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

Comparing Coro/Changes (file contents):
Revision 1.396 by root, Sat Nov 8 13:42:28 2008 UTC vs.
Revision 1.631 by root, Sun Jun 1 19:55:41 2014 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::coroaio (perl leaks like hell). 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: maybe implement a default message channel, very much like Erlang's 4TODO: channel->maxsize(newsize)?
5 actor model (which is cool in a lot of important aspects (failures!), 5TODO: __GCC_HAVE_DWARF2_CFI_ASM
6 but very lacking in others (higher level ipc)). 6TODO: swap_sv, maybe add scope_swap_sv, andallow to unswap by repeat?
7TODO: swap_sv not undone in pool
8TODO: croak when async_pool triesd to runc anceled thread?
9TODO: look into the array placeholder now NULL issue.
7 10
85.0 116.38 Sun Jun 1 21:54:23 CEST 2014
12 - check that perl slots actually have enough space to hold
13 interpreter variables.
14 - untested port to perl 5.19 (64 bit tmps indices) (reported
15 by Andreas König).
16 - croak when cancel is called without a thread context.
17
186.37 Tue Mar 4 13:27:33 CET 2014
19 - *sigh*, removed leftover debugging code from debugging a
20 perl bug, of all things.
21
226.36 Tue Mar 4 07:11:59 CET 2014
23 - semaphores would not clear the destroy hook when interrupted
24 by ->throw, causing segfaults or worse.
25 - ->throw on a thread waiting for a semaphore did not acquire
26 the semaphore, but also didn't wake up other waiters,
27 possibly causing a deadlock.
28 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
29 to report a full stacktrace, which should help find out
30 where the actual call is.
31 - "atomically" destroy data in slf_destroy, because it is
32 the right thing to do, just in case.
33 - disable ecb.h libm dependency, to compile on ancient systems
34 or under adverse conditions.
35
366.33 Mon Nov 18 11:26:27 CET 2013
37 - do not crash when freeing padlists with holes (in 5.18).
38 - tentative SVt_BIND 5.19 port/fix.
39
406.32 Tue Nov 5 15:35:35 CET 2013
41 - use a new algorithm to derive padlists for perl 5.18. The old
42 one could lead to 0-pointer accesses inside perl (reported
43 by Darin McBride).
44
456.31 Thu May 9 07:39:48 CEST 2013
46 - Coro::AIO requests would crash if the thread was ready'd
47 while the request was ongoing.
48
496.29 Wed May 8 02:55:18 CEST 2013
50 - when an on_destroy handler destructs the coro currently being
51 destructed a perl scalar could be accessed after being freed,
52 likely causing a crash.
53
546.28 Wed Mar 6 06:58:02 CET 2013
55 - clean remnants of existing __DIE__ and __WARN__ handlers so
56 they lose their magic and will not cause segfaults later
57 (testcase by Andrey Sagulin).
58 - improved Coro::State documentation a bit.
59 - Coro::Debug::command now flushes the output.
60 - add hack detection code for x32 abi, because the braindead slugs
61 who designed that made it look exactly like x86_64 without
62 providing proper compile time symbols to test for it. as a result,
63 this detection cannot work reliably.
64 - valgrind stack registering was broken.
65 - do not rely on Time::HiRes anymore in Coro::Debug.
66
676.23 Fri Dec 7 23:36:37 CET 2012
68 - use experimental fiber implementation on native windows
69 perls.
70 - use sizeof (void *) as multiplication factor for stack sizes,
71 to accomodate the totally braindamaged microsoft 64 bit "os".
72 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
73 activeperl 5.16 crashes when PerlIO_define_layer is called due
74 to some bug in the perl dll, strawberry perl at least passes
75 the testsuite.
76 - implement Coro::Handle->peeraddr/host/port, for slightly
77 improved compatibility with LWP.
78 - implement 5.17 compatibility by almost blindly applying a
79 good-looking patch by Father Chrysostomos.
80 - move stack management functions into libcoro 3.
81 - libcoro version 3 "released".
82 - support magic values as timed_io_once args.
83 - recommend AnyEvent 7+ or EV 4+, also require EV
84 version 4 or newer for Coro::EV.
85
866.10 Tue Oct 9 01:14:27 CEST 2012
87 - updated ecb.h, it had a typo that caused it to not compile on many
88 big endian systems (reported by many people).
89 - disable memory fences in ecb.h to improve portability.
90
916.09 Sat Oct 6 23:25:02 CEST 2012
92 - Coro::EV I/O watchers were not interruptible by exceptions
93 (Coro::State::throw) (testcase by sten).
94 - ->throw now puts threads into the ready queue, as this seems to
95 be expected by existing code, and code that doesn't cope with spurious
96 wakeups needs fixing anyway.
97 - use fd -1 in mmap.
98 - cast I32 to int in error message printf.
99 - warn about broken so-called "hardened" kernels.
100
1016.08 Fri Apr 13 12:05:47 CEST 2012
102 - be more aggressive about exiting like perl does - formerly,
103 exiting from the non-main thread would not execute END blocks.
104
1056.07 Fri Nov 11 21:21:48 CET 2011
106 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
107 - update ecb.h.
108
1096.06 Mon Aug 8 23:59:48 CEST 2011
110 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
111 backend, which is an order of magnitude slower. unfortunately, cygwins
112 pthread implementation isn't very complete either, so allocate the stack
113 twice just to be sure.
114 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
115 in its ithreaded perl - best recompile cygwin and use the 'w'indows
116 backend for much better performance. also disable ithreads for
117 even better performance...).
118
1196.05 Thu Aug 4 21:36:36 CEST 2011
120 - blush, condvar values would not be propagated from send to recv anymore
121 (reported by Chip Salzenberg).
122 - use exponential increase for the readline buffer length in
123 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
124
1256.04 Wed Aug 3 17:15:45 CEST 2011
126 - use even more efficient and more compatible condvars for
127 compatibility to AnyEvent 6.x :)
128 - more inconsequential ecb.h updates.
129
1306.03 Wed Aug 3 11:41:30 CEST 2011
131 - change how Coro patches AnyEvent condvars for compatibility to
132 AnyEvent 6.x.
133 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
134 hasn't been included.
135
1366.02 Wed Jul 13 04:35:19 CEST 2011
137 - "improve portability to Gentoo" - gentoo manages to put perl variables
138 in memory areas that are farther than 2gb apart, which the jit couldn't
139 handle and barfed. now it's just a bit slower on gentoo and similar
140 systems.
141
1426.01 Sun Jul 3 12:31:14 CEST 2011
143 - workarounds are good, but the test for whether pthreads are used
144 was not good. this one should be better.
145 - check differently whether gcc generates cfi instructions itself.
146
1476.0 Wed Jun 29 19:43:35 CEST 2011
148 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
149 destroyed and cleaned up automatically (e.g. async { schedule }).
150 - implement a JIT compiler for part of the thread switch code,
151 which gives a 50% speed improvement on threaded perls, and
152 about 4% on non-threaded perls (so threaded perls now finally
153 reach about half the speed of non-threaded perls).
154 - slightly modernise Coro::Intro, add section about rouse functions.
155 - avoid DEFSV and ERRSV, giving another 10% improvement
156 in thread switching.
157 - Coro::State->is_destroyed is now called is_zombie.
158 - implement a Coro->safe_cancel method that might fail, but
159 cancels in a "safer" way if it succeeds.
160 - add preliminary support for DEBUGGING perls.
161 - get rid of two hash-accesses when initialising a new Coro - this
162 speeds up coro creation by almost a factor of two.
163 - croak when a coro that is being cancelled tries to block
164 (e.g. while executing a guard block), instead of crashing or
165 deadlocking.
166 - use a more robust and also faster method to identify Coro::State
167 objects - speeds up everything a bit.
168 - implement Coro->cancel in XS for a 20% speed improvement, and to
169 be able to implement mutual cancellation.
170 - speed up context switches by a percent or two by more efficiently
171 allocating context stack entries.
172 - implement Coro->join and Coro->on_destroy in XS for a speedup and
173 a reduction in memory use.
174 - cancelling a coro while it itself is cancelling another coro is
175 now supported and working, instead of triggering an assertion.
176 - be a bit more crash-resistant when calling (buggy) on_destroy
177 callbacks (best effort).
178 - move on_destroy into the slf_frame, to allow extension slf
179 functions to have destructors.
180 - get rid if coro refcounting - simply crash in other interpreter
181 threads by nulling the pointers on clone.
182 - simplify warn/die hook handling when loading Coro - the convoluted
183 logic seems to be no longer neccessary.
184 - use libecb instead of our own home-grown gcc hacks.
185 - document alternatives to Coro::LWP. Please use them :)
186 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
187 sigaltstack. Really. wine suffers from it, erlang suffers from it,
188 and it's known since at least 2006.
189
1905.372 Wed Feb 23 06:14:30 CET 2011
191 - apparently mingw doesn't provide a working gettimeofday, try to
192 work around that by relying on Time::HiRes (indirectly brought to
193 my attention by Max Maischein).
194 - fix some portability issues when Time::HiRes was used.
195
1965.371 Mon Feb 21 14:36:08 CET 2011
197 - backport to windows process emulation code again.
198
1995.37 Sat Feb 19 07:49:44 CET 2011
200 - add a big "Coro thread life cycle" section to "man Coro".
201 - try a tentative workaround against the breakage that 5.13 has
202 introduced without depreciation period. sigh.
203 - no longer use Time::HiRes if gettimeofday is available, which
204 saves quite a lot of memory.
205
2065.36 Sun Feb 13 05:33:41 CET 2011
207 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
208 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
209 method call.
210 - undocument Coro::Timer::sleep and obsolete whole module.
211 - optimise Coro::Timer::timeout memory and cpu usage.
212 - slightly updated Coro::Intro for recent changes.
213 - do not initialise PL_dirty anymore.
214
2155.25 Thu Nov 11 01:08:39 CET 2010
216 - try a different approach on netbsd - netbsd 5 finally has marginally
217 working pthreads, but still broken ucontext/sigaltstack.
218 - openbsd 4.8 finally got their act together, Coro works out of the box
219 with asm, setjmp and pthreads (no change, just informational).
220
2215.24 Sat Oct 23 11:27:12 CEST 2010
222 - port to the EV 4.0 API.
223 - work around bugs in mingw32, making strawberry perl work
224 out of the box.
225 - correctly modify Coro::AIO function prototypes
226 so that they reflect the "no optional parameters" rule.
227 - "ported" libcoro to C++.
228
2295.23 Mon May 17 18:50:42 CEST 2010
230 - be more resistant to ordering changes when initialising
231 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
232 Waldorf).
233 - document that perl 5.12 deliberately removed support for cloning.
234
2355.22 Wed Apr 14 03:55:35 CEST 2010
236 - correctly return udnef on errors in Coro::Handle::read/write
237 (testcase by Marc Mims).
238 - convert Coro::Util into a "perl compatibility wrapper" - the functions
239 are less useful now, but are drop-in replacements for existing
240 functions, listing better alternatives in the documentation. This also
241 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
242 with Coro::Util::inet_aton.
243 - do not override $Coro::idle unconditionally in Coro.pm, as other
244 modules could have provided their own idle coro already
245 (for exmaple, Coro::AnyEvent).
246 - fix Coro::Util::gethost* functions.
247 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
248
2495.21 Wed Dec 16 07:19:51 CET 2009
250 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
251 together.
252 - add some examples on how to combine other event loops with Coro in
253 Coro::AnyEvent, and how to run it (and not to block). Seems to be
254 the most common source of confusion.
255 - try to catch people naively blocking in an event callback.
256 - work around the perl filehandle bug issue in conjunction with
257 older common::sense (as indirectly pointed out by ZSystem).
258 - clarify the "not from signal handlers" section.
259
2605.2 Sun Oct 4 14:54:24 CEST 2009
261 - Coro::Storable destroyed the prototypes of the functions it wrapped.
262 - export rouse_cb and rouse_wait by default now.
263 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
264 - new method $state->swap_sv.
265 - added section on "windows process emulation" to the manpage,
266 after a not-so-fruitful (nor-friendly) "discussion" with chip
267 salzenberg (discussion implies arguments, but his only arguments
268 were ad-hominems, one wonders why he started it in the first
269 place). I hope this explains it well enough for him to understand,
270 and maybe well enough for others to understand.
271 - use common::sense everywhere now.
272 - idle callbacks are no longer supported, use idle coros instead.
273 - print a thread listing when a deadlock is detected.
274
2755.17 Sat Aug 22 23:09:31 CEST 2009
276 - work around a bug in the perl debugger causing crashes
277 when running under the debugger by marking _pool_handler as nodebug.
278 - speed up Coro::async considerably.
279 - try some hacks to get netbsd to work "more often" - their broken
280 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
281 - convert Coro to AE 5.0 API.
282
2835.162 Tue Jul 28 04:04:03 CEST 2009
284 - perl 5.8.2 is now minimum requirement.
285 - skip t/19_handle.t on broken windows perls.
286
2875.161 Wed Jul 22 04:47:38 CEST 2009
288 - Coro::AnyEvent::poll could have a different prototype when EV was
289 used as backend (analyzed by Tatsuhiko Miyagawa).
290 - Coro::AnyEvent errornously initialised the event loop when loaded,
291 not on demand.
292 - try to workaround rare */t/01_unblock.t failures.
293
2945.16 Tue Jul 21 01:44:37 CEST 2009
295 - Coro::AnyEvent failed to hook into the event loop
296 when no threads had been readied between detecting
297 the event loop and actually running it.
298 - considerably speed up Coro::Select by taking avdantage
299 of AnyEvent > 4.8 and some other optimisations.
300 - implement paragraph readline mode in Coro::Handle
301 (based on patches by Zsbán Ambrus).
302 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
303 and analyzed by Yasuhiro MATSUMOTO).
304 - clarified libcoro license and copyright.
305 - someone stole my EXTRA_META!!!
306 - implement Coro::Select::patch_pp_sselect and it's brother,
307 for hardcode select overriding.
308
3095.151 Mon Jul 6 05:41:57 CEST 2009
310 - backport to windows process emulation code again (patch by
311 Yasuhiro MATSUMOTO).
312 - slightly update Coro::MakeMaker.
313
3145.15 Tue Jun 30 10:28:06 CEST 2009
315 - deprecate Coro::Socket, document how to get ipv6 support via
316 AnyEvent::Socket instead.
317 - implement signal->wait ($cb) interface, similar to semaphores.
318 - work around SvOK not supporting getmagic, so we have to getmagic
319 to test for undef :/ (reported by Matthias Waldorf).
320 - load Coro::AnyEvent in all modules using AnyEvent.
321 - work around perl corrupting our internal data structures,
322 reported by Tokuhiro Matsuno.
323 - enable per-coroutine real and cpu time gathering
324 (Coro::State::enable_times).
325
3265.14 Wed Jun 24 01:37:48 CEST 2009
327 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
328 as this seems to have been a difficult concept (poll, sleep, idle,
329 idle_upto).
330 - add Coro::AnyEvent::readable/writable functions.
331 - clarify Coro::EV/Event/AnyEvent manpages.
332 - free per-thread global scalars in the thread calling ->cancel, to
333 avoid crashes when $_, $@ etc., are magical but some of those
334 had already been freed.
335 - "unexperimentalise" the callback interface for Coro::Semaphore.
336 - speed up ready queue management/context switching by using a linked
337 list instead of an array (~5-10%).
338 - implement "watch" command in Coro::Debug shells.
339 - for fun, implement time-slicing as an example in the manpage.
340 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
341 Coro::Event, use the normal AnyEvent handling instead of dieing
342 (the same is true for Coro::Handle).
343 - properly document Coro::EV::timed_io_once.
344 - avoid unneccessary ->cancel calls in Coro::Handle.
345 - maybe make it work on mingw32 with win32 backend
346 (based on patch by Yasuhiro Matsumoto).
347
3485.132 Fri May 29 09:00:39 CEST 2009
349 - do not keep a reference to the argument itself in
350 Coro::Semaphore::guard, as it could change later.
351 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
352 should support a prepare callback.
353
3545.131 Mon Mar 16 23:20:37 CET 2009
355 - implement and document Coro->suspend, Coro->resume.
356 - fix Coro::Select implementation to not (often) close
357 the passed file descriptors (testcase provided by pippijn).
358
3595.13 Mon Dec 15 21:51:42 CET 2008
360 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
361 - don't set diehook to C<undef> but instead to NULL, to avoid
362 spurious warnings.
363 - fix a lot of bugs in Coro::SemaphoreSet.
364 - Coro::SemaphoreSet will less often create a semaphore needlessly.
365 - add Coro::SemaphoreSet::count and wait methods.
366 - take advantage of the new Guard module.
367 - deprecate Coro::guard.
368 - try to fix the dreaded 01_unblock tests once more. I hate it when
369 testsuites need more fixing than the code they are supposed to test.
370 - croak in more cases when a required callback isn't resolvable.
371 - fix some minor issues in Coro::State->call/eval.
372 - use current coroutine context instead of a temporary one
373 when temporarily switching to another coroutine.
374 - do not call C-level on_destroy handlers during global destruction,
375 to avoid needless segfaults.
376
3775.12 Sun Dec 7 13:30:38 CET 2008
378 - add default config for MirOS, which seems to be a bug-to-bug
379 compatible fork of openbsd ("world domination by releasing
380 openbsd cvs before the openbsd folks do it" or so :).
381 - free_padlist did destroy the names pad, not good, but didn't
382 seem to bother perl - this could fix issues such as eval ""
383 inside a function called from multiple coroutines.
384 - use a different method to detect destruction time.
385 - be more careful when freeing padlists just before global
386 destruction.
387 - fixed and expanded the call/cc example.
388 - renamed _terminate to _coro_run.
389 - new method Coro::Channel->shutdown.
390 - try pthreads on openbsd <4.4 (broken sigaltstack, will
391 pthreads fare better?).
392 - be less picky about destroying "a" running coroutine.
393
3945.11 Tue Nov 25 21:49:05 CET 2008
395 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
396 working code with failed assertions, introducing more bugs than
397 it fixes, requiring elaborate workarounds :(
398
3995.1 Mon Nov 24 08:54:59 CET 2008
400 - wrote a small introductory tutorial - Coro::Intro.
401 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
402 - Coro::Select did errornously dup the file descriptors
403 and didn't work with all AnyEvent backends.
404 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
405 LWP data transfers.
406 - disassociate c contexts from coro objects - this is agruably more
407 correct, but mostly allows sharing of cctxs between coro and state
408 objects, for added memory savings and speed increases.
409 - bumped $Coro::POOL_RSS up to 32kb by default.
410 - no longer set the optype to OP_CUSTOM, as B::* understandably
411 doesn't like this very much (and we *are* a type of entersub).
412 - implement state cloning, just to prove that call/cc can be done.
413 - automatically load Coro::AnyEvent in Coro::Handle.
414 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
415 do not have this method (and don't need it either).
416 - speed up generic anyevent methods in Coro::Handle by using rouse
417 callbacks.
418 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
419 others. It also makes the debugger happier, as you can trace
420 through the idle threads now.
421 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
422 - eg/event was pretty broken.
423 - better 5.8.6 compatibility.
424
4255.0 Thu Nov 20 10:35:05 CET 2008
426 - NEW ARCHITECTURE: use the latest 4.x version if you experience
427 stability issues.
428 - bump API version to 7 - all dependents must be recompiled.
429 - removed timed_* functions - they were not being used anyways
430 and should be replaced by a more generic mechanism -
431 and were annoying to support anyways :)
432 - removed SemaphoreSet's waiter method - use sem method instead.
433 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
434 - async_pool did free a scalar value twice
435 ("Attempt to unreference...").
436 - fix a longstanding bug where calling terminate on a coro that
437 was waiting for a semaphore that was just becoming available
438 would cause a deadlock (semaphore would get into a state where
439 it was available but waiters were still blocked).
440 - calling throw on a coroutine that is waiting for a semaphore will
441 no longer make it acquire the semaphore (and thus leak a count).
442 - perl's process emulation is now not even theoretically supported
443 anymore.
444 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
445 conversion of callback-style to blocking-style.
446 - new methods $coro->schedule_to and ->cede_to, to specifically
447 schedule or cede to a specific coroutine.
448 - new function Coro::Semaphore::wait.
449 - use named constants in Coro::Channel (Richard Hundt).
450 - directly patch the entersub opcode calling SLF functions (cede,
451 transfer and so on). this does speed up context switching, but
452 more importanly, it frees us from the hardcoded behaviour of
453 entersub, so we might actually be able to return something from
454 those functions and atcually create new ones.
455 - take advantage of __builtin_frame_address on gcc.
456 - expose THX in coroapi (not sure whether this was a wise decision,
457 as "threaded" perls are running at half speed anyways).
458 - implement execute_slf (schedule-like-function) interface that makes
459 it possible to implement schedule-like-functions in XS.
460 - use new SLF interface to massively speed up Coro::EV by roughly a
461 factor of two.
462 - used new SLF interface to massively speed up Coro::Semaphore by a
463 factor of three.
464 - used new SLF interface to speed up Coro::AIO by roughly a factor of
465 four and reduce its memory usage considerably.
466 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
467 for a nice speedup and vastly more correct behaviour. Also implement
468 a new method "sem" to get at the underlying semaphore object.
469 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
470 (in comparison) 20-40% speedup.
471 - used new SLF interface to reimplement Coro::Signal gaining
472 some unknown (because I was too lazy), but certain, speedup, and also
473 making signals reliable for the first time.
474 - used new SLF interface and other optimisations to speed up async_pool
475 by a factor of two. It also doesn't rely on perl's exception mechanism
476 to exit anymore. The overhead for terminating an async_pool, coro over
477 a normal async is now very small.
478 - sped up coroutine creation/destruction by 40%.
479 - forgot to include Coro/libcoro/README in the dist for all these years.
480 - work around a freebsd pthreads bug (manual testcancel is required as
481 pthread_cond_wait isn't a cancellation point on freebsd).
482 - use new rouse functions to speed up and simplify Coro::BDB.
483 - make "prefer perl native functions" work with threaded perls.
484 - condense Coro::Debug ps output, hint at v and w flags.
485 - (libcoro) lots of minor cleanups and portability improvements.
486
4874.914 Wed Nov 19 12:54:18 CET 2008
488 - fix a disastrous bug in the readline optimisation
489 introduced in 4.801.
490
4914.913 Sat Nov 15 07:58:28 CET 2008
492 - async_pool did free a scalar value twice
493 ("Attempt to unreference...").
494
4954.912 Thu Nov 13 18:31:23 CET 2008
496 - minor cleanups.
497 - use much larger stacks on linux and perl < 5.8.8.
498 - Coro::Debug::new_unix_server did not unlink the socket
499 when destroyed.
500
5014.911 Tue Nov 11 04:26:17 CET 2008
502 - "port" to threaded perls.
503
5044.91 Mon Nov 10 05:36:38 CET 2008
505 - the ->throw exception object no longer leaks.
506 - creating a new cctx leaked a scopestack entry (memleak).
507 - new coroutines didn't get created with a zero flags field
508 (unknown impact).
509 - calling ->throw on a not-yet-started coroutine should now work
510 instead of being ignored.
511 - ->throw is now supported on Coro::State objects.
512 - clean up cctx creation code a bit.
513 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
514
5154.9 Sat Nov 8 17:45:27 CET 2008
9 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 516 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
10 - (libcoro) no longer leak threads in the experimental pthread backend, 517 - (libcoro) no longer leak threads in the experimental pthread backend,
11 also speed it up considerably. 518 also speed it up considerably.
12 - (libcoro) do not rely on makecontext passing void *'s unscathed. 519 - (libcoro) do not rely on makecontext passing void *'s unscathed.
13 - fix compiletime dependencies on libcoro in the Makefile. 520 - fix compiletime dependencies on libcoro in the Makefile.
14 - cctx_count wasn't always updated properly. 521 - cctx_count wasn't always updated properly.
… …
76 - remove debugging code related to MgPV_nolen_const, also try to 583 - remove debugging code related to MgPV_nolen_const, also try to
77 make it compile with perl 5.8.6 (yes, apple apparently loves 584 make it compile with perl 5.8.6 (yes, apple apparently loves
78 outdated software). Reported by John S. 585 outdated software). Reported by John S.
79 586
804.744 Tue Jul 8 22:06:35 CEST 2008 5874.744 Tue Jul 8 22:06:35 CEST 2008
81 - correctly provide default DIE/WARN handlers as documented. 588 - correctly provide default DIE/WARN handlers as documented.
82 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 589 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
83 implementations inside perl work fine for us. 590 implementations inside perl work fine for us.
84 591
854.743 Mon Jun 16 00:21:57 CEST 2008 5924.743 Mon Jun 16 00:21:57 CEST 2008
86 - when using Coro::EV without running EV::loop it could 593 - when using Coro::EV without running EV::loop it could
… …
108 - sprinkle "no warnings" freely over everything, also suppress 615 - sprinkle "no warnings" freely over everything, also suppress
109 warnings for some other modules. 616 warnings for some other modules.
110 - fix typo in WSAEWOULDBLOCK. 617 - fix typo in WSAEWOULDBLOCK.
111 618
1124.72 Sun May 25 05:14:36 CEST 2008 6194.72 Sun May 25 05:14:36 CEST 2008
113 - tweak META.yaml a bit, unfortunately, there is no documented way 620 - tweak META.yaml a bit, unfortunately, there is no documented way
114 to have optional dependencies with CPAN. doh :( 621 to have optional dependencies with CPAN. doh :(
115 - avoid running some tests on windows because they would fail due to 622 - avoid running some tests on windows because they would fail due to
116 perl bug (broken fork, broken pipes...). 623 perl bug (broken fork, broken pipes...).
117 - work around perl on windows bugs where perl returns undocumented 624 - work around perl on windows bugs where perl returns undocumented
118 error codes for sysread, syswrite etc. by taking advantage 625 error codes for sysread, syswrite etc. by taking advantage
… …
122 - use unix domain sockets in testsuite to work around 629 - use unix domain sockets in testsuite to work around
123 common perl implementation bugs on widows (they are emulated by 630 common perl implementation bugs on widows (they are emulated by
124 tcp sockets on windows. ugh.) 631 tcp sockets on windows. ugh.)
125 632
1264.71 Sat May 24 20:01:27 CEST 2008 6334.71 Sat May 24 20:01:27 CEST 2008
127 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 634 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
128 - take advantage of async name resolution of AnyEvent::Util. 635 - take advantage of async name resolution of AnyEvent::Util.
129 - work around brutal inet_aton override in Coro::LWP. 636 - work around brutal inet_aton override in Coro::LWP.
130 - take advantage of the readyhook in Coro::EV, for smoother 637 - take advantage of the readyhook in Coro::EV, for smoother
131 scheduling. 638 scheduling.
132 639
1334.7 Sun May 11 00:32:19 CEST 2008 6404.7 Sun May 11 00:32:19 CEST 2008
134 - completely reworked the Coro manpage. 641 - completely reworked the Coro manpage.
135 - added Coro::AnyEvent, generic event loop integration. 642 - added Coro::AnyEvent, generic event loop integration.
136 - implement cancel, ready and kill commands in Coro::Debug. 643 - implement cancel, ready and kill commands in Coro::Debug.
137 - document find_coro in Coro::Debug. 644 - document find_coro in Coro::Debug.
138 - incompatible change: rename has_stack to has_cctx. 645 - incompatible change: rename has_stack to has_cctx.
139 - Coro::AIO and Coro::BDB no longer force event model detection, 646 - Coro::AIO and Coro::BDB no longer force event model detection,
140 use AnyEvent::AIO and AnyEvent::BDB. 647 use AnyEvent::AIO and AnyEvent::BDB.
… …
264 771
2654.11 Thu Oct 11 02:40:24 CEST 2007 7724.11 Thu Oct 11 02:40:24 CEST 2007
266 - port to threaded perls. 773 - port to threaded perls.
267 774
2684.1 Thu Oct 11 02:38:16 CEST 2007 7754.1 Thu Oct 11 02:38:16 CEST 2007
269 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 776 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
270 be local to each coro (see Coro::State). 777 be local to each coro (see Coro::State).
271 - incompatible change: for very deep reasons, cede and cede_notself 778 - incompatible change: for very deep reasons, cede and cede_notself
272 cannot return anything, so nothing will be returned. 779 cannot return anything, so nothing will be returned.
273 - possibly bring back 5.10 compatibility (untested). 780 - possibly bring back 5.10 compatibility (untested).
274 - work around stupid (and wrong) warning on 5.10 :(. 781 - work around stupid (and wrong) warning on 5.10 :(.
… …
346 but stateful semantics. 853 but stateful semantics.
347 - fixed a lot of typos in Coro.pm (patch submitted by David 854 - fixed a lot of typos in Coro.pm (patch submitted by David
348 Steinbrunner, which applied flawlessly). 855 Steinbrunner, which applied flawlessly).
349 856
3503.6 Sat Apr 14 17:13:31 CEST 2007 8573.6 Sat Apr 14 17:13:31 CEST 2007
351 - added some bugfixes to get eg/myhttpd working again. 858 - added some bugfixes to get eg/myhttpd working again.
352 - added Coro::Storable for often-cede'ing freeze/thaw. 859 - added Coro::Storable for often-cede'ing freeze/thaw.
353 - try to do a clean exit when a coroutine calls exit 860 - try to do a clean exit when a coroutine calls exit
354 (EXPERIMENTAL). 861 (EXPERIMENTAL).
355 - got rid of indirect call through _coro_init. 862 - got rid of indirect call through _coro_init.
356 - updated the partly antique examples in eg/ to 863 - updated the partly antique examples in eg/ to

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines