ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.385
Committed: Thu Oct 30 09:57:00 2008 UTC (15 years, 6 months ago) by root
Branch: MAIN
CVS Tags: rel-4_802
Changes since 1.384: +2 -1 lines
Log Message:
4.802

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48 outdated software).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