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

Comparing Coro/Changes (file contents):
Revision 1.122 by root, Fri Dec 16 01:44:11 2005 UTC vs.
Revision 1.693 by root, Thu Aug 19 22:43:02 2021 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: channel->maxsize(newsize)?
5TODO: __GCC_HAVE_DWARF2_CFI_ASM
6TODO: swap_sv, maybe add scope_swap_sv?
7TODO: croak when async_pool tries to run canceled thread?
8TODO: how to debug safe_cancel failure (vebrose mode? show how to make backtrace?)
9TODO: scalar context for Coro::AIO wrappers?
10
11TODO: State.xs could short-circuit some macros such as CORO_EXECUTE_SLF
12TODO: use frame->destroy to improve Coro::EV
13
14TODO: concurrent ->recv on same condvar
15sub Coro::AnyEvent::CondVar::_send {
16 if ($_[0]{_ae_coro}) {
17 $_->ready
18 for values %{ delete $_[0]{_ae_coro} }
19 }
20};
21
22sub Coro::AnyEvent::CondVar::_wait {
23 until ($_[0]{_ae_sent}) {
24 $AnyEvent::CondVar::Base::WAITING = 0; # avoid recursive check by AnyEvent
25 local $_[0]{_ae_coro}{$Coro::current+0} = $Coro::current;
26 Coro::schedule;
27 }
28};
29
30 - potential fix for an interesting type of recursive coro
31 destruction (testcase by Christopher J. Madsen).
32 - change jit executable strategy, to hopefully appease more
33 hardened environments.
34
356.57 Wed Jul 29 15:08:44 CEST 2020
36 - override AnyEvent condvar methods "properly", as AnyEvent
37 has support for pluggable condvars since 2011. This also
38 avoids problems with recursive waits by avoiding to mix
39 impl methods with coro code.
40 - allow rouse_wait to be called in a different thread.
41 - slightly clarify rouse_cb/rouse_wait documentation.
42 - slightly improve rouse callback performance and memory usage.
43 - try to report correct calller when throwing exception in
44 on $condvar->recv, not a useless line in Coro::AnyEvent.
45
466.56 Thu Feb 20 19:40:07 CET 2020
47 - Coro::EV::time*once functions used ev_once, which is not cancelable.
48 This could cause problems due to I/O watchers not being stopped up
49 on exceptions/cancellation.
50
516.55 Fri Jul 19 17:28:09 CEST 2019
52 - Coro::RWLock woke up only one reader instead of all after a wrlock
53 (analyzed and patch by Maxim Skorobogatov).
54 - slightly improved documentation of Coro::RWLock and RWLock testcase.
55 - quickly (re)-ported to minix 3.3 before minix crashed again.
56
576.54 Sun Dec 16 11:06:07 CET 2018
58 - reinstate Coro::Select::select prototype - it breaks more than it fixes.
59
606.53 Sun Dec 16 10:33:15 CET 2018
61 - the autogenerated prototypes of Coro::AIO now add ";" at the end
62 to avoid generating named unary operators and messing up parsing
63 (mostly affects aio_mlockall 4 | 8 which was parsed as
64 (aio_mlockall 4) | 8).
65 - do not use a prototype for select in Coro::Select, as CORE::select also
66 doesn't have one.
67
686.52 Tue Aug 14 18:49:43 CEST 2018
69 - libcoro update: allow sharing of coro threads using the pthreads
70 backend among different pthreads, as is required by Coro::Multicore.
71 - always allow ->call and ->eval when the coro is the current coroutine. this
72 makes it possible, among other things, to call these on $Coro::main.
73 - minor code cleanups.
74
756.514 Thu Aug 31 18:25:31 CEST 2017
76 - safe_cancel no longer croaks when called on already-destroyed
77 thread, but instead "just works".
78
796.513 Sat Jul 15 01:18:51 CEST 2017
80 - experimental 5.22 comaptibility changes (PadlistNAMES).
81 - removed attributes example and made sure the other examples
82 still work.
83 - fix some statevar types for newer perls, should be of no consequence.
84
856.512 Fri Jul 14 05:03:17 CEST 2017
86 - seems to pass test suite down till 5.8.9, so make this the
87 minimum requirement in Makefile.PL.
88 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
89 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
90 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
91 fortified longjmp variants in some libcs.
92 - re-try JIT allocation without PROT_EXEC, to hopefully improve
93 portability (still doesn't seem to work around broken selinux
94 mmap everywhere).
95 - speling fixes (Xavier Guimard).
96
976.511 Sun Jun 26 23:44:50 CEST 2016
98 - make it compile with threaded perls again.
99 - simplify and speed up __DIE__ and __WARN__ handling.
100
1016.51 Sat Jun 25 20:57:20 CEST 2016
102 - 6.5 release didn't compile on most older perls due to
103 a logic error enabling experimental 5.24 code for all versions.
104 - since various broken or deficient patches float around
105 to work around the vtbl API breakage in perl, let's provide
106 our own workaround, which seems to be much less invasive
107 and more compatible than the approaches seen so far. YMMV.
108 - work around assertion failure in perl_destruct on 5.24
109 (analyzed by Dave Mitchell).
110
1116.5 Wed Jun 22 22:23:50 CEST 2016
112 - swap_sv swapping order was not symmetrical, causing
113 wrong swaps when swap_sv was used multiple times on the
114 same sv in the same thread.
115 - swap_sv calls can now be undone by calling it again
116 with the same variables.
117 - swap_sv calls will now be undone in async_pool threads.
118 - split Coro::Semaphore::up/adjust into separate xs functions
119 for better error reporting, at slight codesize increase.
120 - (libcoro) arm assembly support, please test and report.
121 - adjust to PL_savestack changes in perl 5.24 (adapted from
122 the debian patch, which unfortunately gets it wrong).
123
1246.49 Sat Oct 17 01:40:12 CEST 2015
125 - throwing an exception to a thread waiting in
126 Coro::Handle using Coro::EV did not stop the watchers,
127 causing the next call to fail (testcase by Martin Pritchard).
128 - bump minimum perl version to 5.10.
129
1306.48 Sun Oct 4 19:03:51 CEST 2015
131 - fix memory leak when Coro::Handle uses Coro::EV internally
132 (testcase by Sten Sten).
133 - update code to libev 4 API (internally, Coro still used the
134 version 3 API calls).
135 - remove unused hv_sig.
136
1376.47 Sat Jul 11 03:58:20 CEST 2015
138 - swap_svs - NVs can be in the SV head beginning with 5.20.
139
1406.46 Tue Jun 30 14:36:30 CEST 2015
141 - restore portability to perls with windows fork emulation
142 (patch by Petr Písař).
143
1446.45 Tue Jun 30 01:40:08 CEST 2015
145 - restore portability to perl 5.16 and below.
146
1476.44 Tue Jun 30 00:41:54 CEST 2015
148 - avoid segfaulting (or worse) when tracing a canceled thread.
149 - the Event module silently broke it's public hook API, causing
150 failures on perls compiled with -Duselongdouble. Requiring
151 the latest version of Event because I don't know which release
152 changed this (it's not mentioned in the ChangeLog).
153 - allow xs level enterleave hooks via CoroAPI.
154
1556.43 Thu Jun 4 15:38:14 CEST 2015
156 - use stability canary.
157 - port to stableperl-5.22.0-1.001.
158 - update libecb.
159
1606.42 Wed Feb 11 20:29:52 CET 2015
161 - Coro::SemaphoreSet->try did not actually work (analyzed by
162 SATO Kentaro).
163 - upgrade libecb to be C11 compliant.
164
1656.41 Sat Sep 6 22:08:46 CEST 2014
166 - restore portability to perl 5.8.x.
167 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
168 up some memory, but fixes "use feature" and similar modules
169 when used inside a Coro.
170 - allow zero as argument to Coro::Channel to mean the same thing
171 as no argument. this works with older versions as well,
172 but wasn't legal until now.
173 - slightly better c header file detection.
174
1756.39 Mon Jun 2 00:00:08 CEST 2014
176 - work around more incompatible changes in 5.20.
177
1786.38 Sun Jun 1 21:54:23 CEST 2014
179 - check that perl slots actually have enough space to hold
180 interpreter variables.
181 - untested port to perl 5.19 (64 bit tmps indices) (reported
182 by Andreas König).
183 - croak when cancel is called without a thread context.
184
1856.37 Tue Mar 4 13:27:33 CET 2014
186 - *sigh*, removed leftover debugging code from debugging a
187 perl bug, of all things.
188
1896.36 Tue Mar 4 07:11:59 CET 2014
190 - semaphores would not clear the destroy hook when interrupted
191 by ->throw, causing segfaults or worse.
192 - ->throw on a thread waiting for a semaphore did not acquire
193 the semaphore, but also didn't wake up other waiters,
194 possibly causing a deadlock.
195 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
196 to report a full stacktrace, which should help find out
197 where the actual call is.
198 - "atomically" destroy data in slf_destroy, because it is
199 the right thing to do, just in case.
200 - disable ecb.h libm dependency, to compile on ancient systems
201 or under adverse conditions.
202
2036.33 Mon Nov 18 11:26:27 CET 2013
204 - do not crash when freeing padlists with holes (in 5.18).
205 - tentative SVt_BIND 5.19 port/fix.
206
2076.32 Tue Nov 5 15:35:35 CET 2013
208 - use a new algorithm to derive padlists for perl 5.18. The old
209 one could lead to 0-pointer accesses inside perl (reported
210 by Darin McBride).
211
2126.31 Thu May 9 07:39:48 CEST 2013
213 - Coro::AIO requests would crash if the thread was ready'd
214 while the request was ongoing.
215
2166.29 Wed May 8 02:55:18 CEST 2013
217 - when an on_destroy handler destructs the coro currently being
218 destructed a perl scalar could be accessed after being freed,
219 likely causing a crash.
220
2216.28 Wed Mar 6 06:58:02 CET 2013
222 - clean remnants of existing __DIE__ and __WARN__ handlers so
223 they lose their magic and will not cause segfaults later
224 (testcase by Andrey Sagulin).
225 - improved Coro::State documentation a bit.
226 - Coro::Debug::command now flushes the output.
227 - add hack detection code for x32 abi, because the braindead slugs
228 who designed that made it look exactly like x86_64 without
229 providing proper compile time symbols to test for it. as a result,
230 this detection cannot work reliably.
231 - valgrind stack registering was broken.
232 - do not rely on Time::HiRes anymore in Coro::Debug.
233
2346.23 Fri Dec 7 23:36:37 CET 2012
235 - use experimental fiber implementation on native windows
236 perls.
237 - use sizeof (void *) as multiplication factor for stack sizes,
238 to accomodate the totally braindamaged microsoft 64 bit "os".
239 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
240 activeperl 5.16 crashes when PerlIO_define_layer is called due
241 to some bug in the perl dll, strawberry perl at least passes
242 the testsuite.
243 - implement Coro::Handle->peeraddr/host/port, for slightly
244 improved compatibility with LWP.
245 - implement 5.17 compatibility by almost blindly applying a
246 good-looking patch by Father Chrysostomos.
247 - move stack management functions into libcoro 3.
248 - libcoro version 3 "released".
249 - support magic values as timed_io_once args.
250 - recommend AnyEvent 7+ or EV 4+, also require EV
251 version 4 or newer for Coro::EV.
252
2536.10 Tue Oct 9 01:14:27 CEST 2012
254 - updated ecb.h, it had a typo that caused it to not compile on many
255 big endian systems (reported by many people).
256 - disable memory fences in ecb.h to improve portability.
257
2586.09 Sat Oct 6 23:25:02 CEST 2012
259 - Coro::EV I/O watchers were not interruptible by exceptions
260 (Coro::State::throw) (testcase by sten).
261 - ->throw now puts threads into the ready queue, as this seems to
262 be expected by existing code, and code that doesn't cope with spurious
263 wakeups needs fixing anyway.
264 - use fd -1 in mmap.
265 - cast I32 to int in error message printf.
266 - warn about broken so-called "hardened" kernels.
267
2686.08 Fri Apr 13 12:05:47 CEST 2012
269 - be more aggressive about exiting like perl does - formerly,
270 exiting from the non-main thread would not execute END blocks.
271
2726.07 Fri Nov 11 21:21:48 CET 2011
273 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
274 - update ecb.h.
275
2766.06 Mon Aug 8 23:59:48 CEST 2011
277 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
278 backend, which is an order of magnitude slower. unfortunately, cygwins
279 pthread implementation isn't very complete either, so allocate the stack
280 twice just to be sure.
281 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
282 in its ithreaded perl - best recompile cygwin and use the 'w'indows
283 backend for much better performance. also disable ithreads for
284 even better performance...).
285
2866.05 Thu Aug 4 21:36:36 CEST 2011
287 - blush, condvar values would not be propagated from send to recv anymore
288 (reported by Chip Salzenberg).
289 - use exponential increase for the readline buffer length in
290 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
291
2926.04 Wed Aug 3 17:15:45 CEST 2011
293 - use even more efficient and more compatible condvars for
294 compatibility to AnyEvent 6.x :)
295 - more inconsequential ecb.h updates.
296
2976.03 Wed Aug 3 11:41:30 CEST 2011
298 - change how Coro patches AnyEvent condvars for compatibility to
299 AnyEvent 6.x.
300 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
301 hasn't been included.
302
3036.02 Wed Jul 13 04:35:19 CEST 2011
304 - "improve portability to Gentoo" - gentoo manages to put perl variables
305 in memory areas that are farther than 2gb apart, which the jit couldn't
306 handle and barfed. now it's just a bit slower on gentoo and similar
307 systems.
308
3096.01 Sun Jul 3 12:31:14 CEST 2011
310 - workarounds are good, but the test for whether pthreads are used
311 was not good. this one should be better.
312 - check differently whether gcc generates cfi instructions itself.
313
3146.0 Wed Jun 29 19:43:35 CEST 2011
315 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
316 destroyed and cleaned up automatically (e.g. async { schedule }).
317 - implement a JIT compiler for part of the thread switch code,
318 which gives a 50% speed improvement on threaded perls, and
319 about 4% on non-threaded perls (so threaded perls now finally
320 reach about half the speed of non-threaded perls).
321 - slightly modernise Coro::Intro, add section about rouse functions.
322 - avoid DEFSV and ERRSV, giving another 10% improvement
323 in thread switching.
324 - Coro::State->is_destroyed is now called is_zombie.
325 - implement a Coro->safe_cancel method that might fail, but
326 cancels in a "safer" way if it succeeds.
327 - add preliminary support for DEBUGGING perls.
328 - get rid of two hash-accesses when initialising a new Coro - this
329 speeds up coro creation by almost a factor of two.
330 - croak when a coro that is being cancelled tries to block
331 (e.g. while executing a guard block), instead of crashing or
332 deadlocking.
333 - use a more robust and also faster method to identify Coro::State
334 objects - speeds up everything a bit.
335 - implement Coro->cancel in XS for a 20% speed improvement, and to
336 be able to implement mutual cancellation.
337 - speed up context switches by a percent or two by more efficiently
338 allocating context stack entries.
339 - implement Coro->join and Coro->on_destroy in XS for a speedup and
340 a reduction in memory use.
341 - cancelling a coro while it itself is cancelling another coro is
342 now supported and working, instead of triggering an assertion.
343 - be a bit more crash-resistant when calling (buggy) on_destroy
344 callbacks (best effort).
345 - move on_destroy into the slf_frame, to allow extension slf
346 functions to have destructors.
347 - get rid if coro refcounting - simply crash in other interpreter
348 threads by nulling the pointers on clone.
349 - simplify warn/die hook handling when loading Coro - the convoluted
350 logic seems to be no longer neccessary.
351 - use libecb instead of our own home-grown gcc hacks.
352 - document alternatives to Coro::LWP. Please use them :)
353 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
354 sigaltstack. Really. wine suffers from it, erlang suffers from it,
355 and it's known since at least 2006.
356
3575.372 Wed Feb 23 06:14:30 CET 2011
358 - apparently mingw doesn't provide a working gettimeofday, try to
359 work around that by relying on Time::HiRes (indirectly brought to
360 my attention by Max Maischein).
361 - fix some portability issues when Time::HiRes was used.
362
3635.371 Mon Feb 21 14:36:08 CET 2011
364 - backport to windows process emulation code again.
365
3665.37 Sat Feb 19 07:49:44 CET 2011
367 - add a big "Coro thread life cycle" section to "man Coro".
368 - try a tentative workaround against the breakage that 5.13 has
369 introduced without depreciation period. sigh.
370 - no longer use Time::HiRes if gettimeofday is available, which
371 saves quite a lot of memory.
372
3735.36 Sun Feb 13 05:33:41 CET 2011
374 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
375 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
376 method call.
377 - undocument Coro::Timer::sleep and obsolete whole module.
378 - optimise Coro::Timer::timeout memory and cpu usage.
379 - slightly updated Coro::Intro for recent changes.
380 - do not initialise PL_dirty anymore.
381
3825.25 Thu Nov 11 01:08:39 CET 2010
383 - try a different approach on netbsd - netbsd 5 finally has marginally
384 working pthreads, but still broken ucontext/sigaltstack.
385 - openbsd 4.8 finally got their act together, Coro works out of the box
386 with asm, setjmp and pthreads (no change, just informational).
387
3885.24 Sat Oct 23 11:27:12 CEST 2010
389 - port to the EV 4.0 API.
390 - work around bugs in mingw32, making strawberry perl work
391 out of the box.
392 - correctly modify Coro::AIO function prototypes
393 so that they reflect the "no optional parameters" rule.
394 - "ported" libcoro to C++.
395
3965.23 Mon May 17 18:50:42 CEST 2010
397 - be more resistant to ordering changes when initialising
398 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
399 Waldorf).
400 - document that perl 5.12 deliberately removed support for cloning.
401
4025.22 Wed Apr 14 03:55:35 CEST 2010
403 - correctly return udnef on errors in Coro::Handle::read/write
404 (testcase by Marc Mims).
405 - convert Coro::Util into a "perl compatibility wrapper" - the functions
406 are less useful now, but are drop-in replacements for existing
407 functions, listing better alternatives in the documentation. This also
408 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
409 with Coro::Util::inet_aton.
410 - do not override $Coro::idle unconditionally in Coro.pm, as other
411 modules could have provided their own idle coro already
412 (for exmaple, Coro::AnyEvent).
413 - fix Coro::Util::gethost* functions.
414 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
415
4165.21 Wed Dec 16 07:19:51 CET 2009
417 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
418 together.
419 - add some examples on how to combine other event loops with Coro in
420 Coro::AnyEvent, and how to run it (and not to block). Seems to be
421 the most common source of confusion.
422 - try to catch people naively blocking in an event callback.
423 - work around the perl filehandle bug issue in conjunction with
424 older common::sense (as indirectly pointed out by ZSystem).
425 - clarify the "not from signal handlers" section.
426
4275.2 Sun Oct 4 14:54:24 CEST 2009
428 - Coro::Storable destroyed the prototypes of the functions it wrapped.
429 - export rouse_cb and rouse_wait by default now.
430 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
431 - new method $state->swap_sv.
432 - added section on "windows process emulation" to the manpage,
433 after a not-so-fruitful (nor-friendly) "discussion" with chip
434 salzenberg (discussion implies arguments, but his only arguments
435 were ad-hominems, one wonders why he started it in the first
436 place). I hope this explains it well enough for him to understand,
437 and maybe well enough for others to understand.
438 - use common::sense everywhere now.
439 - idle callbacks are no longer supported, use idle coros instead.
440 - print a thread listing when a deadlock is detected.
441
4425.17 Sat Aug 22 23:09:31 CEST 2009
443 - work around a bug in the perl debugger causing crashes
444 when running under the debugger by marking _pool_handler as nodebug.
445 - speed up Coro::async considerably.
446 - try some hacks to get netbsd to work "more often" - their broken
447 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
448 - convert Coro to AE 5.0 API.
449
4505.162 Tue Jul 28 04:04:03 CEST 2009
451 - perl 5.8.2 is now minimum requirement.
452 - skip t/19_handle.t on broken windows perls.
453
4545.161 Wed Jul 22 04:47:38 CEST 2009
455 - Coro::AnyEvent::poll could have a different prototype when EV was
456 used as backend (analyzed by Tatsuhiko Miyagawa).
457 - Coro::AnyEvent errornously initialised the event loop when loaded,
458 not on demand.
459 - try to workaround rare */t/01_unblock.t failures.
460
4615.16 Tue Jul 21 01:44:37 CEST 2009
462 - Coro::AnyEvent failed to hook into the event loop
463 when no threads had been readied between detecting
464 the event loop and actually running it.
465 - considerably speed up Coro::Select by taking avdantage
466 of AnyEvent > 4.8 and some other optimisations.
467 - implement paragraph readline mode in Coro::Handle
468 (based on patches by Zsbán Ambrus).
469 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
470 and analyzed by Yasuhiro MATSUMOTO).
471 - clarified libcoro license and copyright.
472 - someone stole my EXTRA_META!!!
473 - implement Coro::Select::patch_pp_sselect and it's brother,
474 for hardcode select overriding.
475
4765.151 Mon Jul 6 05:41:57 CEST 2009
477 - backport to windows process emulation code again (patch by
478 Yasuhiro MATSUMOTO).
479 - slightly update Coro::MakeMaker.
480
4815.15 Tue Jun 30 10:28:06 CEST 2009
482 - deprecate Coro::Socket, document how to get ipv6 support via
483 AnyEvent::Socket instead.
484 - implement signal->wait ($cb) interface, similar to semaphores.
485 - work around SvOK not supporting getmagic, so we have to getmagic
486 to test for undef :/ (reported by Matthias Waldorf).
487 - load Coro::AnyEvent in all modules using AnyEvent.
488 - work around perl corrupting our internal data structures,
489 reported by Tokuhiro Matsuno.
490 - enable per-coroutine real and cpu time gathering
491 (Coro::State::enable_times).
492
4935.14 Wed Jun 24 01:37:48 CEST 2009
494 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
495 as this seems to have been a difficult concept (poll, sleep, idle,
496 idle_upto).
497 - add Coro::AnyEvent::readable/writable functions.
498 - clarify Coro::EV/Event/AnyEvent manpages.
499 - free per-thread global scalars in the thread calling ->cancel, to
500 avoid crashes when $_, $@ etc., are magical but some of those
501 had already been freed.
502 - "unexperimentalise" the callback interface for Coro::Semaphore.
503 - speed up ready queue management/context switching by using a linked
504 list instead of an array (~5-10%).
505 - implement "watch" command in Coro::Debug shells.
506 - for fun, implement time-slicing as an example in the manpage.
507 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
508 Coro::Event, use the normal AnyEvent handling instead of dieing
509 (the same is true for Coro::Handle).
510 - properly document Coro::EV::timed_io_once.
511 - avoid unneccessary ->cancel calls in Coro::Handle.
512 - maybe make it work on mingw32 with win32 backend
513 (based on patch by Yasuhiro Matsumoto).
514
5155.132 Fri May 29 09:00:39 CEST 2009
516 - do not keep a reference to the argument itself in
517 Coro::Semaphore::guard, as it could change later.
518 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
519 should support a prepare callback.
520
5215.131 Mon Mar 16 23:20:37 CET 2009
522 - implement and document Coro->suspend, Coro->resume.
523 - fix Coro::Select implementation to not (often) close
524 the passed file descriptors (testcase provided by pippijn).
525
5265.13 Mon Dec 15 21:51:42 CET 2008
527 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
528 - don't set diehook to C<undef> but instead to NULL, to avoid
529 spurious warnings.
530 - fix a lot of bugs in Coro::SemaphoreSet.
531 - Coro::SemaphoreSet will less often create a semaphore needlessly.
532 - add Coro::SemaphoreSet::count and wait methods.
533 - take advantage of the new Guard module.
534 - deprecate Coro::guard.
535 - try to fix the dreaded 01_unblock tests once more. I hate it when
536 testsuites need more fixing than the code they are supposed to test.
537 - croak in more cases when a required callback isn't resolvable.
538 - fix some minor issues in Coro::State->call/eval.
539 - use current coroutine context instead of a temporary one
540 when temporarily switching to another coroutine.
541 - do not call C-level on_destroy handlers during global destruction,
542 to avoid needless segfaults.
543
5445.12 Sun Dec 7 13:30:38 CET 2008
545 - add default config for MirOS, which seems to be a bug-to-bug
546 compatible fork of openbsd ("world domination by releasing
547 openbsd cvs before the openbsd folks do it" or so :).
548 - free_padlist did destroy the names pad, not good, but didn't
549 seem to bother perl - this could fix issues such as eval ""
550 inside a function called from multiple coroutines.
551 - use a different method to detect destruction time.
552 - be more careful when freeing padlists just before global
553 destruction.
554 - fixed and expanded the call/cc example.
555 - renamed _terminate to _coro_run.
556 - new method Coro::Channel->shutdown.
557 - try pthreads on openbsd <4.4 (broken sigaltstack, will
558 pthreads fare better?).
559 - be less picky about destroying "a" running coroutine.
560
5615.11 Tue Nov 25 21:49:05 CET 2008
562 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
563 working code with failed assertions, introducing more bugs than
564 it fixes, requiring elaborate workarounds :(
565
5665.1 Mon Nov 24 08:54:59 CET 2008
567 - wrote a small introductory tutorial - Coro::Intro.
568 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
569 - Coro::Select did errornously dup the file descriptors
570 and didn't work with all AnyEvent backends.
571 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
572 LWP data transfers.
573 - disassociate c contexts from coro objects - this is agruably more
574 correct, but mostly allows sharing of cctxs between coro and state
575 objects, for added memory savings and speed increases.
576 - bumped $Coro::POOL_RSS up to 32kb by default.
577 - no longer set the optype to OP_CUSTOM, as B::* understandably
578 doesn't like this very much (and we *are* a type of entersub).
579 - implement state cloning, just to prove that call/cc can be done.
580 - automatically load Coro::AnyEvent in Coro::Handle.
581 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
582 do not have this method (and don't need it either).
583 - speed up generic anyevent methods in Coro::Handle by using rouse
584 callbacks.
585 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
586 others. It also makes the debugger happier, as you can trace
587 through the idle threads now.
588 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
589 - eg/event was pretty broken.
590 - better 5.8.6 compatibility.
591
5925.0 Thu Nov 20 10:35:05 CET 2008
593 - NEW ARCHITECTURE: use the latest 4.x version if you experience
594 stability issues.
595 - bump API version to 7 - all dependents must be recompiled.
596 - removed timed_* functions - they were not being used anyways
597 and should be replaced by a more generic mechanism -
598 and were annoying to support anyways :)
599 - removed SemaphoreSet's waiter method - use sem method instead.
600 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
601 - async_pool did free a scalar value twice
602 ("Attempt to unreference...").
603 - fix a longstanding bug where calling terminate on a coro that
604 was waiting for a semaphore that was just becoming available
605 would cause a deadlock (semaphore would get into a state where
606 it was available but waiters were still blocked).
607 - calling throw on a coroutine that is waiting for a semaphore will
608 no longer make it acquire the semaphore (and thus leak a count).
609 - perl's process emulation is now not even theoretically supported
610 anymore.
611 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
612 conversion of callback-style to blocking-style.
613 - new methods $coro->schedule_to and ->cede_to, to specifically
614 schedule or cede to a specific coroutine.
615 - new function Coro::Semaphore::wait.
616 - use named constants in Coro::Channel (Richard Hundt).
617 - directly patch the entersub opcode calling SLF functions (cede,
618 transfer and so on). this does speed up context switching, but
619 more importanly, it frees us from the hardcoded behaviour of
620 entersub, so we might actually be able to return something from
621 those functions and atcually create new ones.
622 - take advantage of __builtin_frame_address on gcc.
623 - expose THX in coroapi (not sure whether this was a wise decision,
624 as "threaded" perls are running at half speed anyways).
625 - implement execute_slf (schedule-like-function) interface that makes
626 it possible to implement schedule-like-functions in XS.
627 - use new SLF interface to massively speed up Coro::EV by roughly a
628 factor of two.
629 - used new SLF interface to massively speed up Coro::Semaphore by a
630 factor of three.
631 - used new SLF interface to speed up Coro::AIO by roughly a factor of
632 four and reduce its memory usage considerably.
633 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
634 for a nice speedup and vastly more correct behaviour. Also implement
635 a new method "sem" to get at the underlying semaphore object.
636 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
637 (in comparison) 20-40% speedup.
638 - used new SLF interface to reimplement Coro::Signal gaining
639 some unknown (because I was too lazy), but certain, speedup, and also
640 making signals reliable for the first time.
641 - used new SLF interface and other optimisations to speed up async_pool
642 by a factor of two. It also doesn't rely on perl's exception mechanism
643 to exit anymore. The overhead for terminating an async_pool, coro over
644 a normal async is now very small.
645 - sped up coroutine creation/destruction by 40%.
646 - forgot to include Coro/libcoro/README in the dist for all these years.
647 - work around a freebsd pthreads bug (manual testcancel is required as
648 pthread_cond_wait isn't a cancellation point on freebsd).
649 - use new rouse functions to speed up and simplify Coro::BDB.
650 - make "prefer perl native functions" work with threaded perls.
651 - condense Coro::Debug ps output, hint at v and w flags.
652 - (libcoro) lots of minor cleanups and portability improvements.
653
6544.914 Wed Nov 19 12:54:18 CET 2008
655 - fix a disastrous bug in the readline optimisation
656 introduced in 4.801.
657
6584.913 Sat Nov 15 07:58:28 CET 2008
659 - async_pool did free a scalar value twice
660 ("Attempt to unreference...").
661
6624.912 Thu Nov 13 18:31:23 CET 2008
663 - minor cleanups.
664 - use much larger stacks on linux and perl < 5.8.8.
665 - Coro::Debug::new_unix_server did not unlink the socket
666 when destroyed.
667
6684.911 Tue Nov 11 04:26:17 CET 2008
669 - "port" to threaded perls.
670
6714.91 Mon Nov 10 05:36:38 CET 2008
672 - the ->throw exception object no longer leaks.
673 - creating a new cctx leaked a scopestack entry (memleak).
674 - new coroutines didn't get created with a zero flags field
675 (unknown impact).
676 - calling ->throw on a not-yet-started coroutine should now work
677 instead of being ignored.
678 - ->throw is now supported on Coro::State objects.
679 - clean up cctx creation code a bit.
680 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
681
6824.9 Sat Nov 8 17:45:27 CET 2008
683 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
684 - (libcoro) no longer leak threads in the experimental pthread backend,
685 also speed it up considerably.
686 - (libcoro) do not rely on makecontext passing void *'s unscathed.
687 - fix compiletime dependencies on libcoro in the Makefile.
688 - cctx_count wasn't always updated properly.
689 - Coro::State::cctx_stacksize wasn't applied correctly.
690 - new function Coro::State::cctx_max_idle.
691 - the default max number of idle C contexts is now 4.
692 - (libcoro) try harder to get _setjmp/_longjmp.
693 - (libcoro) cleanup and extend the libcoro API to officially
694 allow the creation of empty source contexts.
695 - very experimental workaround for the totally broken netbsd platform.
696 - tried to hack around openbsd bugs.
697
6984.804 Wed Nov 5 16:36:00 CET 2008
699 - Coro::Debug::new_unix_server would not create a non-blocking listening
700 socket, sometimes causing freezes.
701 - (libcoro) fix misaligned stack points for setjmp and assembly
702 methods, which can cause crashes on x86/x86_64 with a sufficiently
703 aggressive compiler.
704 - new function: Coro::Debug::new_tcp_server.
705 - move ->throw into the Coro class because it only works on coro objects.
706
7074.803 Mon Nov 3 17:16:12 CET 2008
708 - (libcoro) use a global asm statement to become independent of gcc
709 otpimisations for CORO_ASM (thanks to pippijn for the idea).
710 - try to workaround yet another broken bsd, this time dragonfly.
711
7124.802 Thu Oct 30 10:56:12 CET 2008
713 - support -fno-omit-frame-pointer on x86 with the assembly method.
714 - tune 01_unblock.t tests a bit.
715
7164.801 Wed Oct 22 18:33:37 CEST 2008
717 - improve readline speed for very long "lines".
718 - backport to 5.8.8.
719
7204.8 Thu Oct 2 13:34:40 CEST 2008
721 - new function Coro::AIO::aio_wait.
722 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
723 - greatly speed up and reduce memory usage of Coro::AIO requests.
724 - implement some other µ-optimisations.
725
7264.749 Mon Sep 29 14:40:12 CEST 2008
727 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
728
7294.748 Sat Sep 27 14:03:19 CEST 2008
730 - implement, but do not document, PerlIO::cede(granularity).
731 - Coro::Storable forgot to wrap Storable::pstore.
732 - work around the multitude of leaks and memory corruption
733 bugs in PerlIO::via by using our own C-level perliol. As a side
734 effect, Coro::Storable is now much, much, much faster.
735
7364.747 Tue Sep 23 01:59:41 CEST 2008
737 - fix a per-cv memleak (one empty array was leaked per
738 code reference).
739 - avoid a crash in coro->call|rss when the coroutine was already
740 destroyed (most noticably when using Coro::Debug::ps :)
741 - also protect *Storable::FILE.
742 - push up default storable granularity to 20ms.
743
7444.746 Sun Sep 21 03:22:20 CEST 2008
745 - be more insistent on locking Storable against reentrancy
746 in Coro::Storable.
747 - move swap_def?v and throw to Coro::State, as documented.
748
7494.745 Thu Jul 24 00:14:38 CEST 2008
750 - remove debugging code related to MgPV_nolen_const, also try to
751 make it compile with perl 5.8.6 (yes, apple apparently loves
752 outdated software). Reported by John S.
753
7544.744 Tue Jul 8 22:06:35 CEST 2008
755 - correctly provide default DIE/WARN handlers as documented.
756 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
757 implementations inside perl work fine for us.
758
7594.743 Mon Jun 16 00:21:57 CEST 2008
760 - when using Coro::EV without running EV::loop it could
761 result in busy-waiting for events, this has been fixed.
762 - reduce codesize and improve performance by using EV_DEFAULT_UC.
763
7644.742 Sat May 31 14:10:21 CEST 2008
765 - implement a workaround for (some) perl <5.8.8 versions.
766 - require EV 3.3+.
767
7684.741 Fri May 30 23:33:09 CEST 2008
769 - tell netbsd how utterly broken their imitation of an OS is
770 and refuse to build by default if pthreads are in use.
771 - switch to "s" method on all bsd's by default, as their ucontext
772 stuff seems just too broken.
773 - fix a bug in Coro::Select.
774
7754.74 Thu May 29 20:05:31 CEST 2008
776 - do not test Coro::LWP for lack of dependencies.
777 - do not test Coro::Select for lack of working perls.
778
7794.73 Thu May 29 2008
780 - fix a bug in Coro::EV which would cause it to block despite
781 there being runnable coroutines.
782 - sprinkle "no warnings" freely over everything, also suppress
783 warnings for some other modules.
784 - fix typo in WSAEWOULDBLOCK.
785
7864.72 Sun May 25 05:14:36 CEST 2008
787 - tweak META.yaml a bit, unfortunately, there is no documented way
788 to have optional dependencies with CPAN. doh :(
789 - avoid running some tests on windows because they would fail due to
790 perl bug (broken fork, broken pipes...).
791 - work around perl on windows bugs where perl returns undocumented
792 error codes for sysread, syswrite etc. by taking advantage
793 of AnyEvent's workaround for that problem.
794 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
795 a common perl implementation bug on windows.
796 - use unix domain sockets in testsuite to work around
797 common perl implementation bugs on widows (they are emulated by
798 tcp sockets on windows. ugh.)
799
8004.71 Sat May 24 20:01:27 CEST 2008
801 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
802 - take advantage of async name resolution of AnyEvent::Util.
803 - work around brutal inet_aton override in Coro::LWP.
804 - take advantage of the readyhook in Coro::EV, for smoother
805 scheduling.
806
8074.7 Sun May 11 00:32:19 CEST 2008
808 - completely reworked the Coro manpage.
809 - added Coro::AnyEvent, generic event loop integration.
810 - implement cancel, ready and kill commands in Coro::Debug.
811 - document find_coro in Coro::Debug.
812 - incompatible change: rename has_stack to has_cctx.
813 - Coro::AIO and Coro::BDB no longer force event model detection,
814 use AnyEvent::AIO and AnyEvent::BDB.
815
8164.6 Sat Apr 26 10:05:14 CEST 2008
817 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
818 by the Coro module. It was a mistake to have it in the first place.
819 - (experimental) support for activestate perl 5.10 (method "w").
820 - (experimental) support for strawberry perl (method "a").
821 - coro_sigelem_set did not return a value although it had to,
822 actual impact unknown.
823
8244.51 Mon Apr 14 13:28:27 CEST 2008
825 - make it compile again on 5.8.
826
8274.50 Thu Apr 10 09:43:17 CEST 2008
828 - I did it twice! (see 4.49).
829
8304.49 Sun Apr 6 21:23:31 CEST 2008
831 - grr, instead of compiling the recent changes on 5.10 only they
832 were compiled on 5.8 only.
833
8344.48 Sun Apr 6 20:36:46 CEST 2008
835 - allow coroutine switches during eval's under 5.10.x, as it seems
836 the parser is a per-interpreter option now, so this is safe
837 (this might fix the odd crashing bug).
838 - drop support for 5.9.x versions: they are dead, jim.
839
8404.47 Sun Apr 6 00:37:52 CEST 2008
841 - force cctx allocation on API calls: we know we need to force one
842 and gcc actually manages to confuse our heuristic nowadays,
843 leading to crashes and worse.
844 - document force_cctx.
845
8464.46 Fri Apr 4 22:05:43 CEST 2008
847 - upgrade libcoro, resulting in pthread-backend (which was only created
848 to fulfill the rules of the programming languages shootout).
849
8504.45 Thu Mar 13 11:55:36 CET 2008
851 - fix a file leaking bug in eg/mhyttpd that would allow
852 downloading of any file (reported by oesi).
853 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
854 (also add testcase).
855 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
856 apparently having loads of fun with that).
857 - implement $state->swap_defsv, ->swap_defav and document ->throw.
858
8594.4 Wed Feb 13 16:44:29 CET 2008
860 - only assume .cfi pseudo isns exist on GNU/Linux.
861 - add get_timed method to Coro::Channels.
862 - fixed Coro::Debug manpage.
863 - perl 5.11 compatbility improvement based on patch by
864 Andreas König.
865
8664.37 Sun Jan 20 11:25:23 CET 2008
867 - soften the check for an active parser for perl 5.10 (avoiding
868 "Coro::State::transfer called while parsing" in many cases).
869
8704.36 Sun Jan 13 10:53:56 CET 2008
871 - reset diehook when terminating from an async_pool as to not
872 trigger any __DIE__ handlers.
873
8744.35 Sun Jan 13 04:14:13 CET 2008
875 - "bt" debug command now displays any exceptions
876 from longmess and also skips the innermost
877 stackframes, giving more useufl output.
878 - allow backtraces in coroutines blocked in Coro::EV,
879 at a <1% speed hit.
880 - handle localising of $SIG{__WARN__} and __DIE__
881 properly (with a proper amount of dirty hacking).
882
8834.34 Sat Dec 22 17:49:53 CET 2007
884 - upgrade to EV version 2.0 API.
885
8864.33 Mon Dec 17 08:36:12 CET 2007
887 - make Coro::AIO etc. loadable in the absence of EV.
888
8894.32 Mon Dec 17 07:46:02 CET 2007
890 - majorly improved Coro::Handle's performance with Coro::EV.
891 - implemented the until now mythical Coro::BDB module.
892 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
893
8944.31 Wed Dec 5 12:32:39 CET 2007
895 - remove warn statement form Coro::Util.
896
8974.3 Tue Dec 4 20:33:14 CET 2007
898 - calls to the idle function could cause stack corruption
899 when the stack changed.
900 - do no longer rely on the presence of EV::DNS if EV is used
901 (because it is gone), but instead take avdantage of EV::ADNS
902 if available.
903 - add ($) prototypes to all functions in Coro::Storable.
904 - use a conventional (safer) idle callback in Coro::EV.
905 - do accept quickly in Coro::Debug to avoid endless loops.
906
9074.22 Fri Nov 30 16:04:04 CET 2007
908 - really use optimised versions for Event and EV in Coro::Util
909 and Coro::Handle.
910
9114.21 Sun Nov 25 10:48:59 CET 2007
912 - fix a spurious memory read.
913 - Coro::EV no longer keeps the eventloop "alive".
914
9154.2 Fri Nov 9 20:47:05 CET 2007
916 - enable/disable tracing from a new coroutine, not a pooled one.
917 - fix a memleak in Coro::Event.
918 - removed killall call from fork_eval.
919 - made sure store_fd is already loaded so that fork_eval
920 does not have to parse autoload in each subprocess.
921 - only use assembly method if -O switch is in $Config{optimize}.
922 - add (optional) Coro::EV module, so far the best event loop module
923 directly supported by Coro.
924 - if the event model is EV, use EV::DNS to resolve
925 stuff in Coro::Util.
926 - don't get confused by multiple event notifications in Coro::Handle.
927 - initial support for EV (libevent interface).
928 - require Event and EV using configure_requires, to force their existance.
929
9304.13 Wed Oct 24 07:26:45 CEST 2007
931 - add Coro::Storable::blocking_thaw.
932 - use a vastly more complicated technique to localise
933 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
934 - use a coroutine for the idle callback Coro::Event,
935 instead of running Event in the current coroutine context.
936 This also catches recursive invocations.
937 - actually report fork errors in gethostbyname and inet_aton.
938
9394.11 Thu Oct 11 02:40:24 CEST 2007
940 - port to threaded perls.
941
9424.1 Thu Oct 11 02:38:16 CEST 2007
943 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
944 be local to each coro (see Coro::State).
945 - incompatible change: for very deep reasons, cede and cede_notself
946 cannot return anything, so nothing will be returned.
947 - possibly bring back 5.10 compatibility (untested).
948 - work around stupid (and wrong) warning on 5.10 :(.
949 - overlay the saved state over the context stack. This saves
950 a few hundred bytes per coroutine on average and also
951 speeds up context switching a bit.
952 - further tune default stack sizes.
953 - (more) correctly calculate stack usage in coro_rss.
954 - Coro::Storable::blocking_* did not properly lock
955 resulting in races between coroutines.
956 - added Coro::Storable::guard.
957 - stopping to trace a coroutine could destroy the cctx of
958 an unrelated coroutine.
959 - explain the relationship between Perl and C coroutines in
960 more detail in Coro::State.
961 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
962 causing a fork per resolve. This also affected Coro::Socket.
963 - switch to a separate stack in $coro->call/eval to avoid
964 invalidating pointers.
965
9664.03 Sat Oct 6 21:24:00 CEST 2007
967 - added Coro::throw method.
968 - minor code cleanups.
969
9704.02 Sat Oct 6 02:36:47 CEST 2007
971 - fix a very minor per-coroutine memleak (a single codereference).
972 - fixed a bug where the currently in-use c context would be freed
973 prematurely (can happen only when programs change the stacksize
974 or use tracing).
975 - tracing can no longer keep a coro alive after it terminated.
976 - do static branch prediction in the common path for gcc. gives
977 about 2-5% speed improvement here.
978
9794.01 Fri Oct 5 22:10:49 CEST 2007
980 - instead of recreating *a* standard output handle we simply
981 use STDOUT, which is faster and hopefully more robust.
982
9834.0 Fri Oct 5 12:56:00 CEST 2007
984 - incompatibly changed Coro::Storable::freeze.
985 - major new feature: added Coro::Debug, for interactive coroutine
986 debugging, tracing and much more.
987 - major bug fix: unbelievable, but true: $_, $/ and many other
988 "saved" variables actually weren't being saved. This has been fixed,
989 of course, while increasing performance while losing all the save
990 flags.
991 - save flags are gone, and all the api functions dealing with them.
992 - added Coro::Semaphore::adjust.
993 - added Coro::Util::fork_eval.
994 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
995 - added Coro::killall.
996 - reduce initial stack sizes to allow for "micro-coroutines".
997 - better async_pool resource management, moved parts of async_pool
998 handling to XS (major speed improvement).
999 - actually croak before modifying important data structures.
1000 - refuse to transfer while compiling.
1001 - possibly support eval EXPR better now.
1002 - enable assembly per default on linux+bsd x86+amd64.
1003 - all internal members were renamed _something for easier subclassing.
1004 - many minor tweaks.
1005
10063.63 Wed May 16 14:10:06 CEST 2007
1007 - implement handcoded assembly for x86/amd64 SVR ABI.
1008
10093.62 Fri Apr 27 21:36:06 CEST 2007
1010 - upgrade libcoro (which might set unwind info correctly).
1011 - change default on linux to setjmp/longjmp.
1012
10133.61 Thu Apr 19 12:36:18 CEST 2007
1014 - Coro::Storable caused an endless loop when thawing invalid
1015 pst's sometimes.
1016 - use a Semaphore in Coro::Storable, as Storable doesn't
1017 seem to be reentrant (although it is documented to
1018 be threadsafe...).
1019 - fix Coro::Signal to bring back the original unreliable
1020 but stateful semantics.
1021 - fixed a lot of typos in Coro.pm (patch submitted by David
1022 Steinbrunner, which applied flawlessly).
1023
10243.6 Sat Apr 14 17:13:31 CEST 2007
1025 - added some bugfixes to get eg/myhttpd working again.
1026 - added Coro::Storable for often-cede'ing freeze/thaw.
1027 - try to do a clean exit when a coroutine calls exit
1028 (EXPERIMENTAL).
1029 - got rid of indirect call through _coro_init.
1030 - updated the partly antique examples in eg/ to
1031 work again and be a bit less magic, too.
1032 - fixed Coro::Signal semantics to work as documented again.
1033
10343.55 Sun Mar 25 01:20:47 CET 2007
1035 - add SAVE_DEFFH to save the default filehandle and enable
1036 it by default.
1037 - finally move socket-operations from Coro::Socket to Coro::Handle
1038 to be able to unblock foreign sockets.
1039 - add Coro::State::save_also and guarded_save.
1040 - add count accessor to Coro::Semaphore.
1041 - add Coro::State::cctx_stacksize.
1042 - just for the fun of it, do not rely on implicit context,
1043 which can dramatically improve performance, but people
1044 using windows-process-emulation perls do not care much about
1045 performance.
1046
10473.51 Sun Mar 4 14:18:04 CET 2007
1048 - fixed a problem when you weakened references to Coro::State's
1049 (patch by Yuval Kogman).
1050
10513.501 Wed Feb 28 12:44:07 CET 2007
1052 - rename some global symbols as macosx from hell redefines
1053 them without asking.
1054
10553.5 Tue Feb 13 20:22:53 CET 2007
1056 - do AnyEvent model detection earlier, avoiding problems
1057 caused by first using AnyEvent and later Coro::Event.
1058 - implement and document Coro::Event event objects.
1059 - fix a potential problem in Coro::Event causing crashes.
1060 - initialise PL_comppad when creating a new coroutine,
1061 avoids crashes on early coro destruction.
1062
10633.41 Mon Jan 22 19:19:49 CET 2007
1064 - readline on Coro::Handle did not support undefined $/,
1065 nor did it deliver partial lines on EOF or error.
1066 - implement malloc fallback for stack allocation because
1067 stupid broken idiotic OSX has a stupid broken
1068 idiotic fits-the-whole-os mmap "implementation" and
1069 my dick feels longer if Coro is portable even to
1070 obsolete platforms.
1071
10723.4 Fri Jan 19 21:52:54 CET 2007
1073 - remove t/09_timer.t, as it isn't really testing much
1074 but was rather flaky in practise.
1075 - async_pool coro would keep arguments and callback alive until
1076 it was reused.
1077 - cancellation of a coroutine could cause spurious idle calls
1078 in cede_notself.
1079
10803.3 Sat Jan 6 03:45:00 CET 2007
1081 - implement $coro->on_destroy.
1082 - Coro::Event blocking semantics have been changed,
1083 documented and - hopefully - improved.
1084 - fix nice adding, not subtracting, from priority.
1085 - fix ->prio and api_is_ready (patch by Mark Hinds).
1086 - fixed an assert ("... == prev__cctx->idle_te")
1087 that could errronously trigger.
1088 - fix various large and small memleaks.
1089 - use a (hopefully) more stable cancel implementation
1090 that immediately frees the coroutine data.
1091 - cede/cede_notself return a status now.
1092 - added Coro::guard function.
1093 - added a global coroutine pool for jobs (on my machine,
1094 I can create and execute 48k simple coros/s with async,
1095 and 128k coros with async_pool).
1096 - Coro::AIO now uses the coroutine priority as io priority.
1097
10983.2 Fri Dec 22 05:07:09 CET 2006
1099 - improve portability to slightly older perls.
1100 - use cleaner coroutine destruction.
1101 - simplify configuration for users.
1102 - optionally (unrecommended) prefer perl functions over
1103 their coro replacements.
1104
11053.11 Tue Dec 5 13:11:24 CET 2006
1106 - fixed some bogus assert's, but as perl.h disables assert even
1107 without NDEBUG (thank you very much), not too many people should
1108 notice (that did include myself). Andreas König noticed, though :)
1109 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
1110 it. Also noticed by Andreas König :)
1111 - save/restore tainted status.
1112 - verified to pass the testsuite on my 5.9.5.
1113
11143.1 Mon Dec 4 23:03:40 CET 2006
1115 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
1116 - incompatible change: transfer flags are now per-state.
1117 - give a better error message on deadlock.
1118 - document Coro::nready.
1119 - enhanced testsuite.
1120
11213.01 Sun Dec 3 23:47:42 CET 2006
1122 - forgot to include Coro::Timer.
1123
11243.0 Sun Dec 3 22:57:25 CET 2006
1125 - the "FINALLY COMPLETELY STABLE" release (coming soon:
1126 the "FAMOUS LAST WORDS" release).
1127 - implement a new stack sharing algorithm, which uses a stack
1128 pool (size currently hardcoded).
1129 - make stack sharing mandatory (it no longer uses a heuristic).
1130 - eval/die no longer cause weird problems under heavy use.
1131 - Coro::Event could cause livelocks if it was used but
1132 no Coro::Event watchers were used.
1133 - Coro::Event now uses asynccheck as prepare does not
1134 check for changed watchers.
1135 - Coro::Event allows multiple waiting coros and will wake up one
1136 per event.
1137 - Coro::Event should be cleaner and more efficient now.
1138 - new utility function Coro::unblock_sub.
1139 - document the sad fact that Event is no longer reentrant.
1140 - putting a coroutine into the ready queue twice could under
1141 some circumstances lead to stack corruption.
1142 - minor incompatibility: subclassing Coro::State is supported
1143 directly now without going through a _coro_state member.
1144 - state/coro switching is much faster now.
1145 - very minor optimisations and code/documentation cleanup.
1146 - avoid problems due to compiler inlining.
1147 - removed timers from Coro::Timer -> use AnyEvent instead.
1148 - replaced Coro::idle coroutine by (cleaner) idle handler.
1149 - updated to newest libcoro.
1150 - implement enhanced support for valgrind.
1151 - implement is_ready and return value for ready.
1152 - removed Coro::Cont, it was a misnomer (it's generators),
1153 and it was somewhat annoying to get it right. Will come back
1154 if somebody asks for it :)
1155 - many ->wait methods and Coro::Event could return
1156 spuriously without the event having happened.
1157
11582.5 Tue Nov 7 12:22:33 CET 2006
1159 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
1160 use AnyEvent, moved them to Coro/.
1161 - added Coro::LWP which contains all the uglyness required to
1162 make LWP non-blocking.
1163 - should work with perl 5.9.x now (Andreas König made me do it).
1164 - fixed another bug in Coro::Select when the timeout was undef.
1165 - reuse PL_start_env for all coros, saves some memory per coroutine.
1166 - manage PL_top_env differently, hopefully to avoid panic: top_env.
1167 - timeout argument was not properly used in Coro::Socket.
1168 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
1169 - emulate undocumented(!) functionality of IO::Socket required
1170 by LWP(!!).
1171 - updated eg/lwp to work with newer lwp's.
1172 - remove "FATAL: uncaught exception" prefix. Coroutines that die
1173 kill the whole process, just as exceptions in the main "coroutine"
1174 did already.
1175
11762.1 Wed Nov 1 23:01:13 CET 2006
1177 - fix a long-standing bug in Coro::Select where select with
1178 zero timeout would instead change the current default filehandle.
1179 - use a simpler and hopefully more robust way to clone padlists
1180 (uses less memory and a perl function instead of our own).
1181 - coro can now create a stack guard area on many architectures.
1182 - Coro::AIO properly reexports additional functions from IO::AIO.
1183 - updated libcoro with a workaround for OS X,
1184 pach and testing by Michael Schwern.
1185
11862.0 Tue Oct 24 05:47:17 CEST 2006
1187 - support additional aio requests in Coro::AIO.
1188
11891.9 (never properly released due to a glitch)
1190
11911.8 Thu Feb 2 00:59:06 CET 2006
1192 - applied suggested patch by SAMV to avoid problems during stupid
1193 mark & sweep gc run.
1194 - applied patch by Scott Walters for 5.9.3 compatibility.
1195
11961.7 Tue Dec 27 01:41:58 CET 2005
1197 - added Coro::AIO, a thin wrapper around IO::AIO.
3 - improved Makefile.PL explanations. 1198 - improved Makefile.PL explanations.
4 1199
51.6 Mon Dec 12 21:30:05 CET 2005 12001.6 Mon Dec 12 21:30:05 CET 2005
6 - additionally save PM_curpm might fix as-of-yet 1201 - additionally save PM_curpm might fix as-of-yet
7 unseen problems with regex matches being attributed to the 1202 unseen problems with regex matches being attributed to the

Diff Legend

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