ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.433
Committed: Tue Nov 18 23:20:41 2008 UTC (15 years, 6 months ago) by root
Branch: MAIN
Changes since 1.432: +2 -1 lines
Log Message:
*** empty log message ***

File Contents

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