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

Comparing Coro/Changes (file contents):
Revision 1.347 by root, Tue May 20 15:07:03 2008 UTC vs.
Revision 1.635 by root, Wed Sep 3 14:26:34 2014 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO. 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 run canceled thread?
7 9
10 - restore portability to perl 5.8.x.
11 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
12 up some memory, but fixes "use feature" and similar modules
13 when used inside a Coro.
14 - allow zero as argument to Coro::Channel to mean the same thing
15 as no argument. this works with older versions as well,
16 but wasn't legal until now.
17
186.39 Mon Jun 2 00:00:08 CEST 2014
19 - work around more incompatible changes in 5.20.
20
216.38 Sun Jun 1 21:54:23 CEST 2014
22 - check that perl slots actually have enough space to hold
23 interpreter variables.
24 - untested port to perl 5.19 (64 bit tmps indices) (reported
25 by Andreas König).
26 - croak when cancel is called without a thread context.
27
286.37 Tue Mar 4 13:27:33 CET 2014
29 - *sigh*, removed leftover debugging code from debugging a
30 perl bug, of all things.
31
326.36 Tue Mar 4 07:11:59 CET 2014
33 - semaphores would not clear the destroy hook when interrupted
34 by ->throw, causing segfaults or worse.
35 - ->throw on a thread waiting for a semaphore did not acquire
36 the semaphore, but also didn't wake up other waiters,
37 possibly causing a deadlock.
38 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
39 to report a full stacktrace, which should help find out
40 where the actual call is.
41 - "atomically" destroy data in slf_destroy, because it is
42 the right thing to do, just in case.
43 - disable ecb.h libm dependency, to compile on ancient systems
44 or under adverse conditions.
45
466.33 Mon Nov 18 11:26:27 CET 2013
47 - do not crash when freeing padlists with holes (in 5.18).
48 - tentative SVt_BIND 5.19 port/fix.
49
506.32 Tue Nov 5 15:35:35 CET 2013
51 - use a new algorithm to derive padlists for perl 5.18. The old
52 one could lead to 0-pointer accesses inside perl (reported
53 by Darin McBride).
54
556.31 Thu May 9 07:39:48 CEST 2013
56 - Coro::AIO requests would crash if the thread was ready'd
57 while the request was ongoing.
58
596.29 Wed May 8 02:55:18 CEST 2013
60 - when an on_destroy handler destructs the coro currently being
61 destructed a perl scalar could be accessed after being freed,
62 likely causing a crash.
63
646.28 Wed Mar 6 06:58:02 CET 2013
65 - clean remnants of existing __DIE__ and __WARN__ handlers so
66 they lose their magic and will not cause segfaults later
67 (testcase by Andrey Sagulin).
68 - improved Coro::State documentation a bit.
69 - Coro::Debug::command now flushes the output.
70 - add hack detection code for x32 abi, because the braindead slugs
71 who designed that made it look exactly like x86_64 without
72 providing proper compile time symbols to test for it. as a result,
73 this detection cannot work reliably.
74 - valgrind stack registering was broken.
75 - do not rely on Time::HiRes anymore in Coro::Debug.
76
776.23 Fri Dec 7 23:36:37 CET 2012
78 - use experimental fiber implementation on native windows
79 perls.
80 - use sizeof (void *) as multiplication factor for stack sizes,
81 to accomodate the totally braindamaged microsoft 64 bit "os".
82 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
83 activeperl 5.16 crashes when PerlIO_define_layer is called due
84 to some bug in the perl dll, strawberry perl at least passes
85 the testsuite.
86 - implement Coro::Handle->peeraddr/host/port, for slightly
87 improved compatibility with LWP.
88 - implement 5.17 compatibility by almost blindly applying a
89 good-looking patch by Father Chrysostomos.
90 - move stack management functions into libcoro 3.
91 - libcoro version 3 "released".
92 - support magic values as timed_io_once args.
93 - recommend AnyEvent 7+ or EV 4+, also require EV
94 version 4 or newer for Coro::EV.
95
966.10 Tue Oct 9 01:14:27 CEST 2012
97 - updated ecb.h, it had a typo that caused it to not compile on many
98 big endian systems (reported by many people).
99 - disable memory fences in ecb.h to improve portability.
100
1016.09 Sat Oct 6 23:25:02 CEST 2012
102 - Coro::EV I/O watchers were not interruptible by exceptions
103 (Coro::State::throw) (testcase by sten).
104 - ->throw now puts threads into the ready queue, as this seems to
105 be expected by existing code, and code that doesn't cope with spurious
106 wakeups needs fixing anyway.
107 - use fd -1 in mmap.
108 - cast I32 to int in error message printf.
109 - warn about broken so-called "hardened" kernels.
110
1116.08 Fri Apr 13 12:05:47 CEST 2012
112 - be more aggressive about exiting like perl does - formerly,
113 exiting from the non-main thread would not execute END blocks.
114
1156.07 Fri Nov 11 21:21:48 CET 2011
116 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
117 - update ecb.h.
118
1196.06 Mon Aug 8 23:59:48 CEST 2011
120 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
121 backend, which is an order of magnitude slower. unfortunately, cygwins
122 pthread implementation isn't very complete either, so allocate the stack
123 twice just to be sure.
124 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
125 in its ithreaded perl - best recompile cygwin and use the 'w'indows
126 backend for much better performance. also disable ithreads for
127 even better performance...).
128
1296.05 Thu Aug 4 21:36:36 CEST 2011
130 - blush, condvar values would not be propagated from send to recv anymore
131 (reported by Chip Salzenberg).
132 - use exponential increase for the readline buffer length in
133 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
134
1356.04 Wed Aug 3 17:15:45 CEST 2011
136 - use even more efficient and more compatible condvars for
137 compatibility to AnyEvent 6.x :)
138 - more inconsequential ecb.h updates.
139
1406.03 Wed Aug 3 11:41:30 CEST 2011
141 - change how Coro patches AnyEvent condvars for compatibility to
142 AnyEvent 6.x.
143 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
144 hasn't been included.
145
1466.02 Wed Jul 13 04:35:19 CEST 2011
147 - "improve portability to Gentoo" - gentoo manages to put perl variables
148 in memory areas that are farther than 2gb apart, which the jit couldn't
149 handle and barfed. now it's just a bit slower on gentoo and similar
150 systems.
151
1526.01 Sun Jul 3 12:31:14 CEST 2011
153 - workarounds are good, but the test for whether pthreads are used
154 was not good. this one should be better.
155 - check differently whether gcc generates cfi instructions itself.
156
1576.0 Wed Jun 29 19:43:35 CEST 2011
158 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
159 destroyed and cleaned up automatically (e.g. async { schedule }).
160 - implement a JIT compiler for part of the thread switch code,
161 which gives a 50% speed improvement on threaded perls, and
162 about 4% on non-threaded perls (so threaded perls now finally
163 reach about half the speed of non-threaded perls).
164 - slightly modernise Coro::Intro, add section about rouse functions.
165 - avoid DEFSV and ERRSV, giving another 10% improvement
166 in thread switching.
167 - Coro::State->is_destroyed is now called is_zombie.
168 - implement a Coro->safe_cancel method that might fail, but
169 cancels in a "safer" way if it succeeds.
170 - add preliminary support for DEBUGGING perls.
171 - get rid of two hash-accesses when initialising a new Coro - this
172 speeds up coro creation by almost a factor of two.
173 - croak when a coro that is being cancelled tries to block
174 (e.g. while executing a guard block), instead of crashing or
175 deadlocking.
176 - use a more robust and also faster method to identify Coro::State
177 objects - speeds up everything a bit.
178 - implement Coro->cancel in XS for a 20% speed improvement, and to
179 be able to implement mutual cancellation.
180 - speed up context switches by a percent or two by more efficiently
181 allocating context stack entries.
182 - implement Coro->join and Coro->on_destroy in XS for a speedup and
183 a reduction in memory use.
184 - cancelling a coro while it itself is cancelling another coro is
185 now supported and working, instead of triggering an assertion.
186 - be a bit more crash-resistant when calling (buggy) on_destroy
187 callbacks (best effort).
188 - move on_destroy into the slf_frame, to allow extension slf
189 functions to have destructors.
190 - get rid if coro refcounting - simply crash in other interpreter
191 threads by nulling the pointers on clone.
192 - simplify warn/die hook handling when loading Coro - the convoluted
193 logic seems to be no longer neccessary.
194 - use libecb instead of our own home-grown gcc hacks.
195 - document alternatives to Coro::LWP. Please use them :)
196 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
197 sigaltstack. Really. wine suffers from it, erlang suffers from it,
198 and it's known since at least 2006.
199
2005.372 Wed Feb 23 06:14:30 CET 2011
201 - apparently mingw doesn't provide a working gettimeofday, try to
202 work around that by relying on Time::HiRes (indirectly brought to
203 my attention by Max Maischein).
204 - fix some portability issues when Time::HiRes was used.
205
2065.371 Mon Feb 21 14:36:08 CET 2011
207 - backport to windows process emulation code again.
208
2095.37 Sat Feb 19 07:49:44 CET 2011
210 - add a big "Coro thread life cycle" section to "man Coro".
211 - try a tentative workaround against the breakage that 5.13 has
212 introduced without depreciation period. sigh.
213 - no longer use Time::HiRes if gettimeofday is available, which
214 saves quite a lot of memory.
215
2165.36 Sun Feb 13 05:33:41 CET 2011
217 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
218 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
219 method call.
220 - undocument Coro::Timer::sleep and obsolete whole module.
221 - optimise Coro::Timer::timeout memory and cpu usage.
222 - slightly updated Coro::Intro for recent changes.
223 - do not initialise PL_dirty anymore.
224
2255.25 Thu Nov 11 01:08:39 CET 2010
226 - try a different approach on netbsd - netbsd 5 finally has marginally
227 working pthreads, but still broken ucontext/sigaltstack.
228 - openbsd 4.8 finally got their act together, Coro works out of the box
229 with asm, setjmp and pthreads (no change, just informational).
230
2315.24 Sat Oct 23 11:27:12 CEST 2010
232 - port to the EV 4.0 API.
233 - work around bugs in mingw32, making strawberry perl work
234 out of the box.
235 - correctly modify Coro::AIO function prototypes
236 so that they reflect the "no optional parameters" rule.
237 - "ported" libcoro to C++.
238
2395.23 Mon May 17 18:50:42 CEST 2010
240 - be more resistant to ordering changes when initialising
241 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
242 Waldorf).
243 - document that perl 5.12 deliberately removed support for cloning.
244
2455.22 Wed Apr 14 03:55:35 CEST 2010
246 - correctly return udnef on errors in Coro::Handle::read/write
247 (testcase by Marc Mims).
248 - convert Coro::Util into a "perl compatibility wrapper" - the functions
249 are less useful now, but are drop-in replacements for existing
250 functions, listing better alternatives in the documentation. This also
251 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
252 with Coro::Util::inet_aton.
253 - do not override $Coro::idle unconditionally in Coro.pm, as other
254 modules could have provided their own idle coro already
255 (for exmaple, Coro::AnyEvent).
256 - fix Coro::Util::gethost* functions.
257 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
258
2595.21 Wed Dec 16 07:19:51 CET 2009
260 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
261 together.
262 - add some examples on how to combine other event loops with Coro in
263 Coro::AnyEvent, and how to run it (and not to block). Seems to be
264 the most common source of confusion.
265 - try to catch people naively blocking in an event callback.
266 - work around the perl filehandle bug issue in conjunction with
267 older common::sense (as indirectly pointed out by ZSystem).
268 - clarify the "not from signal handlers" section.
269
2705.2 Sun Oct 4 14:54:24 CEST 2009
271 - Coro::Storable destroyed the prototypes of the functions it wrapped.
272 - export rouse_cb and rouse_wait by default now.
273 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
274 - new method $state->swap_sv.
275 - added section on "windows process emulation" to the manpage,
276 after a not-so-fruitful (nor-friendly) "discussion" with chip
277 salzenberg (discussion implies arguments, but his only arguments
278 were ad-hominems, one wonders why he started it in the first
279 place). I hope this explains it well enough for him to understand,
280 and maybe well enough for others to understand.
281 - use common::sense everywhere now.
282 - idle callbacks are no longer supported, use idle coros instead.
283 - print a thread listing when a deadlock is detected.
284
2855.17 Sat Aug 22 23:09:31 CEST 2009
286 - work around a bug in the perl debugger causing crashes
287 when running under the debugger by marking _pool_handler as nodebug.
288 - speed up Coro::async considerably.
289 - try some hacks to get netbsd to work "more often" - their broken
290 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
291 - convert Coro to AE 5.0 API.
292
2935.162 Tue Jul 28 04:04:03 CEST 2009
294 - perl 5.8.2 is now minimum requirement.
295 - skip t/19_handle.t on broken windows perls.
296
2975.161 Wed Jul 22 04:47:38 CEST 2009
298 - Coro::AnyEvent::poll could have a different prototype when EV was
299 used as backend (analyzed by Tatsuhiko Miyagawa).
300 - Coro::AnyEvent errornously initialised the event loop when loaded,
301 not on demand.
302 - try to workaround rare */t/01_unblock.t failures.
303
3045.16 Tue Jul 21 01:44:37 CEST 2009
305 - Coro::AnyEvent failed to hook into the event loop
306 when no threads had been readied between detecting
307 the event loop and actually running it.
308 - considerably speed up Coro::Select by taking avdantage
309 of AnyEvent > 4.8 and some other optimisations.
310 - implement paragraph readline mode in Coro::Handle
311 (based on patches by Zsbán Ambrus).
312 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
313 and analyzed by Yasuhiro MATSUMOTO).
314 - clarified libcoro license and copyright.
315 - someone stole my EXTRA_META!!!
316 - implement Coro::Select::patch_pp_sselect and it's brother,
317 for hardcode select overriding.
318
3195.151 Mon Jul 6 05:41:57 CEST 2009
320 - backport to windows process emulation code again (patch by
321 Yasuhiro MATSUMOTO).
322 - slightly update Coro::MakeMaker.
323
3245.15 Tue Jun 30 10:28:06 CEST 2009
325 - deprecate Coro::Socket, document how to get ipv6 support via
326 AnyEvent::Socket instead.
327 - implement signal->wait ($cb) interface, similar to semaphores.
328 - work around SvOK not supporting getmagic, so we have to getmagic
329 to test for undef :/ (reported by Matthias Waldorf).
330 - load Coro::AnyEvent in all modules using AnyEvent.
331 - work around perl corrupting our internal data structures,
332 reported by Tokuhiro Matsuno.
333 - enable per-coroutine real and cpu time gathering
334 (Coro::State::enable_times).
335
3365.14 Wed Jun 24 01:37:48 CEST 2009
337 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
338 as this seems to have been a difficult concept (poll, sleep, idle,
339 idle_upto).
340 - add Coro::AnyEvent::readable/writable functions.
341 - clarify Coro::EV/Event/AnyEvent manpages.
342 - free per-thread global scalars in the thread calling ->cancel, to
343 avoid crashes when $_, $@ etc., are magical but some of those
344 had already been freed.
345 - "unexperimentalise" the callback interface for Coro::Semaphore.
346 - speed up ready queue management/context switching by using a linked
347 list instead of an array (~5-10%).
348 - implement "watch" command in Coro::Debug shells.
349 - for fun, implement time-slicing as an example in the manpage.
350 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
351 Coro::Event, use the normal AnyEvent handling instead of dieing
352 (the same is true for Coro::Handle).
353 - properly document Coro::EV::timed_io_once.
354 - avoid unneccessary ->cancel calls in Coro::Handle.
355 - maybe make it work on mingw32 with win32 backend
356 (based on patch by Yasuhiro Matsumoto).
357
3585.132 Fri May 29 09:00:39 CEST 2009
359 - do not keep a reference to the argument itself in
360 Coro::Semaphore::guard, as it could change later.
361 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
362 should support a prepare callback.
363
3645.131 Mon Mar 16 23:20:37 CET 2009
365 - implement and document Coro->suspend, Coro->resume.
366 - fix Coro::Select implementation to not (often) close
367 the passed file descriptors (testcase provided by pippijn).
368
3695.13 Mon Dec 15 21:51:42 CET 2008
370 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
371 - don't set diehook to C<undef> but instead to NULL, to avoid
372 spurious warnings.
373 - fix a lot of bugs in Coro::SemaphoreSet.
374 - Coro::SemaphoreSet will less often create a semaphore needlessly.
375 - add Coro::SemaphoreSet::count and wait methods.
376 - take advantage of the new Guard module.
377 - deprecate Coro::guard.
378 - try to fix the dreaded 01_unblock tests once more. I hate it when
379 testsuites need more fixing than the code they are supposed to test.
380 - croak in more cases when a required callback isn't resolvable.
381 - fix some minor issues in Coro::State->call/eval.
382 - use current coroutine context instead of a temporary one
383 when temporarily switching to another coroutine.
384 - do not call C-level on_destroy handlers during global destruction,
385 to avoid needless segfaults.
386
3875.12 Sun Dec 7 13:30:38 CET 2008
388 - add default config for MirOS, which seems to be a bug-to-bug
389 compatible fork of openbsd ("world domination by releasing
390 openbsd cvs before the openbsd folks do it" or so :).
391 - free_padlist did destroy the names pad, not good, but didn't
392 seem to bother perl - this could fix issues such as eval ""
393 inside a function called from multiple coroutines.
394 - use a different method to detect destruction time.
395 - be more careful when freeing padlists just before global
396 destruction.
397 - fixed and expanded the call/cc example.
398 - renamed _terminate to _coro_run.
399 - new method Coro::Channel->shutdown.
400 - try pthreads on openbsd <4.4 (broken sigaltstack, will
401 pthreads fare better?).
402 - be less picky about destroying "a" running coroutine.
403
4045.11 Tue Nov 25 21:49:05 CET 2008
405 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
406 working code with failed assertions, introducing more bugs than
407 it fixes, requiring elaborate workarounds :(
408
4095.1 Mon Nov 24 08:54:59 CET 2008
410 - wrote a small introductory tutorial - Coro::Intro.
411 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
412 - Coro::Select did errornously dup the file descriptors
413 and didn't work with all AnyEvent backends.
414 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
415 LWP data transfers.
416 - disassociate c contexts from coro objects - this is agruably more
417 correct, but mostly allows sharing of cctxs between coro and state
418 objects, for added memory savings and speed increases.
419 - bumped $Coro::POOL_RSS up to 32kb by default.
420 - no longer set the optype to OP_CUSTOM, as B::* understandably
421 doesn't like this very much (and we *are* a type of entersub).
422 - implement state cloning, just to prove that call/cc can be done.
423 - automatically load Coro::AnyEvent in Coro::Handle.
424 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
425 do not have this method (and don't need it either).
426 - speed up generic anyevent methods in Coro::Handle by using rouse
427 callbacks.
428 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
429 others. It also makes the debugger happier, as you can trace
430 through the idle threads now.
431 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
432 - eg/event was pretty broken.
433 - better 5.8.6 compatibility.
434
4355.0 Thu Nov 20 10:35:05 CET 2008
436 - NEW ARCHITECTURE: use the latest 4.x version if you experience
437 stability issues.
438 - bump API version to 7 - all dependents must be recompiled.
439 - removed timed_* functions - they were not being used anyways
440 and should be replaced by a more generic mechanism -
441 and were annoying to support anyways :)
442 - removed SemaphoreSet's waiter method - use sem method instead.
443 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
444 - async_pool did free a scalar value twice
445 ("Attempt to unreference...").
446 - fix a longstanding bug where calling terminate on a coro that
447 was waiting for a semaphore that was just becoming available
448 would cause a deadlock (semaphore would get into a state where
449 it was available but waiters were still blocked).
450 - calling throw on a coroutine that is waiting for a semaphore will
451 no longer make it acquire the semaphore (and thus leak a count).
452 - perl's process emulation is now not even theoretically supported
453 anymore.
454 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
455 conversion of callback-style to blocking-style.
456 - new methods $coro->schedule_to and ->cede_to, to specifically
457 schedule or cede to a specific coroutine.
458 - new function Coro::Semaphore::wait.
459 - use named constants in Coro::Channel (Richard Hundt).
460 - directly patch the entersub opcode calling SLF functions (cede,
461 transfer and so on). this does speed up context switching, but
462 more importanly, it frees us from the hardcoded behaviour of
463 entersub, so we might actually be able to return something from
464 those functions and atcually create new ones.
465 - take advantage of __builtin_frame_address on gcc.
466 - expose THX in coroapi (not sure whether this was a wise decision,
467 as "threaded" perls are running at half speed anyways).
468 - implement execute_slf (schedule-like-function) interface that makes
469 it possible to implement schedule-like-functions in XS.
470 - use new SLF interface to massively speed up Coro::EV by roughly a
471 factor of two.
472 - used new SLF interface to massively speed up Coro::Semaphore by a
473 factor of three.
474 - used new SLF interface to speed up Coro::AIO by roughly a factor of
475 four and reduce its memory usage considerably.
476 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
477 for a nice speedup and vastly more correct behaviour. Also implement
478 a new method "sem" to get at the underlying semaphore object.
479 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
480 (in comparison) 20-40% speedup.
481 - used new SLF interface to reimplement Coro::Signal gaining
482 some unknown (because I was too lazy), but certain, speedup, and also
483 making signals reliable for the first time.
484 - used new SLF interface and other optimisations to speed up async_pool
485 by a factor of two. It also doesn't rely on perl's exception mechanism
486 to exit anymore. The overhead for terminating an async_pool, coro over
487 a normal async is now very small.
488 - sped up coroutine creation/destruction by 40%.
489 - forgot to include Coro/libcoro/README in the dist for all these years.
490 - work around a freebsd pthreads bug (manual testcancel is required as
491 pthread_cond_wait isn't a cancellation point on freebsd).
492 - use new rouse functions to speed up and simplify Coro::BDB.
493 - make "prefer perl native functions" work with threaded perls.
494 - condense Coro::Debug ps output, hint at v and w flags.
495 - (libcoro) lots of minor cleanups and portability improvements.
496
4974.914 Wed Nov 19 12:54:18 CET 2008
498 - fix a disastrous bug in the readline optimisation
499 introduced in 4.801.
500
5014.913 Sat Nov 15 07:58:28 CET 2008
502 - async_pool did free a scalar value twice
503 ("Attempt to unreference...").
504
5054.912 Thu Nov 13 18:31:23 CET 2008
506 - minor cleanups.
507 - use much larger stacks on linux and perl < 5.8.8.
508 - Coro::Debug::new_unix_server did not unlink the socket
509 when destroyed.
510
5114.911 Tue Nov 11 04:26:17 CET 2008
512 - "port" to threaded perls.
513
5144.91 Mon Nov 10 05:36:38 CET 2008
515 - the ->throw exception object no longer leaks.
516 - creating a new cctx leaked a scopestack entry (memleak).
517 - new coroutines didn't get created with a zero flags field
518 (unknown impact).
519 - calling ->throw on a not-yet-started coroutine should now work
520 instead of being ignored.
521 - ->throw is now supported on Coro::State objects.
522 - clean up cctx creation code a bit.
523 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
524
5254.9 Sat Nov 8 17:45:27 CET 2008
526 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
527 - (libcoro) no longer leak threads in the experimental pthread backend,
528 also speed it up considerably.
529 - (libcoro) do not rely on makecontext passing void *'s unscathed.
530 - fix compiletime dependencies on libcoro in the Makefile.
531 - cctx_count wasn't always updated properly.
532 - Coro::State::cctx_stacksize wasn't applied correctly.
533 - new function Coro::State::cctx_max_idle.
534 - the default max number of idle C contexts is now 4.
535 - (libcoro) try harder to get _setjmp/_longjmp.
536 - (libcoro) cleanup and extend the libcoro API to officially
537 allow the creation of empty source contexts.
538 - very experimental workaround for the totally broken netbsd platform.
539 - tried to hack around openbsd bugs.
540
5414.804 Wed Nov 5 16:36:00 CET 2008
542 - Coro::Debug::new_unix_server would not create a non-blocking listening
543 socket, sometimes causing freezes.
544 - (libcoro) fix misaligned stack points for setjmp and assembly
545 methods, which can cause crashes on x86/x86_64 with a sufficiently
546 aggressive compiler.
547 - new function: Coro::Debug::new_tcp_server.
548 - move ->throw into the Coro class because it only works on coro objects.
549
5504.803 Mon Nov 3 17:16:12 CET 2008
551 - (libcoro) use a global asm statement to become independent of gcc
552 otpimisations for CORO_ASM (thanks to pippijn for the idea).
553 - try to workaround yet another broken bsd, this time dragonfly.
554
5554.802 Thu Oct 30 10:56:12 CET 2008
556 - support -fno-omit-frame-pointer on x86 with the assembly method.
557 - tune 01_unblock.t tests a bit.
558
5594.801 Wed Oct 22 18:33:37 CEST 2008
560 - improve readline speed for very long "lines".
561 - backport to 5.8.8.
562
5634.8 Thu Oct 2 13:34:40 CEST 2008
564 - new function Coro::AIO::aio_wait.
565 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
566 - greatly speed up and reduce memory usage of Coro::AIO requests.
567 - implement some other µ-optimisations.
568
5694.749 Mon Sep 29 14:40:12 CEST 2008
570 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
571
5724.748 Sat Sep 27 14:03:19 CEST 2008
573 - implement, but do not document, PerlIO::cede(granularity).
574 - Coro::Storable forgot to wrap Storable::pstore.
575 - work around the multitude of leaks and memory corruption
576 bugs in PerlIO::via by using our own C-level perliol. As a side
577 effect, Coro::Storable is now much, much, much faster.
578
5794.747 Tue Sep 23 01:59:41 CEST 2008
580 - fix a per-cv memleak (one empty array was leaked per
581 code reference).
582 - avoid a crash in coro->call|rss when the coroutine was already
583 destroyed (most noticably when using Coro::Debug::ps :)
584 - also protect *Storable::FILE.
585 - push up default storable granularity to 20ms.
586
5874.746 Sun Sep 21 03:22:20 CEST 2008
588 - be more insistent on locking Storable against reentrancy
589 in Coro::Storable.
590 - move swap_def?v and throw to Coro::State, as documented.
591
5924.745 Thu Jul 24 00:14:38 CEST 2008
593 - remove debugging code related to MgPV_nolen_const, also try to
594 make it compile with perl 5.8.6 (yes, apple apparently loves
595 outdated software). Reported by John S.
596
5974.744 Tue Jul 8 22:06:35 CEST 2008
598 - correctly provide default DIE/WARN handlers as documented.
599 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
600 implementations inside perl work fine for us.
601
6024.743 Mon Jun 16 00:21:57 CEST 2008
603 - when using Coro::EV without running EV::loop it could
604 result in busy-waiting for events, this has been fixed.
605 - reduce codesize and improve performance by using EV_DEFAULT_UC.
606
6074.742 Sat May 31 14:10:21 CEST 2008
608 - implement a workaround for (some) perl <5.8.8 versions.
609 - require EV 3.3+.
610
6114.741 Fri May 30 23:33:09 CEST 2008
612 - tell netbsd how utterly broken their imitation of an OS is
613 and refuse to build by default if pthreads are in use.
614 - switch to "s" method on all bsd's by default, as their ucontext
615 stuff seems just too broken.
616 - fix a bug in Coro::Select.
617
6184.74 Thu May 29 20:05:31 CEST 2008
619 - do not test Coro::LWP for lack of dependencies.
620 - do not test Coro::Select for lack of working perls.
621
6224.73 Thu May 29 2008
623 - fix a bug in Coro::EV which would cause it to block despite
624 there being runnable coroutines.
625 - sprinkle "no warnings" freely over everything, also suppress
626 warnings for some other modules.
627 - fix typo in WSAEWOULDBLOCK.
628
6294.72 Sun May 25 05:14:36 CEST 2008
630 - tweak META.yaml a bit, unfortunately, there is no documented way
631 to have optional dependencies with CPAN. doh :(
632 - avoid running some tests on windows because they would fail due to
633 perl bug (broken fork, broken pipes...).
634 - work around perl on windows bugs where perl returns undocumented
635 error codes for sysread, syswrite etc. by taking advantage
636 of AnyEvent's workaround for that problem.
637 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
638 a common perl implementation bug on windows.
639 - use unix domain sockets in testsuite to work around
640 common perl implementation bugs on widows (they are emulated by
641 tcp sockets on windows. ugh.)
642
6434.71 Sat May 24 20:01:27 CEST 2008
644 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
8 - take advantage of async name resolution of AnyEvent::Util. 645 - take advantage of async name resolution of AnyEvent::Util.
9 - work around brutal inet_aton override in Coro::LWP. 646 - work around brutal inet_aton override in Coro::LWP.
647 - take advantage of the readyhook in Coro::EV, for smoother
648 scheduling.
10 649
114.7 Sun May 11 00:32:19 CEST 2008 6504.7 Sun May 11 00:32:19 CEST 2008
12 - completely reworked the Coro manpage. 651 - completely reworked the Coro manpage.
13 - added Coro::AnyEvent, generic event loop integration. 652 - added Coro::AnyEvent, generic event loop integration.
14 - implement cancel, ready and kill commands in Coro::Debug. 653 - implement cancel, ready and kill commands in Coro::Debug.
15 - document find_coro in Coro::Debug. 654 - document find_coro in Coro::Debug.
16 - incompatible change: rename has_stack to has_cctx. 655 - incompatible change: rename has_stack to has_cctx.
17 - Coro::AIO and Coro::BDB no longer force event model detection, 656 - Coro::AIO and Coro::BDB no longer force event model detection,
18 use AnyEvent::AIO and AnyEvent::BDB. 657 use AnyEvent::AIO and AnyEvent::BDB.
… …
142 781
1434.11 Thu Oct 11 02:40:24 CEST 2007 7824.11 Thu Oct 11 02:40:24 CEST 2007
144 - port to threaded perls. 783 - port to threaded perls.
145 784
1464.1 Thu Oct 11 02:38:16 CEST 2007 7854.1 Thu Oct 11 02:38:16 CEST 2007
147 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 786 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
148 be local to each coro (see Coro::State). 787 be local to each coro (see Coro::State).
149 - incompatible change: for very deep reasons, cede and cede_notself 788 - incompatible change: for very deep reasons, cede and cede_notself
150 cannot return anything, so nothing will be returned. 789 cannot return anything, so nothing will be returned.
151 - possibly bring back 5.10 compatibility (untested). 790 - possibly bring back 5.10 compatibility (untested).
152 - work around stupid (and wrong) warning on 5.10 :(. 791 - work around stupid (and wrong) warning on 5.10 :(.
… …
224 but stateful semantics. 863 but stateful semantics.
225 - fixed a lot of typos in Coro.pm (patch submitted by David 864 - fixed a lot of typos in Coro.pm (patch submitted by David
226 Steinbrunner, which applied flawlessly). 865 Steinbrunner, which applied flawlessly).
227 866
2283.6 Sat Apr 14 17:13:31 CEST 2007 8673.6 Sat Apr 14 17:13:31 CEST 2007
229 - added some bugfixes to get eg/myhttpd working again. 868 - added some bugfixes to get eg/myhttpd working again.
230 - added Coro::Storable for often-cede'ing freeze/thaw. 869 - added Coro::Storable for often-cede'ing freeze/thaw.
231 - try to do a clean exit when a coroutine calls exit 870 - try to do a clean exit when a coroutine calls exit
232 (EXPERIMENTAL). 871 (EXPERIMENTAL).
233 - got rid of indirect call through _coro_init. 872 - got rid of indirect call through _coro_init.
234 - updated the partly antique examples in eg/ to 873 - updated the partly antique examples in eg/ to

Diff Legend

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