ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.545
Committed: Sat Feb 5 21:20:46 2011 UTC (13 years, 3 months ago) by root
Branch: MAIN
Changes since 1.544: +6 -0 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 Revision history for Perl extension Coro.
2
3 TODO: should explore PerlIO::coroaio (perl leaks like hell).
4 TODO: unready_all
5 TODO: myhttpd header parsing
6 TODO: channel->maxsize(newsize)?
7 TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
8
9 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
10 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
11 method call.
12 - undocument Coro::Timer::sleep and obsolete whole module.
13 - optimise Coro::Timer::timeout memory and cpu usage.
14
15 5.25 Thu Nov 11 01:08:39 CET 2010
16 - try a different approach on netbsd - netbsd 5 finally has marginally
17 working pthreads, but still broken ucontext/sigaltstack.
18 - openbsd 4.8 finally got their act together, Coro works out of the box
19 with asm, setjmp and pthreads (no change, just informational).
20
21 5.24 Sat Oct 23 11:27:12 CEST 2010
22 - port to the EV 4.0 API.
23 - work around bugs in mingw32, making strawberry perl work
24 out of the box.
25 - correctly modify Coro::AIO function prototypes
26 so that they reflect the "no optional parameters" rule.
27 - "ported" libcoro to C++.
28
29 5.23 Mon May 17 18:50:42 CEST 2010
30 - be more resistant to ordering changes when initialising
31 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
32 Waldorf).
33 - document that perl 5.12 deliberately removed support for cloning.
34
35 5.22 Wed Apr 14 03:55:35 CEST 2010
36 - correctly return udnef on errors in Coro::Handle::read/write
37 (testcase by Marc Mims).
38 - convert Coro::Util into a "perl compatibility wrapper" - the functions
39 are less useful now, but are drop-in replacements for existing
40 functions, listing better alternatives in the documentation. This also
41 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
42 with Coro::Util::inet_aton.
43 - do not override $Coro::idle unconditionally in Coro.pm, as other
44 modules could have provided their own idle coro already
45 (for exmaple, Coro::AnyEvent).
46 - fix Coro::Util::gethost* functions.
47 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
48
49 5.21 Wed Dec 16 07:19:51 CET 2009
50 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
51 together.
52 - add some examples on how to combine other event loops with Coro in
53 Coro::AnyEvent, and how to run it (and not to block). Seems to be
54 the most common source of confusion.
55 - try to catch people naively blocking in an event callback.
56 - work around the perl filehandle bug issue in conjunction with
57 older common::sense (as indirectly pointed out by ZSystem).
58 - clarify the "not from signal handlers" section.
59
60 5.2 Sun Oct 4 14:54:24 CEST 2009
61 - Coro::Storable destroyed the prototypes of the functions it wrapped.
62 - export rouse_cb and rouse_wait by default now.
63 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
64 - new method $state->swap_sv.
65 - added section on "windows process emulation" to the manpage,
66 after a not-so-fruitful (nor-friendly) "discussion" with chip
67 salzenberg (discussion implies arguments, but his only arguments
68 were ad-hominems, one wonders why he started it in the first
69 place). I hope this explains it well enough for him to understand,
70 and maybe well enough for others to understand.
71 - use common::sense everywhere now.
72 - idle callbacks are no longer supported, use idle coros instead.
73 - print a thread listing when a deadlock is detected.
74
75 5.17 Sat Aug 22 23:09:31 CEST 2009
76 - work around a bug in the perl debugger causing crashes
77 when running under the debugger by marking _pool_handler as nodebug.
78 - speed up Coro::async considerably.
79 - try some hacks to get netbsd to work "more often" - their broken
80 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
81 - convert Coro to AE 5.0 API.
82
83 5.162 Tue Jul 28 04:04:03 CEST 2009
84 - perl 5.8.2 is now minimum requirement.
85 - skip t/19_handle.t on broken windows perls.
86
87 5.161 Wed Jul 22 04:47:38 CEST 2009
88 - Coro::AnyEvent::poll could have a different prototype when EV was
89 used as backend (analyzed by Tatsuhiko Miyagawa).
90 - Coro::AnyEvent errornously initialised the event loop when loaded,
91 not on demand.
92 - try to workaround rare */t/01_unblock.t failures.
93
94 5.16 Tue Jul 21 01:44:37 CEST 2009
95 - Coro::AnyEvent failed to hook into the event loop
96 when no threads had been readied between detecting
97 the event loop and actually running it.
98 - considerably speed up Coro::Select by taking avdantage
99 of AnyEvent > 4.8 and some other optimisations.
100 - implement paragraph readline mode in Coro::Handle
101 (based on patches by Zsbán Ambrus).
102 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
103 and analyzed by Yasuhiro MATSUMOTO).
104 - clarified libcoro license and copyright.
105 - someone stole my EXTRA_META!!!
106 - implement Coro::Select::patch_pp_sselect and it's brother,
107 for hardcode select overriding.
108
109 5.151 Mon Jul 6 05:41:57 CEST 2009
110 - backport to windows process emulation code again (patch by
111 Yasuhiro MATSUMOTO).
112 - slightly update Coro::MakeMaker.
113
114 5.15 Tue Jun 30 10:28:06 CEST 2009
115 - deprecate Coro::Socket, document how to get ipv6 support via
116 AnyEvent::Socket instead.
117 - implement signal->wait ($cb) interface, similar to semaphores.
118 - work around SvOK not supporting getmagic, so we have to getmagic
119 to test for undef :/ (reported by Matthias Waldorf).
120 - load Coro::AnyEvent in all modules using AnyEvent.
121 - work around perl corrupting our internal data structures,
122 reported by Tokuhiro Matsuno.
123 - enable per-coroutine real and cpu time gathering
124 (Coro::State::enable_times).
125
126 5.14 Wed Jun 24 01:37:48 CEST 2009
127 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
128 as this seems to have been a difficult concept (poll, sleep, idle,
129 idle_upto).
130 - add Coro::AnyEvent::readable/writable functions.
131 - clarify Coro::EV/Event/AnyEvent manpages.
132 - free per-thread global scalars in the thread calling ->cancel, to
133 avoid crashes when $_, $@ etc., are magical but some of those
134 had already been freed.
135 - "unexperimentalise" the callback interface for Coro::Semaphore.
136 - speed up ready queue management/context switching by using a linked
137 list instead of an array (~5-10%).
138 - implement "watch" command in Coro::Debug shells.
139 - for fun, implement time-slicing as an example in the manpage.
140 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
141 Coro::Event, use the normal AnyEvent handling instead of dieing
142 (the same is true for Coro::Handle).
143 - properly document Coro::EV::timed_io_once.
144 - avoid unneccessary ->cancel calls in Coro::Handle.
145 - maybe make it work on mingw32 with win32 backend
146 (based on patch by Yasuhiro Matsumoto).
147
148 5.132 Fri May 29 09:00:39 CEST 2009
149 - do not keep a reference to the argument itself in
150 Coro::Semaphore::guard, as it could change later.
151 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
152 should support a prepare callback.
153
154 5.131 Mon Mar 16 23:20:37 CET 2009
155 - implement and document Coro->suspend, Coro->resume.
156 - fix Coro::Select implementation to not (often) close
157 the passed file descriptors (testcase provided by pippijn).
158
159 5.13 Mon Dec 15 21:51:42 CET 2008
160 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
161 - don't set diehook to C<undef> but instead to NULL, to avoid
162 spurious warnings.
163 - fix a lot of bugs in Coro::SemaphoreSet.
164 - Coro::SemaphoreSet will less often create a semaphore needlessly.
165 - add Coro::SemaphoreSet::count and wait methods.
166 - take advantage of the new Guard module.
167 - deprecate Coro::guard.
168 - try to fix the dreaded 01_unblock tests once more. I hate it when
169 testsuites need more fixing than the code they are supposed to test.
170 - croak in more cases when a required callback isn't resolvable.
171 - fix some minor issues in Coro::State->call/eval.
172 - use current coroutine context instead of a temporary one
173 when temporarily switching to another coroutine.
174 - do not call C-level on_destroy handlers during global destruction,
175 to avoid needless segfaults.
176
177 5.12 Sun Dec 7 13:30:38 CET 2008
178 - add default config for MirOS, which seems to be a bug-to-bug
179 compatible fork of openbsd ("world domination by releasing
180 openbsd cvs before the openbsd folks do it" or so :).
181 - free_padlist did destroy the names pad, not good, but didn't
182 seem to bother perl - this could fix issues such as eval ""
183 inside a function called from multiple coroutines.
184 - use a different method to detect destruction time.
185 - be more careful when freeing padlists just before global
186 destruction.
187 - fixed and expanded the call/cc example.
188 - renamed _terminate to _coro_run.
189 - new method Coro::Channel->shutdown.
190 - try pthreads on openbsd <4.4 (broken sigaltstack, will
191 pthreads fare better?).
192 - be less picky about destroying "a" running coroutine.
193
194 5.11 Tue Nov 25 21:49:05 CET 2008
195 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
196 working code with failed assertions, introducing more bugs than
197 it fixes, requiring elaborate workarounds :(
198
199 5.1 Mon Nov 24 08:54:59 CET 2008
200 - wrote a small introductory tutorial - Coro::Intro.
201 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
202 - Coro::Select did errornously dup the file descriptors
203 and didn't work with all AnyEvent backends.
204 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
205 LWP data transfers.
206 - disassociate c contexts from coro objects - this is agruably more
207 correct, but mostly allows sharing of cctxs between coro and state
208 objects, for added memory savings and speed increases.
209 - bumped $Coro::POOL_RSS up to 32kb by default.
210 - no longer set the optype to OP_CUSTOM, as B::* understandably
211 doesn't like this very much (and we *are* a type of entersub).
212 - implement state cloning, just to prove that call/cc can be done.
213 - automatically load Coro::AnyEvent in Coro::Handle.
214 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
215 do not have this method (and don't need it either).
216 - speed up generic anyevent methods in Coro::Handle by using rouse
217 callbacks.
218 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
219 others. It also makes the debugger happier, as you can trace
220 through the idle threads now.
221 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
222 - eg/event was pretty broken.
223 - better 5.8.6 compatibility.
224
225 5.0 Thu Nov 20 10:35:05 CET 2008
226 - NEW ARCHITECTURE: use the latest 4.x version if you experience
227 stability issues.
228 - bump API version to 7 - all dependents must be recompiled.
229 - removed timed_* functions - they were not being used anyways
230 and should be replaced by a more generic mechanism -
231 and were annoying to support anyways :)
232 - removed SemaphoreSet's waiter method - use sem method instead.
233 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
234 - async_pool did free a scalar value twice
235 ("Attempt to unreference...").
236 - fix a longstanding bug where calling terminate on a coro that
237 was waiting for a semaphore that was just becoming available
238 would cause a deadlock (semaphore would get into a state where
239 it was available but waiters were still blocked).
240 - calling throw on a coroutine that is waiting for a semaphore will
241 no longer make it acquire the semaphore (and thus leak a count).
242 - perl's process emulation is now not even theoretically supported
243 anymore.
244 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
245 conversion of callback-style to blocking-style.
246 - new methods $coro->schedule_to and ->cede_to, to specifically
247 schedule or cede to a specific coroutine.
248 - new function Coro::Semaphore::wait.
249 - use named constants in Coro::Channel (Richard Hundt).
250 - directly patch the entersub opcode calling SLF functions (cede,
251 transfer and so on). this does speed up context switching, but
252 more importanly, it frees us from the hardcoded behaviour of
253 entersub, so we might actually be able to return something from
254 those functions and atcually create new ones.
255 - take advantage of __builtin_frame_address on gcc.
256 - expose THX in coroapi (not sure whether this was a wise decision,
257 as "threaded" perls are running at half speed anyways).
258 - implement execute_slf (schedule-like-function) interface that makes
259 it possible to implement schedule-like-functions in XS.
260 - use new SLF interface to massively speed up Coro::EV by roughly a
261 factor of two.
262 - used new SLF interface to massively speed up Coro::Semaphore by a
263 factor of three.
264 - used new SLF interface to speed up Coro::AIO by roughly a factor of
265 four and reduce its memory usage considerably.
266 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
267 for a nice speedup and vastly more correct behaviour. Also implement
268 a new method "sem" to get at the underlying semaphore object.
269 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
270 (in comparison) 20-40% speedup.
271 - used new SLF interface to reimplement Coro::Signal gaining
272 some unknown (because I was too lazy), but certain, speedup, and also
273 making signals reliable for the first time.
274 - used new SLF interface and other optimisations to speed up async_pool
275 by a factor of two. It also doesn't rely on perl's exception mechanism
276 to exit anymore. The overhead for terminating an async_pool, coro over
277 a normal async is now very small.
278 - sped up coroutine creation/destruction by 40%.
279 - forgot to include Coro/libcoro/README in the dist for all these years.
280 - work around a freebsd pthreads bug (manual testcancel is required as
281 pthread_cond_wait isn't a cancellation point on freebsd).
282 - use new rouse functions to speed up and simplify Coro::BDB.
283 - make "prefer perl native functions" work with threaded perls.
284 - condense Coro::Debug ps output, hint at v and w flags.
285 - (libcoro) lots of minor cleanups and portability improvements.
286
287 4.914 Wed Nov 19 12:54:18 CET 2008
288 - fix a disastrous bug in the readline optimisation
289 introduced in 4.801.
290
291 4.913 Sat Nov 15 07:58:28 CET 2008
292 - async_pool did free a scalar value twice
293 ("Attempt to unreference...").
294
295 4.912 Thu Nov 13 18:31:23 CET 2008
296 - minor cleanups.
297 - use much larger stacks on linux and perl < 5.8.8.
298 - Coro::Debug::new_unix_server did not unlink the socket
299 when destroyed.
300
301 4.911 Tue Nov 11 04:26:17 CET 2008
302 - "port" to threaded perls.
303
304 4.91 Mon Nov 10 05:36:38 CET 2008
305 - the ->throw exception object no longer leaks.
306 - creating a new cctx leaked a scopestack entry (memleak).
307 - new coroutines didn't get created with a zero flags field
308 (unknown impact).
309 - calling ->throw on a not-yet-started coroutine should now work
310 instead of being ignored.
311 - ->throw is now supported on Coro::State objects.
312 - clean up cctx creation code a bit.
313 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
314
315 4.9 Sat Nov 8 17:45:27 CET 2008
316 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
317 - (libcoro) no longer leak threads in the experimental pthread backend,
318 also speed it up considerably.
319 - (libcoro) do not rely on makecontext passing void *'s unscathed.
320 - fix compiletime dependencies on libcoro in the Makefile.
321 - cctx_count wasn't always updated properly.
322 - Coro::State::cctx_stacksize wasn't applied correctly.
323 - new function Coro::State::cctx_max_idle.
324 - the default max number of idle C contexts is now 4.
325 - (libcoro) try harder to get _setjmp/_longjmp.
326 - (libcoro) cleanup and extend the libcoro API to officially
327 allow the creation of empty source contexts.
328 - very experimental workaround for the totally broken netbsd platform.
329 - tried to hack around openbsd bugs.
330
331 4.804 Wed Nov 5 16:36:00 CET 2008
332 - Coro::Debug::new_unix_server would not create a non-blocking listening
333 socket, sometimes causing freezes.
334 - (libcoro) fix misaligned stack points for setjmp and assembly
335 methods, which can cause crashes on x86/x86_64 with a sufficiently
336 aggressive compiler.
337 - new function: Coro::Debug::new_tcp_server.
338 - move ->throw into the Coro class because it only works on coro objects.
339
340 4.803 Mon Nov 3 17:16:12 CET 2008
341 - (libcoro) use a global asm statement to become independent of gcc
342 otpimisations for CORO_ASM (thanks to pippijn for the idea).
343 - try to workaround yet another broken bsd, this time dragonfly.
344
345 4.802 Thu Oct 30 10:56:12 CET 2008
346 - support -fno-omit-frame-pointer on x86 with the assembly method.
347 - tune 01_unblock.t tests a bit.
348
349 4.801 Wed Oct 22 18:33:37 CEST 2008
350 - improve readline speed for very long "lines".
351 - backport to 5.8.8.
352
353 4.8 Thu Oct 2 13:34:40 CEST 2008
354 - new function Coro::AIO::aio_wait.
355 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
356 - greatly speed up and reduce memory usage of Coro::AIO requests.
357 - implement some other µ-optimisations.
358
359 4.749 Mon Sep 29 14:40:12 CEST 2008
360 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
361
362 4.748 Sat Sep 27 14:03:19 CEST 2008
363 - implement, but do not document, PerlIO::cede(granularity).
364 - Coro::Storable forgot to wrap Storable::pstore.
365 - work around the multitude of leaks and memory corruption
366 bugs in PerlIO::via by using our own C-level perliol. As a side
367 effect, Coro::Storable is now much, much, much faster.
368
369 4.747 Tue Sep 23 01:59:41 CEST 2008
370 - fix a per-cv memleak (one empty array was leaked per
371 code reference).
372 - avoid a crash in coro->call|rss when the coroutine was already
373 destroyed (most noticably when using Coro::Debug::ps :)
374 - also protect *Storable::FILE.
375 - push up default storable granularity to 20ms.
376
377 4.746 Sun Sep 21 03:22:20 CEST 2008
378 - be more insistent on locking Storable against reentrancy
379 in Coro::Storable.
380 - move swap_def?v and throw to Coro::State, as documented.
381
382 4.745 Thu Jul 24 00:14:38 CEST 2008
383 - remove debugging code related to MgPV_nolen_const, also try to
384 make it compile with perl 5.8.6 (yes, apple apparently loves
385 outdated software). Reported by John S.
386
387 4.744 Tue Jul 8 22:06:35 CEST 2008
388 - correctly provide default DIE/WARN handlers as documented.
389 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
390 implementations inside perl work fine for us.
391
392 4.743 Mon Jun 16 00:21:57 CEST 2008
393 - when using Coro::EV without running EV::loop it could
394 result in busy-waiting for events, this has been fixed.
395 - reduce codesize and improve performance by using EV_DEFAULT_UC.
396
397 4.742 Sat May 31 14:10:21 CEST 2008
398 - implement a workaround for (some) perl <5.8.8 versions.
399 - require EV 3.3+.
400
401 4.741 Fri May 30 23:33:09 CEST 2008
402 - tell netbsd how utterly broken their imitation of an OS is
403 and refuse to build by default if pthreads are in use.
404 - switch to "s" method on all bsd's by default, as their ucontext
405 stuff seems just too broken.
406 - fix a bug in Coro::Select.
407
408 4.74 Thu May 29 20:05:31 CEST 2008
409 - do not test Coro::LWP for lack of dependencies.
410 - do not test Coro::Select for lack of working perls.
411
412 4.73 Thu May 29 2008
413 - fix a bug in Coro::EV which would cause it to block despite
414 there being runnable coroutines.
415 - sprinkle "no warnings" freely over everything, also suppress
416 warnings for some other modules.
417 - fix typo in WSAEWOULDBLOCK.
418
419 4.72 Sun May 25 05:14:36 CEST 2008
420 - tweak META.yaml a bit, unfortunately, there is no documented way
421 to have optional dependencies with CPAN. doh :(
422 - avoid running some tests on windows because they would fail due to
423 perl bug (broken fork, broken pipes...).
424 - work around perl on windows bugs where perl returns undocumented
425 error codes for sysread, syswrite etc. by taking advantage
426 of AnyEvent's workaround for that problem.
427 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
428 a common perl implementation bug on windows.
429 - use unix domain sockets in testsuite to work around
430 common perl implementation bugs on widows (they are emulated by
431 tcp sockets on windows. ugh.)
432
433 4.71 Sat May 24 20:01:27 CEST 2008
434 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
435 - take advantage of async name resolution of AnyEvent::Util.
436 - work around brutal inet_aton override in Coro::LWP.
437 - take advantage of the readyhook in Coro::EV, for smoother
438 scheduling.
439
440 4.7 Sun May 11 00:32:19 CEST 2008
441 - completely reworked the Coro manpage.
442 - added Coro::AnyEvent, generic event loop integration.
443 - implement cancel, ready and kill commands in Coro::Debug.
444 - document find_coro in Coro::Debug.
445 - incompatible change: rename has_stack to has_cctx.
446 - Coro::AIO and Coro::BDB no longer force event model detection,
447 use AnyEvent::AIO and AnyEvent::BDB.
448
449 4.6 Sat Apr 26 10:05:14 CEST 2008
450 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
451 by the Coro module. It was a mistake to have it in the first place.
452 - (experimental) support for activestate perl 5.10 (method "w").
453 - (experimental) support for strawberry perl (method "a").
454 - coro_sigelem_set did not return a value although it had to,
455 actual impact unknown.
456
457 4.51 Mon Apr 14 13:28:27 CEST 2008
458 - make it compile again on 5.8.
459
460 4.50 Thu Apr 10 09:43:17 CEST 2008
461 - I did it twice! (see 4.49).
462
463 4.49 Sun Apr 6 21:23:31 CEST 2008
464 - grr, instead of compiling the recent changes on 5.10 only they
465 were compiled on 5.8 only.
466
467 4.48 Sun Apr 6 20:36:46 CEST 2008
468 - allow coroutine switches during eval's under 5.10.x, as it seems
469 the parser is a per-interpreter option now, so this is safe
470 (this might fix the odd crashing bug).
471 - drop support for 5.9.x versions: they are dead, jim.
472
473 4.47 Sun Apr 6 00:37:52 CEST 2008
474 - force cctx allocation on API calls: we know we need to force one
475 and gcc actually manages to confuse our heuristic nowadays,
476 leading to crashes and worse.
477 - document force_cctx.
478
479 4.46 Fri Apr 4 22:05:43 CEST 2008
480 - upgrade libcoro, resulting in pthread-backend (which was only created
481 to fulfill the rules of the programming languages shootout).
482
483 4.45 Thu Mar 13 11:55:36 CET 2008
484 - fix a file leaking bug in eg/mhyttpd that would allow
485 downloading of any file (reported by oesi).
486 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
487 (also add testcase).
488 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
489 apparently having loads of fun with that).
490 - implement $state->swap_defsv, ->swap_defav and document ->throw.
491
492 4.4 Wed Feb 13 16:44:29 CET 2008
493 - only assume .cfi pseudo isns exist on GNU/Linux.
494 - add get_timed method to Coro::Channels.
495 - fixed Coro::Debug manpage.
496 - perl 5.11 compatbility improvement based on patch by
497 Andreas König.
498
499 4.37 Sun Jan 20 11:25:23 CET 2008
500 - soften the check for an active parser for perl 5.10 (avoiding
501 "Coro::State::transfer called while parsing" in many cases).
502
503 4.36 Sun Jan 13 10:53:56 CET 2008
504 - reset diehook when terminating from an async_pool as to not
505 trigger any __DIE__ handlers.
506
507 4.35 Sun Jan 13 04:14:13 CET 2008
508 - "bt" debug command now displays any exceptions
509 from longmess and also skips the innermost
510 stackframes, giving more useufl output.
511 - allow backtraces in coroutines blocked in Coro::EV,
512 at a <1% speed hit.
513 - handle localising of $SIG{__WARN__} and __DIE__
514 properly (with a proper amount of dirty hacking).
515
516 4.34 Sat Dec 22 17:49:53 CET 2007
517 - upgrade to EV version 2.0 API.
518
519 4.33 Mon Dec 17 08:36:12 CET 2007
520 - make Coro::AIO etc. loadable in the absence of EV.
521
522 4.32 Mon Dec 17 07:46:02 CET 2007
523 - majorly improved Coro::Handle's performance with Coro::EV.
524 - implemented the until now mythical Coro::BDB module.
525 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
526
527 4.31 Wed Dec 5 12:32:39 CET 2007
528 - remove warn statement form Coro::Util.
529
530 4.3 Tue Dec 4 20:33:14 CET 2007
531 - calls to the idle function could cause stack corruption
532 when the stack changed.
533 - do no longer rely on the presence of EV::DNS if EV is used
534 (because it is gone), but instead take avdantage of EV::ADNS
535 if available.
536 - add ($) prototypes to all functions in Coro::Storable.
537 - use a conventional (safer) idle callback in Coro::EV.
538 - do accept quickly in Coro::Debug to avoid endless loops.
539
540 4.22 Fri Nov 30 16:04:04 CET 2007
541 - really use optimised versions for Event and EV in Coro::Util
542 and Coro::Handle.
543
544 4.21 Sun Nov 25 10:48:59 CET 2007
545 - fix a spurious memory read.
546 - Coro::EV no longer keeps the eventloop "alive".
547
548 4.2 Fri Nov 9 20:47:05 CET 2007
549 - enable/disable tracing from a new coroutine, not a pooled one.
550 - fix a memleak in Coro::Event.
551 - removed killall call from fork_eval.
552 - made sure store_fd is already loaded so that fork_eval
553 does not have to parse autoload in each subprocess.
554 - only use assembly method if -O switch is in $Config{optimize}.
555 - add (optional) Coro::EV module, so far the best event loop module
556 directly supported by Coro.
557 - if the event model is EV, use EV::DNS to resolve
558 stuff in Coro::Util.
559 - don't get confused by multiple event notifications in Coro::Handle.
560 - initial support for EV (libevent interface).
561 - require Event and EV using configure_requires, to force their existance.
562
563 4.13 Wed Oct 24 07:26:45 CEST 2007
564 - add Coro::Storable::blocking_thaw.
565 - use a vastly more complicated technique to localise
566 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
567 - use a coroutine for the idle callback Coro::Event,
568 instead of running Event in the current coroutine context.
569 This also catches recursive invocations.
570 - actually report fork errors in gethostbyname and inet_aton.
571
572 4.11 Thu Oct 11 02:40:24 CEST 2007
573 - port to threaded perls.
574
575 4.1 Thu Oct 11 02:38:16 CEST 2007
576 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
577 be local to each coro (see Coro::State).
578 - incompatible change: for very deep reasons, cede and cede_notself
579 cannot return anything, so nothing will be returned.
580 - possibly bring back 5.10 compatibility (untested).
581 - work around stupid (and wrong) warning on 5.10 :(.
582 - overlay the saved state over the context stack. This saves
583 a few hundred bytes per coroutine on average and also
584 speeds up context switching a bit.
585 - further tune default stack sizes.
586 - (more) correctly calculate stack usage in coro_rss.
587 - Coro::Storable::blocking_* did not properly lock
588 resulting in races between coroutines.
589 - added Coro::Storable::guard.
590 - stopping to trace a coroutine could destroy the cctx of
591 an unrelated coroutine.
592 - explain the relationship between Perl and C coroutines in
593 more detail in Coro::State.
594 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
595 causing a fork per resolve. This also affected Coro::Socket.
596 - switch to a separate stack in $coro->call/eval to avoid
597 invalidating pointers.
598
599 4.03 Sat Oct 6 21:24:00 CEST 2007
600 - added Coro::throw method.
601 - minor code cleanups.
602
603 4.02 Sat Oct 6 02:36:47 CEST 2007
604 - fix a very minor per-coroutine memleak (a single codereference).
605 - fixed a bug where the currently in-use c context would be freed
606 prematurely (can happen only when programs change the stacksize
607 or use tracing).
608 - tracing can no longer keep a coro alive after it terminated.
609 - do static branch prediction in the common path for gcc. gives
610 about 2-5% speed improvement here.
611
612 4.01 Fri Oct 5 22:10:49 CEST 2007
613 - instead of recreating *a* standard output handle we simply
614 use STDOUT, which is faster and hopefully more robust.
615
616 4.0 Fri Oct 5 12:56:00 CEST 2007
617 - incompatibly changed Coro::Storable::freeze.
618 - major new feature: added Coro::Debug, for interactive coroutine
619 debugging, tracing and much more.
620 - major bug fix: unbelievable, but true: $_, $/ and many other
621 "saved" variables actually weren't being saved. This has been fixed,
622 of course, while increasing performance while losing all the save
623 flags.
624 - save flags are gone, and all the api functions dealing with them.
625 - added Coro::Semaphore::adjust.
626 - added Coro::Util::fork_eval.
627 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
628 - added Coro::killall.
629 - reduce initial stack sizes to allow for "micro-coroutines".
630 - better async_pool resource management, moved parts of async_pool
631 handling to XS (major speed improvement).
632 - actually croak before modifying important data structures.
633 - refuse to transfer while compiling.
634 - possibly support eval EXPR better now.
635 - enable assembly per default on linux+bsd x86+amd64.
636 - all internal members were renamed _something for easier subclassing.
637 - many minor tweaks.
638
639 3.63 Wed May 16 14:10:06 CEST 2007
640 - implement handcoded assembly for x86/amd64 SVR ABI.
641
642 3.62 Fri Apr 27 21:36:06 CEST 2007
643 - upgrade libcoro (which might set unwind info correctly).
644 - change default on linux to setjmp/longjmp.
645
646 3.61 Thu Apr 19 12:36:18 CEST 2007
647 - Coro::Storable caused an endless loop when thawing invalid
648 pst's sometimes.
649 - use a Semaphore in Coro::Storable, as Storable doesn't
650 seem to be reentrant (although it is documented to
651 be threadsafe...).
652 - fix Coro::Signal to bring back the original unreliable
653 but stateful semantics.
654 - fixed a lot of typos in Coro.pm (patch submitted by David
655 Steinbrunner, which applied flawlessly).
656
657 3.6 Sat Apr 14 17:13:31 CEST 2007
658 - added some bugfixes to get eg/myhttpd working again.
659 - added Coro::Storable for often-cede'ing freeze/thaw.
660 - try to do a clean exit when a coroutine calls exit
661 (EXPERIMENTAL).
662 - got rid of indirect call through _coro_init.
663 - updated the partly antique examples in eg/ to
664 work again and be a bit less magic, too.
665 - fixed Coro::Signal semantics to work as documented again.
666
667 3.55 Sun Mar 25 01:20:47 CET 2007
668 - add SAVE_DEFFH to save the default filehandle and enable
669 it by default.
670 - finally move socket-operations from Coro::Socket to Coro::Handle
671 to be able to unblock foreign sockets.
672 - add Coro::State::save_also and guarded_save.
673 - add count accessor to Coro::Semaphore.
674 - add Coro::State::cctx_stacksize.
675 - just for the fun of it, do not rely on implicit context,
676 which can dramatically improve performance, but people
677 using windows-process-emulation perls do not care much about
678 performance.
679
680 3.51 Sun Mar 4 14:18:04 CET 2007
681 - fixed a problem when you weakened references to Coro::State's
682 (patch by Yuval Kogman).
683
684 3.501 Wed Feb 28 12:44:07 CET 2007
685 - rename some global symbols as macosx from hell redefines
686 them without asking.
687
688 3.5 Tue Feb 13 20:22:53 CET 2007
689 - do AnyEvent model detection earlier, avoiding problems
690 caused by first using AnyEvent and later Coro::Event.
691 - implement and document Coro::Event event objects.
692 - fix a potential problem in Coro::Event causing crashes.
693 - initialise PL_comppad when creating a new coroutine,
694 avoids crashes on early coro destruction.
695
696 3.41 Mon Jan 22 19:19:49 CET 2007
697 - readline on Coro::Handle did not support undefined $/,
698 nor did it deliver partial lines on EOF or error.
699 - implement malloc fallback for stack allocation because
700 stupid broken idiotic OSX has a stupid broken
701 idiotic fits-the-whole-os mmap "implementation" and
702 my dick feels longer if Coro is portable even to
703 obsolete platforms.
704
705 3.4 Fri Jan 19 21:52:54 CET 2007
706 - remove t/09_timer.t, as it isn't really testing much
707 but was rather flaky in practise.
708 - async_pool coro would keep arguments and callback alive until
709 it was reused.
710 - cancellation of a coroutine could cause spurious idle calls
711 in cede_notself.
712
713 3.3 Sat Jan 6 03:45:00 CET 2007
714 - implement $coro->on_destroy.
715 - Coro::Event blocking semantics have been changed,
716 documented and - hopefully - improved.
717 - fix nice adding, not subtracting, from priority.
718 - fix ->prio and api_is_ready (patch by Mark Hinds).
719 - fixed an assert ("... == prev__cctx->idle_te")
720 that could errronously trigger.
721 - fix various large and small memleaks.
722 - use a (hopefully) more stable cancel implementation
723 that immediately frees the coroutine data.
724 - cede/cede_notself return a status now.
725 - added Coro::guard function.
726 - added a global coroutine pool for jobs (on my machine,
727 I can create and execute 48k simple coros/s with async,
728 and 128k coros with async_pool).
729 - Coro::AIO now uses the coroutine priority as io priority.
730
731 3.2 Fri Dec 22 05:07:09 CET 2006
732 - improve portability to slightly older perls.
733 - use cleaner coroutine destruction.
734 - simplify configuration for users.
735 - optionally (unrecommended) prefer perl functions over
736 their coro replacements.
737
738 3.11 Tue Dec 5 13:11:24 CET 2006
739 - fixed some bogus assert's, but as perl.h disables assert even
740 without NDEBUG (thank you very much), not too many people should
741 notice (that did include myself). Andreas König noticed, though :)
742 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
743 it. Also noticed by Andreas König :)
744 - save/restore tainted status.
745 - verified to pass the testsuite on my 5.9.5.
746
747 3.1 Mon Dec 4 23:03:40 CET 2006
748 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
749 - incompatible change: transfer flags are now per-state.
750 - give a better error message on deadlock.
751 - document Coro::nready.
752 - enhanced testsuite.
753
754 3.01 Sun Dec 3 23:47:42 CET 2006
755 - forgot to include Coro::Timer.
756
757 3.0 Sun Dec 3 22:57:25 CET 2006
758 - the "FINALLY COMPLETELY STABLE" release (coming soon:
759 the "FAMOUS LAST WORDS" release).
760 - implement a new stack sharing algorithm, which uses a stack
761 pool (size currently hardcoded).
762 - make stack sharing mandatory (it no longer uses a heuristic).
763 - eval/die no longer cause weird problems under heavy use.
764 - Coro::Event could cause livelocks if it was used but
765 no Coro::Event watchers were used.
766 - Coro::Event now uses asynccheck as prepare does not
767 check for changed watchers.
768 - Coro::Event allows multiple waiting coros and will wake up one
769 per event.
770 - Coro::Event should be cleaner and more efficient now.
771 - new utility function Coro::unblock_sub.
772 - document the sad fact that Event is no longer reentrant.
773 - putting a coroutine into the ready queue twice could under
774 some circumstances lead to stack corruption.
775 - minor incompatibility: subclassing Coro::State is supported
776 directly now without going through a _coro_state member.
777 - state/coro switching is much faster now.
778 - very minor optimisations and code/documentation cleanup.
779 - avoid problems due to compiler inlining.
780 - removed timers from Coro::Timer -> use AnyEvent instead.
781 - replaced Coro::idle coroutine by (cleaner) idle handler.
782 - updated to newest libcoro.
783 - implement enhanced support for valgrind.
784 - implement is_ready and return value for ready.
785 - removed Coro::Cont, it was a misnomer (it's generators),
786 and it was somewhat annoying to get it right. Will come back
787 if somebody asks for it :)
788 - many ->wait methods and Coro::Event could return
789 spuriously without the event having happened.
790
791 2.5 Tue Nov 7 12:22:33 CET 2006
792 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
793 use AnyEvent, moved them to Coro/.
794 - added Coro::LWP which contains all the uglyness required to
795 make LWP non-blocking.
796 - should work with perl 5.9.x now (Andreas König made me do it).
797 - fixed another bug in Coro::Select when the timeout was undef.
798 - reuse PL_start_env for all coros, saves some memory per coroutine.
799 - manage PL_top_env differently, hopefully to avoid panic: top_env.
800 - timeout argument was not properly used in Coro::Socket.
801 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
802 - emulate undocumented(!) functionality of IO::Socket required
803 by LWP(!!).
804 - updated eg/lwp to work with newer lwp's.
805 - remove "FATAL: uncaught exception" prefix. Coroutines that die
806 kill the whole process, just as exceptions in the main "coroutine"
807 did already.
808
809 2.1 Wed Nov 1 23:01:13 CET 2006
810 - fix a long-standing bug in Coro::Select where select with
811 zero timeout would instead change the current default filehandle.
812 - use a simpler and hopefully more robust way to clone padlists
813 (uses less memory and a perl function instead of our own).
814 - coro can now create a stack guard area on many architectures.
815 - Coro::AIO properly reexports additional functions from IO::AIO.
816 - updated libcoro with a workaround for OS X,
817 pach and testing by Michael Schwern.
818
819 2.0 Tue Oct 24 05:47:17 CEST 2006
820 - support additional aio requests in Coro::AIO.
821
822 1.9 (never properly released due to a glitch)
823
824 1.8 Thu Feb 2 00:59:06 CET 2006
825 - applied suggested patch by SAMV to avoid problems during stupid
826 mark & sweep gc run.
827 - applied patch by Scott Walters for 5.9.3 compatibility.
828
829 1.7 Tue Dec 27 01:41:58 CET 2005
830 - added Coro::AIO, a thin wrapper around IO::AIO.
831 - improved Makefile.PL explanations.
832
833 1.6 Mon Dec 12 21:30:05 CET 2005
834 - additionally save PM_curpm might fix as-of-yet
835 unseen problems with regex matches being attributed to the
836 wrong package.
837 - add t/10_bugs.t, which currently checks against imho broken
838 perls that use 0.26MB of stack space per Perl_magic_get
839 invocation (newer linuxes) as opposed to the 0.0002MB perl
840 normally uses.
841 - make stacksize configurable for the ultimate debian experience.
842
843 1.51 Mon Dec 12 18:48:36 CET 2005
844 - remove debugging warn() accidentally left in Coro::Select.
845
846 1.5 Tue Nov 29 13:32:44 CET 2005
847 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
848 - fix Coro::Select 'select'.
849 - strict'ify some modules.
850 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
851 - use XSLoader in selected modules.
852 - remove some 5.6 compatibility cruft.
853
854 1.4 Tue Sep 6 00:11:05 CEST 2005
855 - libcoro did not take into account the trampoline on amd64,
856 when 'l'inux method was used.
857
858 1.31 Tue Aug 30 23:31:33 CEST 2005
859 - some portability fixes/workarounds.
860
861 1.3 Sat Aug 20 03:08:56 CEST 2005
862 - no code changes. module seems to work fine.
863 - improve Coro::State docs, remove reference to nonexisting
864 Coro::State::flush method.
865 - no longer autodetect windows, present it as an option instead.
866
867 1.2 Mon May 16 02:00:55 CEST 2005
868 - included libcoro.c earlier in State.xs, might improve portability.
869 - use a faster and possibly more stable Coro::Cont implementation.
870 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
871 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
872 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
873
874 1.11 Thu Mar 3 18:00:52 CET 2005
875 - change of contact address.
876
877 1.1 Tue Feb 22 20:51:16 CET 2005
878 - support [l]inux method on amd64.
879 - allow some context switches while compiling/eval'ing.
880
881 1.0 Tue Aug 10 03:47:27 CEST 2004
882 - Event 0.89 fixes the bug again.
883 - fix the remaining(??) memleak. I didn't free the cache-AV for
884 padlists, so one AV per created anon-sub leaked.
885
886 0.97 Fri May 14 15:23:32 CEST 2004
887 - remove stupid gcc extension that was an experiment only anyway.
888
889 0.96 Thu May 13 18:09:29 CEST 2004
890 - tell people if they have yet another broken Event variant.
891 - made Coro to work _better_ with threads (it seems that, as long
892 as you use Coros in one thread only everything will work fine.
893 Looking at the perl source the problem might be missing locking
894 between Coro and some perl internal routines. That's not easily
895 fixable).
896 - (hopefully) got rid of the static xnext inside transfer.
897
898 0.95 Sun Feb 15 03:20:28 CET 2004
899 - removed Coro::State::flush, implemented a new and much
900 faster caching mechanism by attaching magic to the CV.
901 - a memleak with (real) closures remains.
902 - dramatically reduced size of stacks to 65536 (linux/x86), after
903 observing that even myhttpd never had more than a kilobyte
904 of stack in use (stack is still only physically allocated in
905 pagesize increments for systems suporting mmap).
906 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
907 - fix for threaded perl.
908
909 0.9 Sun Nov 23 04:32:39 CET 2003
910 - fix a few very hard to track down but (of course) stupid
911 memory leaks.
912 - stringify version number to avoid locale problems :(.
913 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
914 hopefully ignoring it "just works".
915
916 0.8 Wed Nov 5 19:38:40 CET 2003
917 - port to perl5.005_03, but only 5.8.x+ is supported!
918 - honor LocalAddr even without LocalPort in Coro::Socket.
919
920 0.7 Tue May 27 03:12:38 CEST 2003
921 - the version jump indicates some level of testing, not gobs
922 of new features.
923 - uh... I found the prompt function in ExtUtils::MakeMaker.
924 Highly correct stuff, that is...
925 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
926 the fix is not well understood by the author, so beware :(.
927
928 0.652 Thu May 8 02:54:46 CEST 2003
929 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
930 cause version 4 doesn't have ucontext.
931 - Benjamin Reed reported that setjmp works fine on darwin,
932 so preselect it.
933
934 0.651 Sat Mar 29 15:00:23 CET 2003
935 - fix a bug in Coro::Handle where some bytes could get lost
936 on reading, reported by jason@nichego.net.
937
938 0.65 Sun Mar 23 00:08:26 CET 2003
939 - added a README.linux-glibc.
940 - new module Coro::Select.
941 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
942 (I never claimed that I know what I am doing ;).
943 - set default to setjmp/longjmp on non-x86-linux.
944
945 0.6 Thu Nov 21 11:09:06 CET 2002
946 - made lazy context switching the default.
947 - fixed the bug where SemaphoreSet::guard
948 was the same as timed_guard and timed_guard was missing.
949 - fix a memleak in Coro::Timer.
950 - reclassified context sharing as not experimental.
951
952 0.534 Sun Apr 14 03:05:12 CEST 2002
953 - fixed backspace => tab in header parsing (myhttpd).
954 - added eg/readline, for elmex.
955 - Coro::Event now calls ->start on first call instead of
956 ->again, which created wrong timeouts for the first ->next.
957 - fixed a bug where Coro::Socket returned a closed socket instead
958 of nothing on ECONNREFUSED and similar error conditions.
959
960 0.533 Mon Feb 18 18:49:40 CET 2002
961 - adapt to finally working Event-0.85.
962 - fixed buggy event initialization (trapped under windows, but how
963 could it possibly have worked anywhere?)
964
965 0.532 Wed Jan 16 02:45:32 CET 2002
966 - removed the reference to pp_entersub, might work on
967 (some) win32 perls now (testsuite works under cygwin,
968 without Event, which freezes).
969
970 0.531 Mon Dec 10 22:18:44 CET 2001
971 - Fixed a bug in SemaphoreSet::timed_down.
972
973 0.53 Tue Nov 27 21:11:13 CET 2001
974 - some tests for Event.
975 - slightly different internal architecture to get refcnt into the object
976 and not the reference. might fix a few bugs, certainly did introduce
977 new ones.
978 - Coro::Timer (independent of Event).
979 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
980
981 0.52 Tue Nov 6 21:36:18 CET 2001
982 - ported to cygwin (trivial).
983 patch by Gerrit P. Haase <gp@familiehaase.de>.
984 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
985
986 0.51 Thu Nov 1 20:39:01 CET 2001
987 - terminate/cancel now work properly (otherwise termination
988 could cause a "next coroutine is not and contains not..."
989 error.
990 - added Coro::Socket::shutdown.
991 - Coro::Event::loop no is the same as Event::loop.
992 - implemented terminate with args + join.
993
994 0.5 Fri Sep 28 16:15:35 CEST 2001
995 - fixed "print" on a Coro::Handle. The print method worked.
996 - small tweaks (seem to reduce memory consumption a lot)
997 in various modules.
998 - splendid use of "no warnings" scattered throughout.
999 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
1000
1001 0.49 Sun Sep 16 02:42:45 CEST 2001
1002 - changed some method calls to function calls for speed
1003 inside Coro::Handle.
1004 - make Coro::Handle use an array instead of a hash for
1005 speed reasons.
1006 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
1007 sjlj and ucontext should now work.
1008 - IRIX-specific port for libcoro.
1009 - swapped order of accept results to match IO::Socket.
1010 - changed getsock/peername to sock/peername to match IO::Socket.
1011 - fixed a bug that caused segfault when returning to main task
1012 under some circumstances.
1013 - other bugfixes.
1014
1015 0.45 Sun Sep 2 02:54:01 CEST 2001
1016 - new method Coro::Handle::timeout.
1017 - corrected speling of set...name to get...name in Coro::Socket.
1018 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
1019 - Coro::Handle now supports fileno and FILENO.
1020 - added eg/myhttpd, a web server.
1021 - Coro::Socket now accepts numeric ports as well as "port(number)"
1022 syntax.
1023 - moved some scheduling primitives into xs code. more to come.
1024 - new simple priority system.
1025 - implemented Coro::Event using XS, almost four times faster!
1026 - small memory corruption problem fixed (boy that was difficult).
1027
1028 0.13 Wed Aug 8 16:53:07 CEST 2001
1029 - new method Coro::Handle::readline.
1030 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
1031 (see eg/lwp on how to make LWP non-blocking).
1032 - renamed Coro::Socket::new_inet to new.
1033 - added Coro::Util, some utility functions.
1034 - die/eval now works better (still get annoying (but true ;)
1035 "Callback called exit" messages).
1036 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
1037 - much better Coro::Socket support.
1038
1039 0.12 Fri Jul 27 04:19:01 CEST 2001
1040 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
1041 - very experimental stack sharing algorithm. not 100% safe but
1042 should work well in practise ;)
1043 - added Coro::RWLock.
1044 - Coro::Specific now works.
1045
1046 0.11 Tue Jul 24 22:49:21 CEST 2001
1047 - added specialized hack for newer and older linux versions (fast).
1048 - renamed Coro::Event::IO to Coro::Handle.
1049 - new module Coro::Socket.
1050
1051 0.1 Tue Jul 24 01:47:53 CEST 2001
1052 - release candidate 3. A rather new internal structure :(
1053 - the great renaming: Damian Conway gave me a suitable replacement
1054 for yield.
1055 - added Coro/libcoro, a portable coroutine implementation for C
1056 (not even perl-dependent!!!), which can be used by Coro::State.
1057 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
1058 perl-5.6.
1059 - new function Coro::Event::idle.
1060 - the idle process is now overriden
1061 by default in Coro::Event.
1062 - Coro::Channel now enforces the size.
1063 - canceling events now works in all cases.
1064 - Coro state now include $_ and $@.
1065 - yet another bug workaround that I do not really understand :(
1066 - new module Coro::Event::IO (very undocumented) to do
1067 non-blocking i/o.
1068 - performance tuning.
1069
1070 0.08 Thu Jul 19 06:13:25 CEST 2001
1071 - release candidate 2 (new functionality).
1072 - add Coro::State::flush function.
1073 - transfer now supports a flags value (mostly for speed ATM).
1074 - might compile and work in the presence of threads now.
1075 - continuations have a saner syntax.
1076 - no more memleaks.
1077
1078 0.07 Tue Jul 17 17:40:18 CEST 2001
1079 - release candidate 1 ;)
1080 - slightly nicer code.
1081 - fixed a scheduling bug in Coro::Event.
1082
1083 0.06 Tue Jul 17 04:23:24 CEST 2001
1084 - ok, I found the showstopper - the same sub must not be
1085 re-entered in two different coroutines, otherwise => crash. I
1086 see no easy solution to this problem, except by walking the call
1087 chaing and saving/restoring the cv's, which is what I do now.
1088 - memory leaks still latent, especially at thread termination.
1089 - Coro::Event now works (haha).
1090
1091 0.05 Sun Jul 15 17:32:20 CEST 2001
1092 - fixed a few issues in Coro::Event.
1093 - I forgot to include Coro::Event in 0.04 :(:(:(
1094
1095 0.04 Sun Jul 15 05:24:59 CEST 2001
1096 - @_ is now properly localized.
1097 - Coro::State is now easier subclassable.
1098 - Coro::Cont now coroutine-aware.
1099 - Coro::Specific is a low-overhead module to create
1100 coroutine-specific vars.
1101 - Coro::Event provides a simple interface to Event.
1102
1103 0.03 Fri Jul 13 14:51:52 CEST 2001
1104 - transfer() now implemented in XS (beware).
1105 - new module Coro::Cont for really faked continuations.
1106 - big internal architecture changes: Coro::State is now
1107 really low-level and can thus be used to implement other
1108 interesting things, While "Coro::" implements a process-like
1109 model. Still crude and subject to change.
1110 - $_ and $@ are no longer being localized.
1111
1112 0.02 Tue Jul 10 01:38:17 CEST 2001
1113 - implemented "async" attribute.
1114 - $_ and $@ are now localized.
1115 - added Coro::Channel.
1116 - more testcases, still no docs.
1117
1118 0.01 Tue Jul 3 02:18:41 CEST 2001
1119 - original version; copied from Convert::Scalar.
1120