaught exception" prefix. Coroutines that die
797 kill the whole process, just as exceptions in the main "coroutine"
798 did already.
799
800 2.1 Wed Nov 1 23:01:13 CET 2006
801 - fix a long-standing bug in Coro::Select where select with
802 zero timeout would instead change the current default filehandle.
803 - use a simpler and hopefully more robust way to clone padlists
804 (uses less memory and a perl function instead of our own).
805 - coro can now create a stack guard area on many architectures.
806 - Coro::AIO properly reexports additional functions from IO::AIO.
807 - updated libcoro with a workaround for OS X,
808 pach and testing by Michael Schwern.
809
810 2.0 Tue Oct 24 05:47:17 CEST 2006
811 - support additional aio requests in Coro::AIO.
812
813 1.9 (never properly released due to a glitch)
814
815 1.8 Thu Feb 2 00:59:06 CET 2006
816 - applied suggested patch by SAMV to avoid problems during stupid
817 mark & sweep gc run.
818 - applied patch by Scott Walters for 5.9.3 compatibility.
819
820 1.7 Tue Dec 27 01:41:58 CET 2005
821 - added Coro::AIO, a thin wrapper around IO::AIO.
822 - improved Makefile.PL explanations.
823
824 1.6 Mon Dec 12 21:30:05 CET 2005
825 - additionally save PM_curpm might fix as-of-yet
826 unseen problems with regex matches being attributed to the
827 wrong package.
828 - add t/10_bugs.t, which currently checks against imho broken
829 perls that use 0.26MB of stack space per Perl_magic_get
830 invocation (newer linuxes) as opposed to the 0.0002MB perl
831 normally uses.
832 - make stacksize configurable for the ultimate debian experience.
833
834 1.51 Mon Dec 12 18:48:36 CET 2005
835 - remove debugging warn() accidentally left in Coro::Select.
836
837 1.5 Tue Nov 29 13:32:44 CET 2005
838 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
839 - fix Coro::Select 'select'.
840 - strict'ify some modules.
841 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
842 - use XSLoader in selected modules.
843 - remove some 5.6 compatibility cruft.
844
845 1.4 Tue Sep 6 00:11:05 CEST 2005
846 - libcoro did not take into account the trampoline on amd64,
847 when 'l'inux method was used.
848
849 1.31 Tue Aug 30 23:31:33 CEST 2005
850 - some portability fixes/workarounds.
851
852 1.3 Sat Aug 20 03:08:56 CEST 2005
853 - no code changes. module seems to work fine.
854 - improve Coro::State docs, remove reference to nonexisting
855 Coro::State::flush method.
856 - no longer autodetect windows, present it as an option instead.
857
858 1.2 Mon May 16 02:00:55 CEST 2005
859 - included libcoro.c earlier in State.xs, might improve portability.
860 - use a faster and possibly more stable Coro::Cont implementation.
861 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
862 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
863 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
864
865 1.11 Thu Mar 3 18:00:52 CET 2005
866 - change of contact address.
867
868 1.1 Tue Feb 22 20:51:16 CET 2005
869 - support [l]inux method on amd64.
870 - allow some context switches while compiling/eval'ing.
871
872 1.0 Tue Aug 10 03:47:27 CEST 2004
873 - Event 0.89 fixes the bug again.
874 - fix the remaining(??) memleak. I didn't free the cache-AV for
875 padlists, so one AV per created anon-sub leaked.
876
877 0.97 Fri May 14 15:23:32 CEST 2004
878 - remove stupid gcc extension that was an experiment only anyway.
879
880 0.96 Thu May 13 18:09:29 CEST 2004
881 - tell people if they have yet another broken Event variant.
882 - made Coro to work _better_ with threads (it seems that, as long
883 as you use Coros in one thread only everything will work fine.
884 Looking at the perl source the problem might be missing locking
885 between Coro and some perl internal routines. That's not easily
886 fixable).
887 - (hopefully) got rid of the static xnext inside transfer.
888
889 0.95 Sun Feb 15 03:20:28 CET 2004
890 - removed Coro::State::flush, implemented a new and much
891 faster caching mechanism by attaching magic to the CV.
892 - a memleak with (real) closures remains.
893 - dramatically reduced size of stacks to 65536 (linux/x86), after
894 observing that even myhttpd never had more than a kilobyte
895 of stack in use (stack is still only physically allocated in
896 pagesize increments for systems suporting mmap).
897 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
898 - fix for threaded perl.
899
900 0.9 Sun Nov 23 04:32:39 CET 2003
901 - fix a few very hard to track down but (of course) stupid
902 memory leaks.
903 - stringify version number to avoid locale problems :(.
904 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
905 hopefully ignoring it "just works".
906
907 0.8 Wed Nov 5 19:38:40 CET 2003
908 - port to perl5.005_03, but only 5.8.x+ is supported!
909 - honor LocalAddr even without LocalPort in Coro::Socket.
910
911 0.7 Tue May 27 03:12:38 CEST 2003
912 - the version jump indicates some level of testing, not gobs
913 of new features.
914 - uh... I found the prompt function in ExtUtils::MakeMaker.
915 Highly correct stuff, that is...
916 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
917 the fix is not well understood by the author, so beware :(.
918
919 0.652 Thu May 8 02:54:46 CEST 2003
920 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
921 cause version 4 doesn't have ucontext.
922 - Benjamin Reed reported that setjmp works fine on darwin,
923 so preselect it.
924
925 0.651 Sat Mar 29 15:00:23 CET 2003
926 - fix a bug in Coro::Handle where some bytes could get lost
927 on reading, reported by jason@nichego.net.
928
929 0.65 Sun Mar 23 00:08:26 CET 2003
930 - added a README.linux-glibc.
931 - new module Coro::Select.
932 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
933 (I never claimed that I know what I am doing ;).
934 - set default to setjmp/longjmp on non-x86-linux.
935
936 0.6 Thu Nov 21 11:09:06 CET 2002
937 - made lazy context switching the default.
938 - fixed the bug where SemaphoreSet::guard
939 was the same as timed_guard and timed_guard was missing.
940 - fix a memleak in Coro::Timer.
941 - reclassified context sharing as not experimental.
942
943 0.534 Sun Apr 14 03:05:12 CEST 2002
944 - fixed backspace => tab in header parsing (myhttpd).
945 - added eg/readline, for elmex.
946 - Coro::Event now calls ->start on first call instead of
947 ->again, which created wrong timeouts for the first ->next.
948 - fixed a bug where Coro::Socket returned a closed socket instead
949 of nothing on ECONNREFUSED and similar error conditions.
950
951 0.533 Mon Feb 18 18:49:40 CET 2002
952 - adapt to finally working Event-0.85.
953 - fixed buggy event initialization (trapped under windows, but how
954 could it possibly have worked anywhere?)
955
956 0.532 Wed Jan 16 02:45:32 CET 2002
957 - removed the reference to pp_entersub, might work on
958 (some) win32 perls now (testsuite works under cygwin,
959 without Event, which freezes).
960
961 0.531 Mon Dec 10 22:18:44 CET 2001
962 - Fixed a bug in SemaphoreSet::timed_down.
963
964 0.53 Tue Nov 27 21:11:13 CET 2001
965 - some tests for Event.
966 - slightly different internal architecture to get refcnt into the object
967 and not the reference. might fix a few bugs, certainly did introduce
968 new ones.
969 - Coro::Timer (independent of Event).
970 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
971
972 0.52 Tue Nov 6 21:36:18 CET 2001
973 - ported to cygwin (trivial).
974 patch by Gerrit P. Haase <gp@familiehaase.de>.
975 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
976
977 0.51 Thu Nov 1 20:39:01 CET 2001
978 - terminate/cancel now work properly (otherwise termination
979 could cause a "next coroutine is not and contains not..."
980 error.
981 - added Coro::Socket::shutdown.
982 - Coro::Event::loop no is the same as Event::loop.
983 - implemented terminate with args + join.
984
985 0.5 Fri Sep 28 16:15:35 CEST 2001
986 - fixed "print" on a Coro::Handle. The print method worked.
987 - small tweaks (seem to reduce memory consumption a lot)
988 in various modules.
989 - splendid use of "no warnings" scattered throughout.
990 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
991
992 0.49 Sun Sep 16 02:42:45 CEST 2001
993 - changed some method calls to function calls for speed
994 inside Coro::Handle.
995 - make Coro::Handle use an array instead of a hash for
996 speed reasons.
997 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
998 sjlj and ucontext should now work.
999 - IRIX-specific port for libcoro.
1000 - swapped order of accept results to match IO::Socket.
1001 - changed getsock/peername to sock/peername to match IO::Socket.
1002 - fixed a bug that caused segfault when returning to main task
1003 under some circumstances.
1004 - other bugfixes.
1005
1006 0.45 Sun Sep 2 02:54:01 CEST 2001
1007 - new method Coro::Handle::timeout.
1008 - corrected speling of set...name to get...name in Coro::Socket.
1009 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
1010 - Coro::Handle now supports fileno and FILENO.
1011 - added eg/myhttpd, a web server.
1012 - Coro::Socket now accepts numeric ports as well as "port(number)"
1013 syntax.
1014 - moved some scheduling primitives into xs code. more to come.
1015 - new simple priority system.
1016 - implemented Coro::Event using XS, almost four times faster!
1017 - small memory corruption problem fixed (boy that was difficult).
1018
1019 0.13 Wed Aug 8 16:53:07 CEST 2001
1020 - new method Coro::Handle::readline.
1021 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
1022 (see eg/lwp on how to make LWP non-blocking).
1023 - renamed Coro::Socket::new_inet to new.
1024 - added Coro::Util, some utility functions.
1025 - die/eval now works better (still get annoying (but true ;)
1026 "Callback called exit" messages).
1027 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
1028 - much better Coro::Socket support.
1029
1030 0.12 Fri Jul 27 04:19:01 CEST 2001
1031 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
1032 - very experimental stack sharing algorithm. not 100% safe but
1033 should work well in practise ;)
1034 - added Coro::RWLock.
1035 - Coro::Specific now works.
1036
1037 0.11 Tue Jul 24 22:49:21 CEST 2001
1038 - added specialized hack for newer and older linux versions (fast).
1039 - renamed Coro::Event::IO to Coro::Handle.
1040 - new module Coro::Socket.
1041
1042 0.1 Tue Jul 24 01:47:53 CEST 2001
1043 - release candidate 3. A rather new internal structure :(
1044 - the great renaming: Damian Conway gave me a suitable replacement
1045 for yield.
1046 - added Coro/libcoro, a portable coroutine implementation for C
1047 (not even perl-dependent!!!), which can be used by Coro::State.
1048 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
1049 perl-5.6.
1050 - new function Coro::Event::idle.
1051 - the idle process is now overriden
1052 by default in Coro::Event.
1053 - Coro::Channel now enforces the size.
1054 - canceling events now works in all cases.
1055 - Coro state now include $_ and $@.
1056 - yet another bug workaround that I do not really understand :(
1057 - new module Coro::Event::IO (very undocumented) to do
1058 non-blocking i/o.
1059 - performance tuning.
1060
1061 0.08 Thu Jul 19 06:13:25 CEST 2001
1062 - release candidate 2 (new functionality).
1063 - add Coro::State::flush function.
1064 - transfer now supports a flags value (mostly for speed ATM).
1065 - might compile and work in the presence of threads now.
1066 - continuations have a saner syntax.
1067 - no more memleaks.
1068
1069 0.07 Tue Jul 17 17:40:18 CEST 2001
1070 - release candidate 1 ;)
1071 - slightly nicer code.
1072 - fixed a scheduling bug in Coro::Event.
1073
1074 0.06 Tue Jul 17 04:23:24 CEST 2001
1075 - ok, I found the showstopper - the same sub must not be
1076 re-entered in two different coroutines, otherwise => crash. I
1077 see no easy solution to this problem, except by walking the call
1078 chaing and saving/restoring the cv's, which is what I do now.
1079 - memory leaks still latent, especially at thread termination.
1080 - Coro::Event now works (haha).
1081
1082 0.05 Sun Jul 15 17:32:20 CEST 2001
1083 - fixed a few issues in Coro::Event.
1084 - I forgot to include Coro::Event in 0.04 :(:(:(
1085
1086 0.04 Sun Jul 15 05:24:59 CEST 2001
1087 - @_ is now properly localized.
1088 - Coro::State is now easier subclassable.
1089 - Coro::Cont now coroutine-aware.
1090 - Coro::Specific is a low-overhead module to create
1091 coroutine-specific vars.
1092 - Coro::Event provides a simple interface to Event.
1093
1094 0.03 Fri Jul 13 14:51:52 CEST 2001
1095 - transfer() now implemented in XS (beware).
1096 - new module Coro::Cont for really faked continuations.
1097 - big internal architecture changes: Coro::State is now
1098 really low-level and can thus be used to implement other
1099 interesting things, While "Coro::" implements a process-like
1100 model. Still crude and subject to change.
1101 - $_ and $@ are no longer being localized.
1102
1103 0.02 Tue Jul 10 01:38:17 CEST 2001
1104 - implemented "async" attribute.
1105 - $_ and $@ are now localized.
1106 - added Coro::Channel.
1107 - more testcases, still no docs.
1108
1109 0.01 Tue Jul 3 02:18:41 CEST 2001
1110 - original version; copied from Convert::Scalar.
1111