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

Comparing Coro/Changes (file contents):
Revision 1.351 by root, Sun May 25 01:32:15 2008 UTC vs.
Revision 1.694 by root, Sat May 21 17:09:22 2022 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?
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?
7 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 - add connected method as alias for peername in Coro::Handle, as
33 the inherited version does not work for us (analyzed by Simon816).
34 - change jit executable strategy, to hopefully appease more
35 hardened environments.
36
376.57 Wed Jul 29 15:08:44 CEST 2020
38 - override AnyEvent condvar methods "properly", as AnyEvent
39 has support for pluggable condvars since 2011. This also
40 avoids problems with recursive waits by avoiding to mix
41 impl methods with coro code.
42 - allow rouse_wait to be called in a different thread.
43 - slightly clarify rouse_cb/rouse_wait documentation.
44 - slightly improve rouse callback performance and memory usage.
45 - try to report correct calller when throwing exception in
46 on $condvar->recv, not a useless line in Coro::AnyEvent.
47
486.56 Thu Feb 20 19:40:07 CET 2020
49 - Coro::EV::time*once functions used ev_once, which is not cancelable.
50 This could cause problems due to I/O watchers not being stopped up
51 on exceptions/cancellation.
52
536.55 Fri Jul 19 17:28:09 CEST 2019
54 - Coro::RWLock woke up only one reader instead of all after a wrlock
55 (analyzed and patch by Maxim Skorobogatov).
56 - slightly improved documentation of Coro::RWLock and RWLock testcase.
57 - quickly (re)-ported to minix 3.3 before minix crashed again.
58
596.54 Sun Dec 16 11:06:07 CET 2018
60 - reinstate Coro::Select::select prototype - it breaks more than it fixes.
61
626.53 Sun Dec 16 10:33:15 CET 2018
63 - the autogenerated prototypes of Coro::AIO now add ";" at the end
64 to avoid generating named unary operators and messing up parsing
65 (mostly affects aio_mlockall 4 | 8 which was parsed as
66 (aio_mlockall 4) | 8).
67 - do not use a prototype for select in Coro::Select, as CORE::select also
68 doesn't have one.
69
706.52 Tue Aug 14 18:49:43 CEST 2018
71 - libcoro update: allow sharing of coro threads using the pthreads
72 backend among different pthreads, as is required by Coro::Multicore.
73 - always allow ->call and ->eval when the coro is the current coroutine. this
74 makes it possible, among other things, to call these on $Coro::main.
75 - minor code cleanups.
76
776.514 Thu Aug 31 18:25:31 CEST 2017
78 - safe_cancel no longer croaks when called on already-destroyed
79 thread, but instead "just works".
80
816.513 Sat Jul 15 01:18:51 CEST 2017
82 - experimental 5.22 comaptibility changes (PadlistNAMES).
83 - removed attributes example and made sure the other examples
84 still work.
85 - fix some statevar types for newer perls, should be of no consequence.
86
876.512 Fri Jul 14 05:03:17 CEST 2017
88 - seems to pass test suite down till 5.8.9, so make this the
89 minimum requirement in Makefile.PL.
90 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
91 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
92 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
93 fortified longjmp variants in some libcs.
94 - re-try JIT allocation without PROT_EXEC, to hopefully improve
95 portability (still doesn't seem to work around broken selinux
96 mmap everywhere).
97 - speling fixes (Xavier Guimard).
98
996.511 Sun Jun 26 23:44:50 CEST 2016
100 - make it compile with threaded perls again.
101 - simplify and speed up __DIE__ and __WARN__ handling.
102
1036.51 Sat Jun 25 20:57:20 CEST 2016
104 - 6.5 release didn't compile on most older perls due to
105 a logic error enabling experimental 5.24 code for all versions.
106 - since various broken or deficient patches float around
107 to work around the vtbl API breakage in perl, let's provide
108 our own workaround, which seems to be much less invasive
109 and more compatible than the approaches seen so far. YMMV.
110 - work around assertion failure in perl_destruct on 5.24
111 (analyzed by Dave Mitchell).
112
1136.5 Wed Jun 22 22:23:50 CEST 2016
114 - swap_sv swapping order was not symmetrical, causing
115 wrong swaps when swap_sv was used multiple times on the
116 same sv in the same thread.
117 - swap_sv calls can now be undone by calling it again
118 with the same variables.
119 - swap_sv calls will now be undone in async_pool threads.
120 - split Coro::Semaphore::up/adjust into separate xs functions
121 for better error reporting, at slight codesize increase.
122 - (libcoro) arm assembly support, please test and report.
123 - adjust to PL_savestack changes in perl 5.24 (adapted from
124 the debian patch, which unfortunately gets it wrong).
125
1266.49 Sat Oct 17 01:40:12 CEST 2015
127 - throwing an exception to a thread waiting in
128 Coro::Handle using Coro::EV did not stop the watchers,
129 causing the next call to fail (testcase by Martin Pritchard).
130 - bump minimum perl version to 5.10.
131
1326.48 Sun Oct 4 19:03:51 CEST 2015
133 - fix memory leak when Coro::Handle uses Coro::EV internally
134 (testcase by Sten Sten).
135 - update code to libev 4 API (internally, Coro still used the
136 version 3 API calls).
137 - remove unused hv_sig.
138
1396.47 Sat Jul 11 03:58:20 CEST 2015
140 - swap_svs - NVs can be in the SV head beginning with 5.20.
141
1426.46 Tue Jun 30 14:36:30 CEST 2015
143 - restore portability to perls with windows fork emulation
144 (patch by Petr Písař).
145
1466.45 Tue Jun 30 01:40:08 CEST 2015
147 - restore portability to perl 5.16 and below.
148
1496.44 Tue Jun 30 00:41:54 CEST 2015
150 - avoid segfaulting (or worse) when tracing a canceled thread.
151 - the Event module silently broke it's public hook API, causing
152 failures on perls compiled with -Duselongdouble. Requiring
153 the latest version of Event because I don't know which release
154 changed this (it's not mentioned in the ChangeLog).
155 - allow xs level enterleave hooks via CoroAPI.
156
1576.43 Thu Jun 4 15:38:14 CEST 2015
158 - use stability canary.
159 - port to stableperl-5.22.0-1.001.
160 - update libecb.
161
1626.42 Wed Feb 11 20:29:52 CET 2015
163 - Coro::SemaphoreSet->try did not actually work (analyzed by
164 SATO Kentaro).
165 - upgrade libecb to be C11 compliant.
166
1676.41 Sat Sep 6 22:08:46 CEST 2014
168 - restore portability to perl 5.8.x.
169 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
170 up some memory, but fixes "use feature" and similar modules
171 when used inside a Coro.
172 - allow zero as argument to Coro::Channel to mean the same thing
173 as no argument. this works with older versions as well,
174 but wasn't legal until now.
175 - slightly better c header file detection.
176
1776.39 Mon Jun 2 00:00:08 CEST 2014
178 - work around more incompatible changes in 5.20.
179
1806.38 Sun Jun 1 21:54:23 CEST 2014
181 - check that perl slots actually have enough space to hold
182 interpreter variables.
183 - untested port to perl 5.19 (64 bit tmps indices) (reported
184 by Andreas König).
185 - croak when cancel is called without a thread context.
186
1876.37 Tue Mar 4 13:27:33 CET 2014
188 - *sigh*, removed leftover debugging code from debugging a
189 perl bug, of all things.
190
1916.36 Tue Mar 4 07:11:59 CET 2014
192 - semaphores would not clear the destroy hook when interrupted
193 by ->throw, causing segfaults or worse.
194 - ->throw on a thread waiting for a semaphore did not acquire
195 the semaphore, but also didn't wake up other waiters,
196 possibly causing a deadlock.
197 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
198 to report a full stacktrace, which should help find out
199 where the actual call is.
200 - "atomically" destroy data in slf_destroy, because it is
201 the right thing to do, just in case.
202 - disable ecb.h libm dependency, to compile on ancient systems
203 or under adverse conditions.
204
2056.33 Mon Nov 18 11:26:27 CET 2013
206 - do not crash when freeing padlists with holes (in 5.18).
207 - tentative SVt_BIND 5.19 port/fix.
208
2096.32 Tue Nov 5 15:35:35 CET 2013
210 - use a new algorithm to derive padlists for perl 5.18. The old
211 one could lead to 0-pointer accesses inside perl (reported
212 by Darin McBride).
213
2146.31 Thu May 9 07:39:48 CEST 2013
215 - Coro::AIO requests would crash if the thread was ready'd
216 while the request was ongoing.
217
2186.29 Wed May 8 02:55:18 CEST 2013
219 - when an on_destroy handler destructs the coro currently being
220 destructed a perl scalar could be accessed after being freed,
221 likely causing a crash.
222
2236.28 Wed Mar 6 06:58:02 CET 2013
224 - clean remnants of existing __DIE__ and __WARN__ handlers so
225 they lose their magic and will not cause segfaults later
226 (testcase by Andrey Sagulin).
227 - improved Coro::State documentation a bit.
228 - Coro::Debug::command now flushes the output.
229 - add hack detection code for x32 abi, because the braindead slugs
230 who designed that made it look exactly like x86_64 without
231 providing proper compile time symbols to test for it. as a result,
232 this detection cannot work reliably.
233 - valgrind stack registering was broken.
234 - do not rely on Time::HiRes anymore in Coro::Debug.
235
2366.23 Fri Dec 7 23:36:37 CET 2012
237 - use experimental fiber implementation on native windows
238 perls.
239 - use sizeof (void *) as multiplication factor for stack sizes,
240 to accomodate the totally braindamaged microsoft 64 bit "os".
241 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
242 activeperl 5.16 crashes when PerlIO_define_layer is called due
243 to some bug in the perl dll, strawberry perl at least passes
244 the testsuite.
245 - implement Coro::Handle->peeraddr/host/port, for slightly
246 improved compatibility with LWP.
247 - implement 5.17 compatibility by almost blindly applying a
248 good-looking patch by Father Chrysostomos.
249 - move stack management functions into libcoro 3.
250 - libcoro version 3 "released".
251 - support magic values as timed_io_once args.
252 - recommend AnyEvent 7+ or EV 4+, also require EV
253 version 4 or newer for Coro::EV.
254
2556.10 Tue Oct 9 01:14:27 CEST 2012
256 - updated ecb.h, it had a typo that caused it to not compile on many
257 big endian systems (reported by many people).
258 - disable memory fences in ecb.h to improve portability.
259
2606.09 Sat Oct 6 23:25:02 CEST 2012
261 - Coro::EV I/O watchers were not interruptible by exceptions
262 (Coro::State::throw) (testcase by sten).
263 - ->throw now puts threads into the ready queue, as this seems to
264 be expected by existing code, and code that doesn't cope with spurious
265 wakeups needs fixing anyway.
266 - use fd -1 in mmap.
267 - cast I32 to int in error message printf.
268 - warn about broken so-called "hardened" kernels.
269
2706.08 Fri Apr 13 12:05:47 CEST 2012
271 - be more aggressive about exiting like perl does - formerly,
272 exiting from the non-main thread would not execute END blocks.
273
2746.07 Fri Nov 11 21:21:48 CET 2011
275 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
276 - update ecb.h.
277
2786.06 Mon Aug 8 23:59:48 CEST 2011
279 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
280 backend, which is an order of magnitude slower. unfortunately, cygwins
281 pthread implementation isn't very complete either, so allocate the stack
282 twice just to be sure.
283 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
284 in its ithreaded perl - best recompile cygwin and use the 'w'indows
285 backend for much better performance. also disable ithreads for
286 even better performance...).
287
2886.05 Thu Aug 4 21:36:36 CEST 2011
289 - blush, condvar values would not be propagated from send to recv anymore
290 (reported by Chip Salzenberg).
291 - use exponential increase for the readline buffer length in
292 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
293
2946.04 Wed Aug 3 17:15:45 CEST 2011
295 - use even more efficient and more compatible condvars for
296 compatibility to AnyEvent 6.x :)
297 - more inconsequential ecb.h updates.
298
2996.03 Wed Aug 3 11:41:30 CEST 2011
300 - change how Coro patches AnyEvent condvars for compatibility to
301 AnyEvent 6.x.
302 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
303 hasn't been included.
304
3056.02 Wed Jul 13 04:35:19 CEST 2011
306 - "improve portability to Gentoo" - gentoo manages to put perl variables
307 in memory areas that are farther than 2gb apart, which the jit couldn't
308 handle and barfed. now it's just a bit slower on gentoo and similar
309 systems.
310
3116.01 Sun Jul 3 12:31:14 CEST 2011
312 - workarounds are good, but the test for whether pthreads are used
313 was not good. this one should be better.
314 - check differently whether gcc generates cfi instructions itself.
315
3166.0 Wed Jun 29 19:43:35 CEST 2011
317 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
318 destroyed and cleaned up automatically (e.g. async { schedule }).
319 - implement a JIT compiler for part of the thread switch code,
320 which gives a 50% speed improvement on threaded perls, and
321 about 4% on non-threaded perls (so threaded perls now finally
322 reach about half the speed of non-threaded perls).
323 - slightly modernise Coro::Intro, add section about rouse functions.
324 - avoid DEFSV and ERRSV, giving another 10% improvement
325 in thread switching.
326 - Coro::State->is_destroyed is now called is_zombie.
327 - implement a Coro->safe_cancel method that might fail, but
328 cancels in a "safer" way if it succeeds.
329 - add preliminary support for DEBUGGING perls.
330 - get rid of two hash-accesses when initialising a new Coro - this
331 speeds up coro creation by almost a factor of two.
332 - croak when a coro that is being cancelled tries to block
333 (e.g. while executing a guard block), instead of crashing or
334 deadlocking.
335 - use a more robust and also faster method to identify Coro::State
336 objects - speeds up everything a bit.
337 - implement Coro->cancel in XS for a 20% speed improvement, and to
338 be able to implement mutual cancellation.
339 - speed up context switches by a percent or two by more efficiently
340 allocating context stack entries.
341 - implement Coro->join and Coro->on_destroy in XS for a speedup and
342 a reduction in memory use.
343 - cancelling a coro while it itself is cancelling another coro is
344 now supported and working, instead of triggering an assertion.
345 - be a bit more crash-resistant when calling (buggy) on_destroy
346 callbacks (best effort).
347 - move on_destroy into the slf_frame, to allow extension slf
348 functions to have destructors.
349 - get rid if coro refcounting - simply crash in other interpreter
350 threads by nulling the pointers on clone.
351 - simplify warn/die hook handling when loading Coro - the convoluted
352 logic seems to be no longer neccessary.
353 - use libecb instead of our own home-grown gcc hacks.
354 - document alternatives to Coro::LWP. Please use them :)
355 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
356 sigaltstack. Really. wine suffers from it, erlang suffers from it,
357 and it's known since at least 2006.
358
3595.372 Wed Feb 23 06:14:30 CET 2011
360 - apparently mingw doesn't provide a working gettimeofday, try to
361 work around that by relying on Time::HiRes (indirectly brought to
362 my attention by Max Maischein).
363 - fix some portability issues when Time::HiRes was used.
364
3655.371 Mon Feb 21 14:36:08 CET 2011
366 - backport to windows process emulation code again.
367
3685.37 Sat Feb 19 07:49:44 CET 2011
369 - add a big "Coro thread life cycle" section to "man Coro".
370 - try a tentative workaround against the breakage that 5.13 has
371 introduced without depreciation period. sigh.
372 - no longer use Time::HiRes if gettimeofday is available, which
373 saves quite a lot of memory.
374
3755.36 Sun Feb 13 05:33:41 CET 2011
376 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
377 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
378 method call.
379 - undocument Coro::Timer::sleep and obsolete whole module.
380 - optimise Coro::Timer::timeout memory and cpu usage.
381 - slightly updated Coro::Intro for recent changes.
382 - do not initialise PL_dirty anymore.
383
3845.25 Thu Nov 11 01:08:39 CET 2010
385 - try a different approach on netbsd - netbsd 5 finally has marginally
386 working pthreads, but still broken ucontext/sigaltstack.
387 - openbsd 4.8 finally got their act together, Coro works out of the box
388 with asm, setjmp and pthreads (no change, just informational).
389
3905.24 Sat Oct 23 11:27:12 CEST 2010
391 - port to the EV 4.0 API.
392 - work around bugs in mingw32, making strawberry perl work
393 out of the box.
394 - correctly modify Coro::AIO function prototypes
395 so that they reflect the "no optional parameters" rule.
396 - "ported" libcoro to C++.
397
3985.23 Mon May 17 18:50:42 CEST 2010
399 - be more resistant to ordering changes when initialising
400 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
401 Waldorf).
402 - document that perl 5.12 deliberately removed support for cloning.
403
4045.22 Wed Apr 14 03:55:35 CEST 2010
405 - correctly return udnef on errors in Coro::Handle::read/write
406 (testcase by Marc Mims).
407 - convert Coro::Util into a "perl compatibility wrapper" - the functions
408 are less useful now, but are drop-in replacements for existing
409 functions, listing better alternatives in the documentation. This also
410 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
411 with Coro::Util::inet_aton.
412 - do not override $Coro::idle unconditionally in Coro.pm, as other
413 modules could have provided their own idle coro already
414 (for exmaple, Coro::AnyEvent).
415 - fix Coro::Util::gethost* functions.
416 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
417
4185.21 Wed Dec 16 07:19:51 CET 2009
419 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
420 together.
421 - add some examples on how to combine other event loops with Coro in
422 Coro::AnyEvent, and how to run it (and not to block). Seems to be
423 the most common source of confusion.
424 - try to catch people naively blocking in an event callback.
425 - work around the perl filehandle bug issue in conjunction with
426 older common::sense (as indirectly pointed out by ZSystem).
427 - clarify the "not from signal handlers" section.
428
4295.2 Sun Oct 4 14:54:24 CEST 2009
430 - Coro::Storable destroyed the prototypes of the functions it wrapped.
431 - export rouse_cb and rouse_wait by default now.
432 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
433 - new method $state->swap_sv.
434 - added section on "windows process emulation" to the manpage,
435 after a not-so-fruitful (nor-friendly) "discussion" with chip
436 salzenberg (discussion implies arguments, but his only arguments
437 were ad-hominems, one wonders why he started it in the first
438 place). I hope this explains it well enough for him to understand,
439 and maybe well enough for others to understand.
440 - use common::sense everywhere now.
441 - idle callbacks are no longer supported, use idle coros instead.
442 - print a thread listing when a deadlock is detected.
443
4445.17 Sat Aug 22 23:09:31 CEST 2009
445 - work around a bug in the perl debugger causing crashes
446 when running under the debugger by marking _pool_handler as nodebug.
447 - speed up Coro::async considerably.
448 - try some hacks to get netbsd to work "more often" - their broken
449 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
450 - convert Coro to AE 5.0 API.
451
4525.162 Tue Jul 28 04:04:03 CEST 2009
453 - perl 5.8.2 is now minimum requirement.
454 - skip t/19_handle.t on broken windows perls.
455
4565.161 Wed Jul 22 04:47:38 CEST 2009
457 - Coro::AnyEvent::poll could have a different prototype when EV was
458 used as backend (analyzed by Tatsuhiko Miyagawa).
459 - Coro::AnyEvent errornously initialised the event loop when loaded,
460 not on demand.
461 - try to workaround rare */t/01_unblock.t failures.
462
4635.16 Tue Jul 21 01:44:37 CEST 2009
464 - Coro::AnyEvent failed to hook into the event loop
465 when no threads had been readied between detecting
466 the event loop and actually running it.
467 - considerably speed up Coro::Select by taking avdantage
468 of AnyEvent > 4.8 and some other optimisations.
469 - implement paragraph readline mode in Coro::Handle
470 (based on patches by Zsbán Ambrus).
471 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
472 and analyzed by Yasuhiro MATSUMOTO).
473 - clarified libcoro license and copyright.
474 - someone stole my EXTRA_META!!!
475 - implement Coro::Select::patch_pp_sselect and it's brother,
476 for hardcode select overriding.
477
4785.151 Mon Jul 6 05:41:57 CEST 2009
479 - backport to windows process emulation code again (patch by
480 Yasuhiro MATSUMOTO).
481 - slightly update Coro::MakeMaker.
482
4835.15 Tue Jun 30 10:28:06 CEST 2009
484 - deprecate Coro::Socket, document how to get ipv6 support via
485 AnyEvent::Socket instead.
486 - implement signal->wait ($cb) interface, similar to semaphores.
487 - work around SvOK not supporting getmagic, so we have to getmagic
488 to test for undef :/ (reported by Matthias Waldorf).
489 - load Coro::AnyEvent in all modules using AnyEvent.
490 - work around perl corrupting our internal data structures,
491 reported by Tokuhiro Matsuno.
492 - enable per-coroutine real and cpu time gathering
493 (Coro::State::enable_times).
494
4955.14 Wed Jun 24 01:37:48 CEST 2009
496 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
497 as this seems to have been a difficult concept (poll, sleep, idle,
498 idle_upto).
499 - add Coro::AnyEvent::readable/writable functions.
500 - clarify Coro::EV/Event/AnyEvent manpages.
501 - free per-thread global scalars in the thread calling ->cancel, to
502 avoid crashes when $_, $@ etc., are magical but some of those
503 had already been freed.
504 - "unexperimentalise" the callback interface for Coro::Semaphore.
505 - speed up ready queue management/context switching by using a linked
506 list instead of an array (~5-10%).
507 - implement "watch" command in Coro::Debug shells.
508 - for fun, implement time-slicing as an example in the manpage.
509 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
510 Coro::Event, use the normal AnyEvent handling instead of dieing
511 (the same is true for Coro::Handle).
512 - properly document Coro::EV::timed_io_once.
513 - avoid unneccessary ->cancel calls in Coro::Handle.
514 - maybe make it work on mingw32 with win32 backend
515 (based on patch by Yasuhiro Matsumoto).
516
5175.132 Fri May 29 09:00:39 CEST 2009
518 - do not keep a reference to the argument itself in
519 Coro::Semaphore::guard, as it could change later.
520 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
521 should support a prepare callback.
522
5235.131 Mon Mar 16 23:20:37 CET 2009
524 - implement and document Coro->suspend, Coro->resume.
525 - fix Coro::Select implementation to not (often) close
526 the passed file descriptors (testcase provided by pippijn).
527
5285.13 Mon Dec 15 21:51:42 CET 2008
529 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
530 - don't set diehook to C<undef> but instead to NULL, to avoid
531 spurious warnings.
532 - fix a lot of bugs in Coro::SemaphoreSet.
533 - Coro::SemaphoreSet will less often create a semaphore needlessly.
534 - add Coro::SemaphoreSet::count and wait methods.
535 - take advantage of the new Guard module.
536 - deprecate Coro::guard.
537 - try to fix the dreaded 01_unblock tests once more. I hate it when
538 testsuites need more fixing than the code they are supposed to test.
539 - croak in more cases when a required callback isn't resolvable.
540 - fix some minor issues in Coro::State->call/eval.
541 - use current coroutine context instead of a temporary one
542 when temporarily switching to another coroutine.
543 - do not call C-level on_destroy handlers during global destruction,
544 to avoid needless segfaults.
545
5465.12 Sun Dec 7 13:30:38 CET 2008
547 - add default config for MirOS, which seems to be a bug-to-bug
548 compatible fork of openbsd ("world domination by releasing
549 openbsd cvs before the openbsd folks do it" or so :).
550 - free_padlist did destroy the names pad, not good, but didn't
551 seem to bother perl - this could fix issues such as eval ""
552 inside a function called from multiple coroutines.
553 - use a different method to detect destruction time.
554 - be more careful when freeing padlists just before global
555 destruction.
556 - fixed and expanded the call/cc example.
557 - renamed _terminate to _coro_run.
558 - new method Coro::Channel->shutdown.
559 - try pthreads on openbsd <4.4 (broken sigaltstack, will
560 pthreads fare better?).
561 - be less picky about destroying "a" running coroutine.
562
5635.11 Tue Nov 25 21:49:05 CET 2008
564 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
565 working code with failed assertions, introducing more bugs than
566 it fixes, requiring elaborate workarounds :(
567
5685.1 Mon Nov 24 08:54:59 CET 2008
569 - wrote a small introductory tutorial - Coro::Intro.
570 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
571 - Coro::Select did errornously dup the file descriptors
572 and didn't work with all AnyEvent backends.
573 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
574 LWP data transfers.
575 - disassociate c contexts from coro objects - this is agruably more
576 correct, but mostly allows sharing of cctxs between coro and state
577 objects, for added memory savings and speed increases.
578 - bumped $Coro::POOL_RSS up to 32kb by default.
579 - no longer set the optype to OP_CUSTOM, as B::* understandably
580 doesn't like this very much (and we *are* a type of entersub).
581 - implement state cloning, just to prove that call/cc can be done.
582 - automatically load Coro::AnyEvent in Coro::Handle.
583 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
584 do not have this method (and don't need it either).
585 - speed up generic anyevent methods in Coro::Handle by using rouse
586 callbacks.
587 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
588 others. It also makes the debugger happier, as you can trace
589 through the idle threads now.
590 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
591 - eg/event was pretty broken.
592 - better 5.8.6 compatibility.
593
5945.0 Thu Nov 20 10:35:05 CET 2008
595 - NEW ARCHITECTURE: use the latest 4.x version if you experience
596 stability issues.
597 - bump API version to 7 - all dependents must be recompiled.
598 - removed timed_* functions - they were not being used anyways
599 and should be replaced by a more generic mechanism -
600 and were annoying to support anyways :)
601 - removed SemaphoreSet's waiter method - use sem method instead.
602 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
603 - async_pool did free a scalar value twice
604 ("Attempt to unreference...").
605 - fix a longstanding bug where calling terminate on a coro that
606 was waiting for a semaphore that was just becoming available
607 would cause a deadlock (semaphore would get into a state where
608 it was available but waiters were still blocked).
609 - calling throw on a coroutine that is waiting for a semaphore will
610 no longer make it acquire the semaphore (and thus leak a count).
611 - perl's process emulation is now not even theoretically supported
612 anymore.
613 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
614 conversion of callback-style to blocking-style.
615 - new methods $coro->schedule_to and ->cede_to, to specifically
616 schedule or cede to a specific coroutine.
617 - new function Coro::Semaphore::wait.
618 - use named constants in Coro::Channel (Richard Hundt).
619 - directly patch the entersub opcode calling SLF functions (cede,
620 transfer and so on). this does speed up context switching, but
621 more importanly, it frees us from the hardcoded behaviour of
622 entersub, so we might actually be able to return something from
623 those functions and atcually create new ones.
624 - take advantage of __builtin_frame_address on gcc.
625 - expose THX in coroapi (not sure whether this was a wise decision,
626 as "threaded" perls are running at half speed anyways).
627 - implement execute_slf (schedule-like-function) interface that makes
628 it possible to implement schedule-like-functions in XS.
629 - use new SLF interface to massively speed up Coro::EV by roughly a
630 factor of two.
631 - used new SLF interface to massively speed up Coro::Semaphore by a
632 factor of three.
633 - used new SLF interface to speed up Coro::AIO by roughly a factor of
634 four and reduce its memory usage considerably.
635 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
636 for a nice speedup and vastly more correct behaviour. Also implement
637 a new method "sem" to get at the underlying semaphore object.
638 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
639 (in comparison) 20-40% speedup.
640 - used new SLF interface to reimplement Coro::Signal gaining
641 some unknown (because I was too lazy), but certain, speedup, and also
642 making signals reliable for the first time.
643 - used new SLF interface and other optimisations to speed up async_pool
644 by a factor of two. It also doesn't rely on perl's exception mechanism
645 to exit anymore. The overhead for terminating an async_pool, coro over
646 a normal async is now very small.
647 - sped up coroutine creation/destruction by 40%.
648 - forgot to include Coro/libcoro/README in the dist for all these years.
649 - work around a freebsd pthreads bug (manual testcancel is required as
650 pthread_cond_wait isn't a cancellation point on freebsd).
651 - use new rouse functions to speed up and simplify Coro::BDB.
652 - make "prefer perl native functions" work with threaded perls.
653 - condense Coro::Debug ps output, hint at v and w flags.
654 - (libcoro) lots of minor cleanups and portability improvements.
655
6564.914 Wed Nov 19 12:54:18 CET 2008
657 - fix a disastrous bug in the readline optimisation
658 introduced in 4.801.
659
6604.913 Sat Nov 15 07:58:28 CET 2008
661 - async_pool did free a scalar value twice
662 ("Attempt to unreference...").
663
6644.912 Thu Nov 13 18:31:23 CET 2008
665 - minor cleanups.
666 - use much larger stacks on linux and perl < 5.8.8.
667 - Coro::Debug::new_unix_server did not unlink the socket
668 when destroyed.
669
6704.911 Tue Nov 11 04:26:17 CET 2008
671 - "port" to threaded perls.
672
6734.91 Mon Nov 10 05:36:38 CET 2008
674 - the ->throw exception object no longer leaks.
675 - creating a new cctx leaked a scopestack entry (memleak).
676 - new coroutines didn't get created with a zero flags field
677 (unknown impact).
678 - calling ->throw on a not-yet-started coroutine should now work
679 instead of being ignored.
680 - ->throw is now supported on Coro::State objects.
681 - clean up cctx creation code a bit.
682 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
683
6844.9 Sat Nov 8 17:45:27 CET 2008
685 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
686 - (libcoro) no longer leak threads in the experimental pthread backend,
687 also speed it up considerably.
688 - (libcoro) do not rely on makecontext passing void *'s unscathed.
689 - fix compiletime dependencies on libcoro in the Makefile.
690 - cctx_count wasn't always updated properly.
691 - Coro::State::cctx_stacksize wasn't applied correctly.
692 - new function Coro::State::cctx_max_idle.
693 - the default max number of idle C contexts is now 4.
694 - (libcoro) try harder to get _setjmp/_longjmp.
695 - (libcoro) cleanup and extend the libcoro API to officially
696 allow the creation of empty source contexts.
697 - very experimental workaround for the totally broken netbsd platform.
698 - tried to hack around openbsd bugs.
699
7004.804 Wed Nov 5 16:36:00 CET 2008
701 - Coro::Debug::new_unix_server would not create a non-blocking listening
702 socket, sometimes causing freezes.
703 - (libcoro) fix misaligned stack points for setjmp and assembly
704 methods, which can cause crashes on x86/x86_64 with a sufficiently
705 aggressive compiler.
706 - new function: Coro::Debug::new_tcp_server.
707 - move ->throw into the Coro class because it only works on coro objects.
708
7094.803 Mon Nov 3 17:16:12 CET 2008
710 - (libcoro) use a global asm statement to become independent of gcc
711 otpimisations for CORO_ASM (thanks to pippijn for the idea).
712 - try to workaround yet another broken bsd, this time dragonfly.
713
7144.802 Thu Oct 30 10:56:12 CET 2008
715 - support -fno-omit-frame-pointer on x86 with the assembly method.
716 - tune 01_unblock.t tests a bit.
717
7184.801 Wed Oct 22 18:33:37 CEST 2008
719 - improve readline speed for very long "lines".
720 - backport to 5.8.8.
721
7224.8 Thu Oct 2 13:34:40 CEST 2008
723 - new function Coro::AIO::aio_wait.
724 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
725 - greatly speed up and reduce memory usage of Coro::AIO requests.
726 - implement some other µ-optimisations.
727
7284.749 Mon Sep 29 14:40:12 CEST 2008
729 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
730
7314.748 Sat Sep 27 14:03:19 CEST 2008
732 - implement, but do not document, PerlIO::cede(granularity).
733 - Coro::Storable forgot to wrap Storable::pstore.
734 - work around the multitude of leaks and memory corruption
735 bugs in PerlIO::via by using our own C-level perliol. As a side
736 effect, Coro::Storable is now much, much, much faster.
737
7384.747 Tue Sep 23 01:59:41 CEST 2008
739 - fix a per-cv memleak (one empty array was leaked per
740 code reference).
741 - avoid a crash in coro->call|rss when the coroutine was already
742 destroyed (most noticably when using Coro::Debug::ps :)
743 - also protect *Storable::FILE.
744 - push up default storable granularity to 20ms.
745
7464.746 Sun Sep 21 03:22:20 CEST 2008
747 - be more insistent on locking Storable against reentrancy
748 in Coro::Storable.
749 - move swap_def?v and throw to Coro::State, as documented.
750
7514.745 Thu Jul 24 00:14:38 CEST 2008
752 - remove debugging code related to MgPV_nolen_const, also try to
753 make it compile with perl 5.8.6 (yes, apple apparently loves
754 outdated software). Reported by John S.
755
7564.744 Tue Jul 8 22:06:35 CEST 2008
757 - correctly provide default DIE/WARN handlers as documented.
758 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
759 implementations inside perl work fine for us.
760
7614.743 Mon Jun 16 00:21:57 CEST 2008
762 - when using Coro::EV without running EV::loop it could
763 result in busy-waiting for events, this has been fixed.
764 - reduce codesize and improve performance by using EV_DEFAULT_UC.
765
7664.742 Sat May 31 14:10:21 CEST 2008
767 - implement a workaround for (some) perl <5.8.8 versions.
768 - require EV 3.3+.
769
7704.741 Fri May 30 23:33:09 CEST 2008
771 - tell netbsd how utterly broken their imitation of an OS is
772 and refuse to build by default if pthreads are in use.
773 - switch to "s" method on all bsd's by default, as their ucontext
774 stuff seems just too broken.
775 - fix a bug in Coro::Select.
776
7774.74 Thu May 29 20:05:31 CEST 2008
778 - do not test Coro::LWP for lack of dependencies.
779 - do not test Coro::Select for lack of working perls.
780
7814.73 Thu May 29 2008
782 - fix a bug in Coro::EV which would cause it to block despite
783 there being runnable coroutines.
784 - sprinkle "no warnings" freely over everything, also suppress
785 warnings for some other modules.
786 - fix typo in WSAEWOULDBLOCK.
787
7884.72 Sun May 25 05:14:36 CEST 2008
8 - tweak META.yaml a bit, unfortunately, there is no documented way 789 - tweak META.yaml a bit, unfortunately, there is no documented way
9 to have optional dependencies with CPAN. doh :( 790 to have optional dependencies with CPAN. doh :(
10 - avoid running some tests on windows because they would fail due to 791 - avoid running some tests on windows because they would fail due to
11 perl bug (broken fork, broken pipes...). 792 perl bug (broken fork, broken pipes...).
12 - work around perl on windows bugs where perl returns undocumented 793 - work around perl on windows bugs where perl returns undocumented
13 error codes for sysread, syswrite etc. by taking advantage 794 error codes for sysread, syswrite etc. by taking advantage
… …
17 - use unix domain sockets in testsuite to work around 798 - use unix domain sockets in testsuite to work around
18 common perl implementation bugs on widows (they are emulated by 799 common perl implementation bugs on widows (they are emulated by
19 tcp sockets on windows. ugh.) 800 tcp sockets on windows. ugh.)
20 801
214.71 Sat May 24 20:01:27 CEST 2008 8024.71 Sat May 24 20:01:27 CEST 2008
22 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 803 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
23 - take advantage of async name resolution of AnyEvent::Util. 804 - take advantage of async name resolution of AnyEvent::Util.
24 - work around brutal inet_aton override in Coro::LWP. 805 - work around brutal inet_aton override in Coro::LWP.
25 - take advantage of the readyhook in Coro::EV, for smoother 806 - take advantage of the readyhook in Coro::EV, for smoother
26 scheduling. 807 scheduling.
27 808
284.7 Sun May 11 00:32:19 CEST 2008 8094.7 Sun May 11 00:32:19 CEST 2008
29 - completely reworked the Coro manpage. 810 - completely reworked the Coro manpage.
30 - added Coro::AnyEvent, generic event loop integration. 811 - added Coro::AnyEvent, generic event loop integration.
31 - implement cancel, ready and kill commands in Coro::Debug. 812 - implement cancel, ready and kill commands in Coro::Debug.
32 - document find_coro in Coro::Debug. 813 - document find_coro in Coro::Debug.
33 - incompatible change: rename has_stack to has_cctx. 814 - incompatible change: rename has_stack to has_cctx.
34 - Coro::AIO and Coro::BDB no longer force event model detection, 815 - Coro::AIO and Coro::BDB no longer force event model detection,
35 use AnyEvent::AIO and AnyEvent::BDB. 816 use AnyEvent::AIO and AnyEvent::BDB.
… …
159 940
1604.11 Thu Oct 11 02:40:24 CEST 2007 9414.11 Thu Oct 11 02:40:24 CEST 2007
161 - port to threaded perls. 942 - port to threaded perls.
162 943
1634.1 Thu Oct 11 02:38:16 CEST 2007 9444.1 Thu Oct 11 02:38:16 CEST 2007
164 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 945 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
165 be local to each coro (see Coro::State). 946 be local to each coro (see Coro::State).
166 - incompatible change: for very deep reasons, cede and cede_notself 947 - incompatible change: for very deep reasons, cede and cede_notself
167 cannot return anything, so nothing will be returned. 948 cannot return anything, so nothing will be returned.
168 - possibly bring back 5.10 compatibility (untested). 949 - possibly bring back 5.10 compatibility (untested).
169 - work around stupid (and wrong) warning on 5.10 :(. 950 - work around stupid (and wrong) warning on 5.10 :(.
… …
241 but stateful semantics. 1022 but stateful semantics.
242 - fixed a lot of typos in Coro.pm (patch submitted by David 1023 - fixed a lot of typos in Coro.pm (patch submitted by David
243 Steinbrunner, which applied flawlessly). 1024 Steinbrunner, which applied flawlessly).
244 1025
2453.6 Sat Apr 14 17:13:31 CEST 2007 10263.6 Sat Apr 14 17:13:31 CEST 2007
246 - added some bugfixes to get eg/myhttpd working again. 1027 - added some bugfixes to get eg/myhttpd working again.
247 - added Coro::Storable for often-cede'ing freeze/thaw. 1028 - added Coro::Storable for often-cede'ing freeze/thaw.
248 - try to do a clean exit when a coroutine calls exit 1029 - try to do a clean exit when a coroutine calls exit
249 (EXPERIMENTAL). 1030 (EXPERIMENTAL).
250 - got rid of indirect call through _coro_init. 1031 - got rid of indirect call through _coro_init.
251 - updated the partly antique examples in eg/ to 1032 - updated the partly antique examples in eg/ to

Diff Legend

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