ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.382
Committed: Wed Oct 22 16:34:07 2008 UTC (15 years, 7 months ago) by root
Branch: MAIN
CVS Tags: rel-4_801
Changes since 1.381: +3 -1 lines
Log Message:
4.801

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44 outdated software).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