ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
(Generate patch)

Comparing Coro/Changes (file contents):
Revision 1.68 by root, Mon Feb 18 21:25:00 2002 UTC vs.
Revision 1.377 by root, Mon Sep 29 12:40:50 2008 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2
3TODO: should explore PerlIO::coroaio.
4TODO: 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
84.749 Mon Sep 29 14:40:12 CEST 2008
9 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
10
114.748 Sat Sep 27 14:03:19 CEST 2008
12 - implement, but do not document, PerlIO::cede(granularity).
13 - Coro::Storable forgot to wrap Storable::pstore.
14 - work around the multitude of leaks and memory corruption
15 bugs in PerlIO::via by using our own C-level perliol. As a side
16 effect, Coro::Storable is now much, much, much faster.
17
184.747 Tue Sep 23 01:59:41 CEST 2008
19 - fix a per-cv memleak (one empty array was leaked per
20 code reference).
21 - avoid a crash in coro->call|rss when the coroutine was already
22 destroyed (most noticably when using Coro::Debug::ps :)
23 - also protect *Storable::FILE.
24 - push up default storable granularity to 20ms.
25
264.746 Sun Sep 21 03:22:20 CEST 2008
27 - be more insistent on locking Storable against reentrancy
28 in Coro::Storable.
29 - move swap_def?v and throw to Coro::State, as documented.
30
314.745 Thu Jul 24 00:14:38 CEST 2008
32 - remove debugging code related to MgPV_nolen_const, also try to
33 make it compile with perl 5.8.6 (yes, apple apparently loves
34 outdated software). Reported by John S.
35
364.744 Tue Jul 8 22:06:35 CEST 2008
37 - correctly provide default DIE/WARN handlers as documented.
38 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
39 implementations inside perl work fine for us.
40
414.743 Mon Jun 16 00:21:57 CEST 2008
42 - when using Coro::EV without running EV::loop it could
43 result in busy-waiting for events, this has been fixed.
44 - reduce codesize and improve performance by using EV_DEFAULT_UC.
45
464.742 Sat May 31 14:10:21 CEST 2008
47 - implement a workaround for (some) perl <5.8.8 versions.
48 - require EV 3.3+.
49
504.741 Fri May 30 23:33:09 CEST 2008
51 - tell netbsd how utterly broken their imitation of an OS is
52 and refuse to build by default if pthreads are in use.
53 - switch to "s" method on all bsd's by default, as their ucontext
54 stuff seems just too broken.
55 - fix a bug in Coro::Select.
56
574.74 Thu May 29 20:05:31 CEST 2008
58 - do not test Coro::LWP for lack of dependencies.
59 - do not test Coro::Select for lack of working perls.
60
614.73 Thu May 29 2008
62 - fix a bug in Coro::EV which would cause it to block despite
63 there being runnable coroutines.
64 - sprinkle "no warnings" freely over everything, also suppress
65 warnings for some other modules.
66 - fix typo in WSAEWOULDBLOCK.
67
684.72 Sun May 25 05:14:36 CEST 2008
69 - tweak META.yaml a bit, unfortunately, there is no documented way
70 to have optional dependencies with CPAN. doh :(
71 - avoid running some tests on windows because they would fail due to
72 perl bug (broken fork, broken pipes...).
73 - work around perl on windows bugs where perl returns undocumented
74 error codes for sysread, syswrite etc. by taking advantage
75 of AnyEvent's workaround for that problem.
76 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
77 a common perl implementation bug on windows.
78 - use unix domain sockets in testsuite to work around
79 common perl implementation bugs on widows (they are emulated by
80 tcp sockets on windows. ugh.)
81
824.71 Sat May 24 20:01:27 CEST 2008
83 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
84 - take advantage of async name resolution of AnyEvent::Util.
85 - work around brutal inet_aton override in Coro::LWP.
86 - take advantage of the readyhook in Coro::EV, for smoother
87 scheduling.
88
894.7 Sun May 11 00:32:19 CEST 2008
90 - completely reworked the Coro manpage.
91 - added Coro::AnyEvent, generic event loop integration.
92 - implement cancel, ready and kill commands in Coro::Debug.
93 - document find_coro in Coro::Debug.
94 - incompatible change: rename has_stack to has_cctx.
95 - Coro::AIO and Coro::BDB no longer force event model detection,
96 use AnyEvent::AIO and AnyEvent::BDB.
97
984.6 Sat Apr 26 10:05:14 CEST 2008
99 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
100 by the Coro module. It was a mistake to have it in the first place.
101 - (experimental) support for activestate perl 5.10 (method "w").
102 - (experimental) support for strawberry perl (method "a").
103 - coro_sigelem_set did not return a value although it had to,
104 actual impact unknown.
105
1064.51 Mon Apr 14 13:28:27 CEST 2008
107 - make it compile again on 5.8.
108
1094.50 Thu Apr 10 09:43:17 CEST 2008
110 - I did it twice! (see 4.49).
111
1124.49 Sun Apr 6 21:23:31 CEST 2008
113 - grr, instead of compiling the recent changes on 5.10 only they
114 were compiled on 5.8 only.
115
1164.48 Sun Apr 6 20:36:46 CEST 2008
117 - allow coroutine switches during eval's under 5.10.x, as it seems
118 the parser is a per-interpreter option now, so this is safe
119 (this might fix the odd crashing bug).
120 - drop support for 5.9.x versions: they are dead, jim.
121
1224.47 Sun Apr 6 00:37:52 CEST 2008
123 - force cctx allocation on API calls: we know we need to force one
124 and gcc actually manages to confuse our heuristic nowadays,
125 leading to crashes and worse.
126 - document force_cctx.
127
1284.46 Fri Apr 4 22:05:43 CEST 2008
129 - upgrade libcoro, resulting in pthread-backend (which was only created
130 to fulfill the rules of the programming languages shootout).
131
1324.45 Thu Mar 13 11:55:36 CET 2008
133 - fix a file leaking bug in eg/mhyttpd that would allow
134 downloading of any file (reported by oesi).
135 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
136 (also add testcase).
137 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
138 apparently having loads of fun with that).
139 - implement $state->swap_defsv, ->swap_defav and document ->throw.
140
1414.4 Wed Feb 13 16:44:29 CET 2008
142 - only assume .cfi pseudo isns exist on GNU/Linux.
143 - add get_timed method to Coro::Channels.
144 - fixed Coro::Debug manpage.
145 - perl 5.11 compatbility improvement based on patch by
146 Andreas König.
147
1484.37 Sun Jan 20 11:25:23 CET 2008
149 - soften the check for an active parser for perl 5.10 (avoiding
150 "Coro::State::transfer called while parsing" in many cases).
151
1524.36 Sun Jan 13 10:53:56 CET 2008
153 - reset diehook when terminating from an async_pool as to not
154 trigger any __DIE__ handlers.
155
1564.35 Sun Jan 13 04:14:13 CET 2008
157 - "bt" debug command now displays any exceptions
158 from longmess and also skips the innermost
159 stackframes, giving more useufl output.
160 - allow backtraces in coroutines blocked in Coro::EV,
161 at a <1% speed hit.
162 - handle localising of $SIG{__WARN__} and __DIE__
163 properly (with a proper amount of dirty hacking).
164
1654.34 Sat Dec 22 17:49:53 CET 2007
166 - upgrade to EV version 2.0 API.
167
1684.33 Mon Dec 17 08:36:12 CET 2007
169 - make Coro::AIO etc. loadable in the absence of EV.
170
1714.32 Mon Dec 17 07:46:02 CET 2007
172 - majorly improved Coro::Handle's performance with Coro::EV.
173 - implemented the until now mythical Coro::BDB module.
174 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
175
1764.31 Wed Dec 5 12:32:39 CET 2007
177 - remove warn statement form Coro::Util.
178
1794.3 Tue Dec 4 20:33:14 CET 2007
180 - calls to the idle function could cause stack corruption
181 when the stack changed.
182 - do no longer rely on the presence of EV::DNS if EV is used
183 (because it is gone), but instead take avdantage of EV::ADNS
184 if available.
185 - add ($) prototypes to all functions in Coro::Storable.
186 - use a conventional (safer) idle callback in Coro::EV.
187 - do accept quickly in Coro::Debug to avoid endless loops.
188
1894.22 Fri Nov 30 16:04:04 CET 2007
190 - really use optimised versions for Event and EV in Coro::Util
191 and Coro::Handle.
192
1934.21 Sun Nov 25 10:48:59 CET 2007
194 - fix a spurious memory read.
195 - Coro::EV no longer keeps the eventloop "alive".
196
1974.2 Fri Nov 9 20:47:05 CET 2007
198 - enable/disable tracing from a new coroutine, not a pooled one.
199 - fix a memleak in Coro::Event.
200 - removed killall call from fork_eval.
201 - made sure store_fd is already loaded so that fork_eval
202 does not have to parse autoload in each subprocess.
203 - only use assembly method if -O switch is in $Config{optimize}.
204 - add (optional) Coro::EV module, so far the best event loop module
205 directly supported by Coro.
206 - if the event model is EV, use EV::DNS to resolve
207 stuff in Coro::Util.
208 - don't get confused by multiple event notifications in Coro::Handle.
209 - initial support for EV (libevent interface).
210 - require Event and EV using configure_requires, to force their existance.
211
2124.13 Wed Oct 24 07:26:45 CEST 2007
213 - add Coro::Storable::blocking_thaw.
214 - use a vastly more complicated technique to localise
215 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
216 - use a coroutine for the idle callback Coro::Event,
217 instead of running Event in the current coroutine context.
218 This also catches recursive invocations.
219 - actually report fork errors in gethostbyname and inet_aton.
220
2214.11 Thu Oct 11 02:40:24 CEST 2007
222 - port to threaded perls.
223
2244.1 Thu Oct 11 02:38:16 CEST 2007
225 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
226 be local to each coro (see Coro::State).
227 - incompatible change: for very deep reasons, cede and cede_notself
228 cannot return anything, so nothing will be returned.
229 - possibly bring back 5.10 compatibility (untested).
230 - work around stupid (and wrong) warning on 5.10 :(.
231 - overlay the saved state over the context stack. This saves
232 a few hundred bytes per coroutine on average and also
233 speeds up context switching a bit.
234 - further tune default stack sizes.
235 - (more) correctly calculate stack usage in coro_rss.
236 - Coro::Storable::blocking_* did not properly lock
237 resulting in races between coroutines.
238 - added Coro::Storable::guard.
239 - stopping to trace a coroutine could destroy the cctx of
240 an unrelated coroutine.
241 - explain the relationship between Perl and C coroutines in
242 more detail in Coro::State.
243 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
244 causing a fork per resolve. This also affected Coro::Socket.
245 - switch to a separate stack in $coro->call/eval to avoid
246 invalidating pointers.
247
2484.03 Sat Oct 6 21:24:00 CEST 2007
249 - added Coro::throw method.
250 - minor code cleanups.
251
2524.02 Sat Oct 6 02:36:47 CEST 2007
253 - fix a very minor per-coroutine memleak (a single codereference).
254 - fixed a bug where the currently in-use c context would be freed
255 prematurely (can happen only when programs change the stacksize
256 or use tracing).
257 - tracing can no longer keep a coro alive after it terminated.
258 - do static branch prediction in the common path for gcc. gives
259 about 2-5% speed improvement here.
260
2614.01 Fri Oct 5 22:10:49 CEST 2007
262 - instead of recreating *a* standard output handle we simply
263 use STDOUT, which is faster and hopefully more robust.
264
2654.0 Fri Oct 5 12:56:00 CEST 2007
266 - incompatibly changed Coro::Storable::freeze.
267 - major new feature: added Coro::Debug, for interactive coroutine
268 debugging, tracing and much more.
269 - major bug fix: unbelievable, but true: $_, $/ and many other
270 "saved" variables actually weren't being saved. This has been fixed,
271 of course, while increasing performance while losing all the save
272 flags.
273 - save flags are gone, and all the api functions dealing with them.
274 - added Coro::Semaphore::adjust.
275 - added Coro::Util::fork_eval.
276 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
277 - added Coro::killall.
278 - reduce initial stack sizes to allow for "micro-coroutines".
279 - better async_pool resource management, moved parts of async_pool
280 handling to XS (major speed improvement).
281 - actually croak before modifying important data structures.
282 - refuse to transfer while compiling.
283 - possibly support eval EXPR better now.
284 - enable assembly per default on linux+bsd x86+amd64.
285 - all internal members were renamed _something for easier subclassing.
286 - many minor tweaks.
287
2883.63 Wed May 16 14:10:06 CEST 2007
289 - implement handcoded assembly for x86/amd64 SVR ABI.
290
2913.62 Fri Apr 27 21:36:06 CEST 2007
292 - upgrade libcoro (which might set unwind info correctly).
293 - change default on linux to setjmp/longjmp.
294
2953.61 Thu Apr 19 12:36:18 CEST 2007
296 - Coro::Storable caused an endless loop when thawing invalid
297 pst's sometimes.
298 - use a Semaphore in Coro::Storable, as Storable doesn't
299 seem to be reentrant (although it is documented to
300 be threadsafe...).
301 - fix Coro::Signal to bring back the original unreliable
302 but stateful semantics.
303 - fixed a lot of typos in Coro.pm (patch submitted by David
304 Steinbrunner, which applied flawlessly).
305
3063.6 Sat Apr 14 17:13:31 CEST 2007
307 - added some bugfixes to get eg/myhttpd working again.
308 - added Coro::Storable for often-cede'ing freeze/thaw.
309 - try to do a clean exit when a coroutine calls exit
310 (EXPERIMENTAL).
311 - got rid of indirect call through _coro_init.
312 - updated the partly antique examples in eg/ to
313 work again and be a bit less magic, too.
314 - fixed Coro::Signal semantics to work as documented again.
315
3163.55 Sun Mar 25 01:20:47 CET 2007
317 - add SAVE_DEFFH to save the default filehandle and enable
318 it by default.
319 - finally move socket-operations from Coro::Socket to Coro::Handle
320 to be able to unblock foreign sockets.
321 - add Coro::State::save_also and guarded_save.
322 - add count accessor to Coro::Semaphore.
323 - add Coro::State::cctx_stacksize.
324 - just for the fun of it, do not rely on implicit context,
325 which can dramatically improve performance, but people
326 using windows-process-emulation perls do not care much about
327 performance.
328
3293.51 Sun Mar 4 14:18:04 CET 2007
330 - fixed a problem when you weakened references to Coro::State's
331 (patch by Yuval Kogman).
332
3333.501 Wed Feb 28 12:44:07 CET 2007
334 - rename some global symbols as macosx from hell redefines
335 them without asking.
336
3373.5 Tue Feb 13 20:22:53 CET 2007
338 - do AnyEvent model detection earlier, avoiding problems
339 caused by first using AnyEvent and later Coro::Event.
340 - implement and document Coro::Event event objects.
341 - fix a potential problem in Coro::Event causing crashes.
342 - initialise PL_comppad when creating a new coroutine,
343 avoids crashes on early coro destruction.
344
3453.41 Mon Jan 22 19:19:49 CET 2007
346 - readline on Coro::Handle did not support undefined $/,
347 nor did it deliver partial lines on EOF or error.
348 - implement malloc fallback for stack allocation because
349 stupid broken idiotic OSX has a stupid broken
350 idiotic fits-the-whole-os mmap "implementation" and
351 my dick feels longer if Coro is portable even to
352 obsolete platforms.
353
3543.4 Fri Jan 19 21:52:54 CET 2007
355 - remove t/09_timer.t, as it isn't really testing much
356 but was rather flaky in practise.
357 - async_pool coro would keep arguments and callback alive until
358 it was reused.
359 - cancellation of a coroutine could cause spurious idle calls
360 in cede_notself.
361
3623.3 Sat Jan 6 03:45:00 CET 2007
363 - implement $coro->on_destroy.
364 - Coro::Event blocking semantics have been changed,
365 documented and - hopefully - improved.
366 - fix nice adding, not subtracting, from priority.
367 - fix ->prio and api_is_ready (patch by Mark Hinds).
368 - fixed an assert ("... == prev__cctx->idle_te")
369 that could errronously trigger.
370 - fix various large and small memleaks.
371 - use a (hopefully) more stable cancel implementation
372 that immediately frees the coroutine data.
373 - cede/cede_notself return a status now.
374 - added Coro::guard function.
375 - added a global coroutine pool for jobs (on my machine,
376 I can create and execute 48k simple coros/s with async,
377 and 128k coros with async_pool).
378 - Coro::AIO now uses the coroutine priority as io priority.
379
3803.2 Fri Dec 22 05:07:09 CET 2006
381 - improve portability to slightly older perls.
382 - use cleaner coroutine destruction.
383 - simplify configuration for users.
384 - optionally (unrecommended) prefer perl functions over
385 their coro replacements.
386
3873.11 Tue Dec 5 13:11:24 CET 2006
388 - fixed some bogus assert's, but as perl.h disables assert even
389 without NDEBUG (thank you very much), not too many people should
390 notice (that did include myself). Andreas König noticed, though :)
391 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
392 it. Also noticed by Andreas König :)
393 - save/restore tainted status.
394 - verified to pass the testsuite on my 5.9.5.
395
3963.1 Mon Dec 4 23:03:40 CET 2006
397 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
398 - incompatible change: transfer flags are now per-state.
399 - give a better error message on deadlock.
400 - document Coro::nready.
401 - enhanced testsuite.
402
4033.01 Sun Dec 3 23:47:42 CET 2006
404 - forgot to include Coro::Timer.
405
4063.0 Sun Dec 3 22:57:25 CET 2006
407 - the "FINALLY COMPLETELY STABLE" release (coming soon:
408 the "FAMOUS LAST WORDS" release).
409 - implement a new stack sharing algorithm, which uses a stack
410 pool (size currently hardcoded).
411 - make stack sharing mandatory (it no longer uses a heuristic).
412 - eval/die no longer cause weird problems under heavy use.
413 - Coro::Event could cause livelocks if it was used but
414 no Coro::Event watchers were used.
415 - Coro::Event now uses asynccheck as prepare does not
416 check for changed watchers.
417 - Coro::Event allows multiple waiting coros and will wake up one
418 per event.
419 - Coro::Event should be cleaner and more efficient now.
420 - new utility function Coro::unblock_sub.
421 - document the sad fact that Event is no longer reentrant.
422 - putting a coroutine into the ready queue twice could under
423 some circumstances lead to stack corruption.
424 - minor incompatibility: subclassing Coro::State is supported
425 directly now without going through a _coro_state member.
426 - state/coro switching is much faster now.
427 - very minor optimisations and code/documentation cleanup.
428 - avoid problems due to compiler inlining.
429 - removed timers from Coro::Timer -> use AnyEvent instead.
430 - replaced Coro::idle coroutine by (cleaner) idle handler.
431 - updated to newest libcoro.
432 - implement enhanced support for valgrind.
433 - implement is_ready and return value for ready.
434 - removed Coro::Cont, it was a misnomer (it's generators),
435 and it was somewhat annoying to get it right. Will come back
436 if somebody asks for it :)
437 - many ->wait methods and Coro::Event could return
438 spuriously without the event having happened.
439
4402.5 Tue Nov 7 12:22:33 CET 2006
441 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
442 use AnyEvent, moved them to Coro/.
443 - added Coro::LWP which contains all the uglyness required to
444 make LWP non-blocking.
445 - should work with perl 5.9.x now (Andreas König made me do it).
446 - fixed another bug in Coro::Select when the timeout was undef.
447 - reuse PL_start_env for all coros, saves some memory per coroutine.
448 - manage PL_top_env differently, hopefully to avoid panic: top_env.
449 - timeout argument was not properly used in Coro::Socket.
450 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
451 - emulate undocumented(!) functionality of IO::Socket required
452 by LWP(!!).
453 - updated eg/lwp to work with newer lwp's.
454 - remove "FATAL: uncaught exception" prefix. Coroutines that die
455 kill the whole process, just as exceptions in the main "coroutine"
456 did already.
457
4582.1 Wed Nov 1 23:01:13 CET 2006
459 - fix a long-standing bug in Coro::Select where select with
460 zero timeout would instead change the current default filehandle.
461 - use a simpler and hopefully more robust way to clone padlists
462 (uses less memory and a perl function instead of our own).
463 - coro can now create a stack guard area on many architectures.
464 - Coro::AIO properly reexports additional functions from IO::AIO.
465 - updated libcoro with a workaround for OS X,
466 pach and testing by Michael Schwern.
467
4682.0 Tue Oct 24 05:47:17 CEST 2006
469 - support additional aio requests in Coro::AIO.
470
4711.9 (never properly released due to a glitch)
472
4731.8 Thu Feb 2 00:59:06 CET 2006
474 - applied suggested patch by SAMV to avoid problems during stupid
475 mark & sweep gc run.
476 - applied patch by Scott Walters for 5.9.3 compatibility.
477
4781.7 Tue Dec 27 01:41:58 CET 2005
479 - added Coro::AIO, a thin wrapper around IO::AIO.
480 - improved Makefile.PL explanations.
481
4821.6 Mon Dec 12 21:30:05 CET 2005
483 - additionally save PM_curpm might fix as-of-yet
484 unseen problems with regex matches being attributed to the
485 wrong package.
486 - add t/10_bugs.t, which currently checks against imho broken
487 perls that use 0.26MB of stack space per Perl_magic_get
488 invocation (newer linuxes) as opposed to the 0.0002MB perl
489 normally uses.
490 - make stacksize configurable for the ultimate debian experience.
491
4921.51 Mon Dec 12 18:48:36 CET 2005
493 - remove debugging warn() accidentally left in Coro::Select.
494
4951.5 Tue Nov 29 13:32:44 CET 2005
496 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
497 - fix Coro::Select 'select'.
498 - strict'ify some modules.
499 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
500 - use XSLoader in selected modules.
501 - remove some 5.6 compatibility cruft.
502
5031.4 Tue Sep 6 00:11:05 CEST 2005
504 - libcoro did not take into account the trampoline on amd64,
505 when 'l'inux method was used.
506
5071.31 Tue Aug 30 23:31:33 CEST 2005
508 - some portability fixes/workarounds.
509
5101.3 Sat Aug 20 03:08:56 CEST 2005
511 - no code changes. module seems to work fine.
512 - improve Coro::State docs, remove reference to nonexisting
513 Coro::State::flush method.
514 - no longer autodetect windows, present it as an option instead.
515
5161.2 Mon May 16 02:00:55 CEST 2005
517 - included libcoro.c earlier in State.xs, might improve portability.
518 - use a faster and possibly more stable Coro::Cont implementation.
519 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
520 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
521 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
522
5231.11 Thu Mar 3 18:00:52 CET 2005
524 - change of contact address.
525
5261.1 Tue Feb 22 20:51:16 CET 2005
527 - support [l]inux method on amd64.
528 - allow some context switches while compiling/eval'ing.
529
5301.0 Tue Aug 10 03:47:27 CEST 2004
531 - Event 0.89 fixes the bug again.
532 - fix the remaining(??) memleak. I didn't free the cache-AV for
533 padlists, so one AV per created anon-sub leaked.
534
5350.97 Fri May 14 15:23:32 CEST 2004
536 - remove stupid gcc extension that was an experiment only anyway.
537
5380.96 Thu May 13 18:09:29 CEST 2004
539 - tell people if they have yet another broken Event variant.
540 - made Coro to work _better_ with threads (it seems that, as long
541 as you use Coros in one thread only everything will work fine.
542 Looking at the perl source the problem might be missing locking
543 between Coro and some perl internal routines. That's not easily
544 fixable).
545 - (hopefully) got rid of the static xnext inside transfer.
546
5470.95 Sun Feb 15 03:20:28 CET 2004
548 - removed Coro::State::flush, implemented a new and much
549 faster caching mechanism by attaching magic to the CV.
550 - a memleak with (real) closures remains.
551 - dramatically reduced size of stacks to 65536 (linux/x86), after
552 observing that even myhttpd never had more than a kilobyte
553 of stack in use (stack is still only physically allocated in
554 pagesize increments for systems suporting mmap).
555 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
556 - fix for threaded perl.
557
5580.9 Sun Nov 23 04:32:39 CET 2003
559 - fix a few very hard to track down but (of course) stupid
560 memory leaks.
561 - stringify version number to avoid locale problems :(.
562 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
563 hopefully ignoring it "just works".
564
5650.8 Wed Nov 5 19:38:40 CET 2003
566 - port to perl5.005_03, but only 5.8.x+ is supported!
567 - honor LocalAddr even without LocalPort in Coro::Socket.
568
5690.7 Tue May 27 03:12:38 CEST 2003
570 - the version jump indicates some level of testing, not gobs
571 of new features.
572 - uh... I found the prompt function in ExtUtils::MakeMaker.
573 Highly correct stuff, that is...
574 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
575 the fix is not well understood by the author, so beware :(.
576
5770.652 Thu May 8 02:54:46 CEST 2003
578 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
579 cause version 4 doesn't have ucontext.
580 - Benjamin Reed reported that setjmp works fine on darwin,
581 so preselect it.
582
5830.651 Sat Mar 29 15:00:23 CET 2003
584 - fix a bug in Coro::Handle where some bytes could get lost
585 on reading, reported by jason@nichego.net.
586
5870.65 Sun Mar 23 00:08:26 CET 2003
588 - added a README.linux-glibc.
589 - new module Coro::Select.
590 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
591 (I never claimed that I know what I am doing ;).
592 - set default to setjmp/longjmp on non-x86-linux.
593
5940.6 Thu Nov 21 11:09:06 CET 2002
595 - made lazy context switching the default.
596 - fixed the bug where SemaphoreSet::guard
597 was the same as timed_guard and timed_guard was missing.
598 - fix a memleak in Coro::Timer.
599 - reclassified context sharing as not experimental.
600
6010.534 Sun Apr 14 03:05:12 CEST 2002
602 - fixed backspace => tab in header parsing (myhttpd).
603 - added eg/readline, for elmex.
604 - Coro::Event now calls ->start on first call instead of
605 ->again, which created wrong timeouts for the first ->next.
606 - fixed a bug where Coro::Socket returned a closed socket instead
607 of nothing on ECONNREFUSED and similar error conditions.
2 608
30.533 Mon Feb 18 18:49:40 CET 2002 6090.533 Mon Feb 18 18:49:40 CET 2002
4 - adapt to finally working Event-0.85. 610 - adapt to finally working Event-0.85.
5 - fixed buggy event initialization (trapped under windows, but how 611 - fixed buggy event initialization (trapped under windows, but how
6 could it possibly have worked anywhere?) 612 could it possibly have worked anywhere?)

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines