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

Comparing Coro/Changes (file contents):
Revision 1.412 by root, Fri Nov 14 20:35:49 2008 UTC vs.
Revision 1.680 by root, Fri Jul 12 06:33:33 2019 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::coroaio (perl leaks like hell). 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: maybe implement a default message channel, very much like Erlang's 4TODO: channel->maxsize(newsize)?
5 actor model (which is cool in a lot of important aspects (failures!), 5TODO: __GCC_HAVE_DWARF2_CFI_ASM
6 but very lacking in others (higher level ipc)). 6TODO: swap_sv, maybe add scope_swap_sv?
7TODO: MAP_STACK 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?
8 10
9// tobench 1.500 11 - quickly (re)-ported to minix 3.3 before minix crashed again.
10// evbench 305436 12
115.0 136.54 Sun Dec 16 11:06:07 CET 2018
12 - NEW ARCHITECTURE: use 4.912 for the stable version. 14 - reinstate Coro::Select::select prototype - it breaks more than it fixes.
15
166.53 Sun Dec 16 10:33:15 CET 2018
17 - the autogenerated prototypes of Coro::AIO now add ";" at the end
18 to avoid generating named unary operators and messing up parsing
19 (mostly affects aio_mlockall 4 | 8 which was parsed as
20 (aio_mlockall 4) | 8).
21 - do not use a prototype for select in Coro::Select, as CORE::select also
22 doesn't have one.
23
246.52 Tue Aug 14 18:49:43 CEST 2018
25 - libcoro update: allow sharing of coro threads using the pthreads
26 backend among different pthreads, as is required by Coro::Multicore.
27 - always allow ->call and ->eval when the coro is the current coroutine. this
28 makes it possible, among other things, to call these on $Coro::main.
29 - minor code cleanups.
30
316.514 Thu Aug 31 18:25:31 CEST 2017
32 - safe_cancel no longer croaks when called on already-destroyed
33 thread, but instead "just works".
34
356.513 Sat Jul 15 01:18:51 CEST 2017
36 - experimental 5.22 comaptibility changes (PadlistNAMES).
37 - removed attributes example and made sure the other examples
38 still work.
39 - fix some statevar types for newer perls, should be of no consequence.
40
416.512 Fri Jul 14 05:03:17 CEST 2017
42 - seems to pass test suite down till 5.8.9, so make this the
43 minimum requirement in Makefile.PL.
44 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
45 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
46 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
47 fortified longjmp variants in some libcs.
48 - re-try JIT allocation without PROT_EXEC, to hopefully improve
49 portability (still doesn't seem to work around broken selinux
50 mmap everywhere).
51 - speling fixes (Xavier Guimard).
52
536.511 Sun Jun 26 23:44:50 CEST 2016
54 - make it compile with threaded perls again.
55 - simplify and speed up __DIE__ and __WARN__ handling.
56
576.51 Sat Jun 25 20:57:20 CEST 2016
58 - 6.5 release didn't compile on most older perls due to
59 a logic error enabling experimental 5.24 code for all versions.
60 - since various broken or deficient patches float around
61 to work around the vtbl API breakage in perl, let's provide
62 our own workaround, which seems to be much less invasive
63 and more compatible than the approaches seen so far. YMMV.
64 - work around assertion failure in perl_destruct on 5.24
65 (analyzed by Dave Mitchell).
66
676.5 Wed Jun 22 22:23:50 CEST 2016
68 - swap_sv swapping order was not symmetrical, causing
69 wrong swaps when swap_sv was used multiple times on the
70 same sv in the same thread.
71 - swap_sv calls can now be undone by calling it again
72 with the same variables.
73 - swap_sv calls will now be undone in async_pool threads.
74 - split Coro::Semaphore::up/adjust into separate xs functions
75 for better error reporting, at slight codesize increase.
76 - (libcoro) arm assembly support, please test and report.
77 - adjust to PL_savestack changes in perl 5.24 (adapted from
78 the debian patch, which unfortunately gets it wrong).
79
806.49 Sat Oct 17 01:40:12 CEST 2015
81 - throwing an exception to a thread waiting in
82 Coro::Handle using Coro::EV did not stop the watchers,
83 causing the next call to fail (testcase by Martin Pritchard).
84 - bump minimum perl version to 5.10.
85
866.48 Sun Oct 4 19:03:51 CEST 2015
87 - fix memory leak when Coro::Handle uses Coro::EV internally
88 (testcase by Sten Sten).
89 - update code to libev 4 API (internally, Coro still used the
90 version 3 API calls).
91 - remove unused hv_sig.
92
936.47 Sat Jul 11 03:58:20 CEST 2015
94 - swap_svs - NVs can be in the SV head beginning with 5.20.
95
966.46 Tue Jun 30 14:36:30 CEST 2015
97 - restore portability to perls with windows fork emulation
98 (patch by Petr Písař).
99
1006.45 Tue Jun 30 01:40:08 CEST 2015
101 - restore portability to perl 5.16 and below.
102
1036.44 Tue Jun 30 00:41:54 CEST 2015
104 - avoid segfaulting (or worse) when tracing a canceled thread.
105 - the Event module silently broke it's public hook API, causing
106 failures on perls compiled with -Duselongdouble. Requiring
107 the latest version of Event because I don't know which release
108 changed this (it's not mentioned in the ChangeLog).
109 - allow xs level enterleave hooks via CoroAPI.
110
1116.43 Thu Jun 4 15:38:14 CEST 2015
112 - use stability canary.
113 - port to stableperl-5.22.0-1.001.
114 - update libecb.
115
1166.42 Wed Feb 11 20:29:52 CET 2015
117 - Coro::SemaphoreSet->try did not actually work (analyzed by
118 SATO Kentaro).
119 - upgrade libecb to be C11 compliant.
120
1216.41 Sat Sep 6 22:08:46 CEST 2014
122 - restore portability to perl 5.8.x.
123 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
124 up some memory, but fixes "use feature" and similar modules
125 when used inside a Coro.
126 - allow zero as argument to Coro::Channel to mean the same thing
127 as no argument. this works with older versions as well,
128 but wasn't legal until now.
129 - slightly better c header file detection.
130
1316.39 Mon Jun 2 00:00:08 CEST 2014
132 - work around more incompatible changes in 5.20.
133
1346.38 Sun Jun 1 21:54:23 CEST 2014
135 - check that perl slots actually have enough space to hold
136 interpreter variables.
137 - untested port to perl 5.19 (64 bit tmps indices) (reported
138 by Andreas König).
139 - croak when cancel is called without a thread context.
140
1416.37 Tue Mar 4 13:27:33 CET 2014
142 - *sigh*, removed leftover debugging code from debugging a
143 perl bug, of all things.
144
1456.36 Tue Mar 4 07:11:59 CET 2014
146 - semaphores would not clear the destroy hook when interrupted
147 by ->throw, causing segfaults or worse.
148 - ->throw on a thread waiting for a semaphore did not acquire
149 the semaphore, but also didn't wake up other waiters,
150 possibly causing a deadlock.
151 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
152 to report a full stacktrace, which should help find out
153 where the actual call is.
154 - "atomically" destroy data in slf_destroy, because it is
155 the right thing to do, just in case.
156 - disable ecb.h libm dependency, to compile on ancient systems
157 or under adverse conditions.
158
1596.33 Mon Nov 18 11:26:27 CET 2013
160 - do not crash when freeing padlists with holes (in 5.18).
161 - tentative SVt_BIND 5.19 port/fix.
162
1636.32 Tue Nov 5 15:35:35 CET 2013
164 - use a new algorithm to derive padlists for perl 5.18. The old
165 one could lead to 0-pointer accesses inside perl (reported
166 by Darin McBride).
167
1686.31 Thu May 9 07:39:48 CEST 2013
169 - Coro::AIO requests would crash if the thread was ready'd
170 while the request was ongoing.
171
1726.29 Wed May 8 02:55:18 CEST 2013
173 - when an on_destroy handler destructs the coro currently being
174 destructed a perl scalar could be accessed after being freed,
175 likely causing a crash.
176
1776.28 Wed Mar 6 06:58:02 CET 2013
178 - clean remnants of existing __DIE__ and __WARN__ handlers so
179 they lose their magic and will not cause segfaults later
180 (testcase by Andrey Sagulin).
181 - improved Coro::State documentation a bit.
182 - Coro::Debug::command now flushes the output.
183 - add hack detection code for x32 abi, because the braindead slugs
184 who designed that made it look exactly like x86_64 without
185 providing proper compile time symbols to test for it. as a result,
186 this detection cannot work reliably.
187 - valgrind stack registering was broken.
188 - do not rely on Time::HiRes anymore in Coro::Debug.
189
1906.23 Fri Dec 7 23:36:37 CET 2012
191 - use experimental fiber implementation on native windows
192 perls.
193 - use sizeof (void *) as multiplication factor for stack sizes,
194 to accomodate the totally braindamaged microsoft 64 bit "os".
195 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
196 activeperl 5.16 crashes when PerlIO_define_layer is called due
197 to some bug in the perl dll, strawberry perl at least passes
198 the testsuite.
199 - implement Coro::Handle->peeraddr/host/port, for slightly
200 improved compatibility with LWP.
201 - implement 5.17 compatibility by almost blindly applying a
202 good-looking patch by Father Chrysostomos.
203 - move stack management functions into libcoro 3.
204 - libcoro version 3 "released".
205 - support magic values as timed_io_once args.
206 - recommend AnyEvent 7+ or EV 4+, also require EV
207 version 4 or newer for Coro::EV.
208
2096.10 Tue Oct 9 01:14:27 CEST 2012
210 - updated ecb.h, it had a typo that caused it to not compile on many
211 big endian systems (reported by many people).
212 - disable memory fences in ecb.h to improve portability.
213
2146.09 Sat Oct 6 23:25:02 CEST 2012
215 - Coro::EV I/O watchers were not interruptible by exceptions
216 (Coro::State::throw) (testcase by sten).
217 - ->throw now puts threads into the ready queue, as this seems to
218 be expected by existing code, and code that doesn't cope with spurious
219 wakeups needs fixing anyway.
220 - use fd -1 in mmap.
221 - cast I32 to int in error message printf.
222 - warn about broken so-called "hardened" kernels.
223
2246.08 Fri Apr 13 12:05:47 CEST 2012
225 - be more aggressive about exiting like perl does - formerly,
226 exiting from the non-main thread would not execute END blocks.
227
2286.07 Fri Nov 11 21:21:48 CET 2011
229 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
230 - update ecb.h.
231
2326.06 Mon Aug 8 23:59:48 CEST 2011
233 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
234 backend, which is an order of magnitude slower. unfortunately, cygwins
235 pthread implementation isn't very complete either, so allocate the stack
236 twice just to be sure.
237 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
238 in its ithreaded perl - best recompile cygwin and use the 'w'indows
239 backend for much better performance. also disable ithreads for
240 even better performance...).
241
2426.05 Thu Aug 4 21:36:36 CEST 2011
243 - blush, condvar values would not be propagated from send to recv anymore
244 (reported by Chip Salzenberg).
245 - use exponential increase for the readline buffer length in
246 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
247
2486.04 Wed Aug 3 17:15:45 CEST 2011
249 - use even more efficient and more compatible condvars for
250 compatibility to AnyEvent 6.x :)
251 - more inconsequential ecb.h updates.
252
2536.03 Wed Aug 3 11:41:30 CEST 2011
254 - change how Coro patches AnyEvent condvars for compatibility to
255 AnyEvent 6.x.
256 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
257 hasn't been included.
258
2596.02 Wed Jul 13 04:35:19 CEST 2011
260 - "improve portability to Gentoo" - gentoo manages to put perl variables
261 in memory areas that are farther than 2gb apart, which the jit couldn't
262 handle and barfed. now it's just a bit slower on gentoo and similar
263 systems.
264
2656.01 Sun Jul 3 12:31:14 CEST 2011
266 - workarounds are good, but the test for whether pthreads are used
267 was not good. this one should be better.
268 - check differently whether gcc generates cfi instructions itself.
269
2706.0 Wed Jun 29 19:43:35 CEST 2011
271 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
272 destroyed and cleaned up automatically (e.g. async { schedule }).
273 - implement a JIT compiler for part of the thread switch code,
274 which gives a 50% speed improvement on threaded perls, and
275 about 4% on non-threaded perls (so threaded perls now finally
276 reach about half the speed of non-threaded perls).
277 - slightly modernise Coro::Intro, add section about rouse functions.
278 - avoid DEFSV and ERRSV, giving another 10% improvement
279 in thread switching.
280 - Coro::State->is_destroyed is now called is_zombie.
281 - implement a Coro->safe_cancel method that might fail, but
282 cancels in a "safer" way if it succeeds.
283 - add preliminary support for DEBUGGING perls.
284 - get rid of two hash-accesses when initialising a new Coro - this
285 speeds up coro creation by almost a factor of two.
286 - croak when a coro that is being cancelled tries to block
287 (e.g. while executing a guard block), instead of crashing or
288 deadlocking.
289 - use a more robust and also faster method to identify Coro::State
290 objects - speeds up everything a bit.
291 - implement Coro->cancel in XS for a 20% speed improvement, and to
292 be able to implement mutual cancellation.
293 - speed up context switches by a percent or two by more efficiently
294 allocating context stack entries.
295 - implement Coro->join and Coro->on_destroy in XS for a speedup and
296 a reduction in memory use.
297 - cancelling a coro while it itself is cancelling another coro is
298 now supported and working, instead of triggering an assertion.
299 - be a bit more crash-resistant when calling (buggy) on_destroy
300 callbacks (best effort).
301 - move on_destroy into the slf_frame, to allow extension slf
302 functions to have destructors.
303 - get rid if coro refcounting - simply crash in other interpreter
304 threads by nulling the pointers on clone.
305 - simplify warn/die hook handling when loading Coro - the convoluted
306 logic seems to be no longer neccessary.
307 - use libecb instead of our own home-grown gcc hacks.
308 - document alternatives to Coro::LWP. Please use them :)
309 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
310 sigaltstack. Really. wine suffers from it, erlang suffers from it,
311 and it's known since at least 2006.
312
3135.372 Wed Feb 23 06:14:30 CET 2011
314 - apparently mingw doesn't provide a working gettimeofday, try to
315 work around that by relying on Time::HiRes (indirectly brought to
316 my attention by Max Maischein).
317 - fix some portability issues when Time::HiRes was used.
318
3195.371 Mon Feb 21 14:36:08 CET 2011
320 - backport to windows process emulation code again.
321
3225.37 Sat Feb 19 07:49:44 CET 2011
323 - add a big "Coro thread life cycle" section to "man Coro".
324 - try a tentative workaround against the breakage that 5.13 has
325 introduced without depreciation period. sigh.
326 - no longer use Time::HiRes if gettimeofday is available, which
327 saves quite a lot of memory.
328
3295.36 Sun Feb 13 05:33:41 CET 2011
330 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
331 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
332 method call.
333 - undocument Coro::Timer::sleep and obsolete whole module.
334 - optimise Coro::Timer::timeout memory and cpu usage.
335 - slightly updated Coro::Intro for recent changes.
336 - do not initialise PL_dirty anymore.
337
3385.25 Thu Nov 11 01:08:39 CET 2010
339 - try a different approach on netbsd - netbsd 5 finally has marginally
340 working pthreads, but still broken ucontext/sigaltstack.
341 - openbsd 4.8 finally got their act together, Coro works out of the box
342 with asm, setjmp and pthreads (no change, just informational).
343
3445.24 Sat Oct 23 11:27:12 CEST 2010
345 - port to the EV 4.0 API.
346 - work around bugs in mingw32, making strawberry perl work
347 out of the box.
348 - correctly modify Coro::AIO function prototypes
349 so that they reflect the "no optional parameters" rule.
350 - "ported" libcoro to C++.
351
3525.23 Mon May 17 18:50:42 CEST 2010
353 - be more resistant to ordering changes when initialising
354 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
355 Waldorf).
356 - document that perl 5.12 deliberately removed support for cloning.
357
3585.22 Wed Apr 14 03:55:35 CEST 2010
359 - correctly return udnef on errors in Coro::Handle::read/write
360 (testcase by Marc Mims).
361 - convert Coro::Util into a "perl compatibility wrapper" - the functions
362 are less useful now, but are drop-in replacements for existing
363 functions, listing better alternatives in the documentation. This also
364 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
365 with Coro::Util::inet_aton.
366 - do not override $Coro::idle unconditionally in Coro.pm, as other
367 modules could have provided their own idle coro already
368 (for exmaple, Coro::AnyEvent).
369 - fix Coro::Util::gethost* functions.
370 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
371
3725.21 Wed Dec 16 07:19:51 CET 2009
373 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
374 together.
375 - add some examples on how to combine other event loops with Coro in
376 Coro::AnyEvent, and how to run it (and not to block). Seems to be
377 the most common source of confusion.
378 - try to catch people naively blocking in an event callback.
379 - work around the perl filehandle bug issue in conjunction with
380 older common::sense (as indirectly pointed out by ZSystem).
381 - clarify the "not from signal handlers" section.
382
3835.2 Sun Oct 4 14:54:24 CEST 2009
384 - Coro::Storable destroyed the prototypes of the functions it wrapped.
385 - export rouse_cb and rouse_wait by default now.
386 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
387 - new method $state->swap_sv.
388 - added section on "windows process emulation" to the manpage,
389 after a not-so-fruitful (nor-friendly) "discussion" with chip
390 salzenberg (discussion implies arguments, but his only arguments
391 were ad-hominems, one wonders why he started it in the first
392 place). I hope this explains it well enough for him to understand,
393 and maybe well enough for others to understand.
394 - use common::sense everywhere now.
395 - idle callbacks are no longer supported, use idle coros instead.
396 - print a thread listing when a deadlock is detected.
397
3985.17 Sat Aug 22 23:09:31 CEST 2009
399 - work around a bug in the perl debugger causing crashes
400 when running under the debugger by marking _pool_handler as nodebug.
401 - speed up Coro::async considerably.
402 - try some hacks to get netbsd to work "more often" - their broken
403 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
404 - convert Coro to AE 5.0 API.
405
4065.162 Tue Jul 28 04:04:03 CEST 2009
407 - perl 5.8.2 is now minimum requirement.
408 - skip t/19_handle.t on broken windows perls.
409
4105.161 Wed Jul 22 04:47:38 CEST 2009
411 - Coro::AnyEvent::poll could have a different prototype when EV was
412 used as backend (analyzed by Tatsuhiko Miyagawa).
413 - Coro::AnyEvent errornously initialised the event loop when loaded,
414 not on demand.
415 - try to workaround rare */t/01_unblock.t failures.
416
4175.16 Tue Jul 21 01:44:37 CEST 2009
418 - Coro::AnyEvent failed to hook into the event loop
419 when no threads had been readied between detecting
420 the event loop and actually running it.
421 - considerably speed up Coro::Select by taking avdantage
422 of AnyEvent > 4.8 and some other optimisations.
423 - implement paragraph readline mode in Coro::Handle
424 (based on patches by Zsbán Ambrus).
425 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
426 and analyzed by Yasuhiro MATSUMOTO).
427 - clarified libcoro license and copyright.
428 - someone stole my EXTRA_META!!!
429 - implement Coro::Select::patch_pp_sselect and it's brother,
430 for hardcode select overriding.
431
4325.151 Mon Jul 6 05:41:57 CEST 2009
433 - backport to windows process emulation code again (patch by
434 Yasuhiro MATSUMOTO).
435 - slightly update Coro::MakeMaker.
436
4375.15 Tue Jun 30 10:28:06 CEST 2009
438 - deprecate Coro::Socket, document how to get ipv6 support via
439 AnyEvent::Socket instead.
440 - implement signal->wait ($cb) interface, similar to semaphores.
441 - work around SvOK not supporting getmagic, so we have to getmagic
442 to test for undef :/ (reported by Matthias Waldorf).
443 - load Coro::AnyEvent in all modules using AnyEvent.
444 - work around perl corrupting our internal data structures,
445 reported by Tokuhiro Matsuno.
446 - enable per-coroutine real and cpu time gathering
447 (Coro::State::enable_times).
448
4495.14 Wed Jun 24 01:37:48 CEST 2009
450 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
451 as this seems to have been a difficult concept (poll, sleep, idle,
452 idle_upto).
453 - add Coro::AnyEvent::readable/writable functions.
454 - clarify Coro::EV/Event/AnyEvent manpages.
455 - free per-thread global scalars in the thread calling ->cancel, to
456 avoid crashes when $_, $@ etc., are magical but some of those
457 had already been freed.
458 - "unexperimentalise" the callback interface for Coro::Semaphore.
459 - speed up ready queue management/context switching by using a linked
460 list instead of an array (~5-10%).
461 - implement "watch" command in Coro::Debug shells.
462 - for fun, implement time-slicing as an example in the manpage.
463 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
464 Coro::Event, use the normal AnyEvent handling instead of dieing
465 (the same is true for Coro::Handle).
466 - properly document Coro::EV::timed_io_once.
467 - avoid unneccessary ->cancel calls in Coro::Handle.
468 - maybe make it work on mingw32 with win32 backend
469 (based on patch by Yasuhiro Matsumoto).
470
4715.132 Fri May 29 09:00:39 CEST 2009
472 - do not keep a reference to the argument itself in
473 Coro::Semaphore::guard, as it could change later.
474 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
475 should support a prepare callback.
476
4775.131 Mon Mar 16 23:20:37 CET 2009
478 - implement and document Coro->suspend, Coro->resume.
479 - fix Coro::Select implementation to not (often) close
480 the passed file descriptors (testcase provided by pippijn).
481
4825.13 Mon Dec 15 21:51:42 CET 2008
483 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
484 - don't set diehook to C<undef> but instead to NULL, to avoid
485 spurious warnings.
486 - fix a lot of bugs in Coro::SemaphoreSet.
487 - Coro::SemaphoreSet will less often create a semaphore needlessly.
488 - add Coro::SemaphoreSet::count and wait methods.
489 - take advantage of the new Guard module.
490 - deprecate Coro::guard.
491 - try to fix the dreaded 01_unblock tests once more. I hate it when
492 testsuites need more fixing than the code they are supposed to test.
493 - croak in more cases when a required callback isn't resolvable.
494 - fix some minor issues in Coro::State->call/eval.
495 - use current coroutine context instead of a temporary one
496 when temporarily switching to another coroutine.
497 - do not call C-level on_destroy handlers during global destruction,
498 to avoid needless segfaults.
499
5005.12 Sun Dec 7 13:30:38 CET 2008
501 - add default config for MirOS, which seems to be a bug-to-bug
502 compatible fork of openbsd ("world domination by releasing
503 openbsd cvs before the openbsd folks do it" or so :).
504 - free_padlist did destroy the names pad, not good, but didn't
505 seem to bother perl - this could fix issues such as eval ""
506 inside a function called from multiple coroutines.
507 - use a different method to detect destruction time.
508 - be more careful when freeing padlists just before global
509 destruction.
510 - fixed and expanded the call/cc example.
511 - renamed _terminate to _coro_run.
512 - new method Coro::Channel->shutdown.
513 - try pthreads on openbsd <4.4 (broken sigaltstack, will
514 pthreads fare better?).
515 - be less picky about destroying "a" running coroutine.
516
5175.11 Tue Nov 25 21:49:05 CET 2008
518 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
519 working code with failed assertions, introducing more bugs than
520 it fixes, requiring elaborate workarounds :(
521
5225.1 Mon Nov 24 08:54:59 CET 2008
523 - wrote a small introductory tutorial - Coro::Intro.
524 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
525 - Coro::Select did errornously dup the file descriptors
526 and didn't work with all AnyEvent backends.
527 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
528 LWP data transfers.
529 - disassociate c contexts from coro objects - this is agruably more
530 correct, but mostly allows sharing of cctxs between coro and state
531 objects, for added memory savings and speed increases.
532 - bumped $Coro::POOL_RSS up to 32kb by default.
533 - no longer set the optype to OP_CUSTOM, as B::* understandably
534 doesn't like this very much (and we *are* a type of entersub).
535 - implement state cloning, just to prove that call/cc can be done.
536 - automatically load Coro::AnyEvent in Coro::Handle.
537 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
538 do not have this method (and don't need it either).
539 - speed up generic anyevent methods in Coro::Handle by using rouse
540 callbacks.
541 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
542 others. It also makes the debugger happier, as you can trace
543 through the idle threads now.
544 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
545 - eg/event was pretty broken.
546 - better 5.8.6 compatibility.
547
5485.0 Thu Nov 20 10:35:05 CET 2008
549 - NEW ARCHITECTURE: use the latest 4.x version if you experience
550 stability issues.
13 - bump API version to 7 - all dependents must be recompiled. 551 - bump API version to 7 - all dependents must be recompiled.
552 - removed timed_* functions - they were not being used anyways
553 and should be replaced by a more generic mechanism -
554 and were annoying to support anyways :)
555 - removed SemaphoreSet's waiter method - use sem method instead.
556 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
557 - async_pool did free a scalar value twice
558 ("Attempt to unreference...").
559 - fix a longstanding bug where calling terminate on a coro that
560 was waiting for a semaphore that was just becoming available
561 would cause a deadlock (semaphore would get into a state where
562 it was available but waiters were still blocked).
563 - calling throw on a coroutine that is waiting for a semaphore will
564 no longer make it acquire the semaphore (and thus leak a count).
14 - perl's process emulation is now not even theoretically supported 565 - perl's process emulation is now not even theoretically supported
15 anymore. 566 anymore.
567 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
568 conversion of callback-style to blocking-style.
569 - new methods $coro->schedule_to and ->cede_to, to specifically
570 schedule or cede to a specific coroutine.
571 - new function Coro::Semaphore::wait.
572 - use named constants in Coro::Channel (Richard Hundt).
16 - directly patch the entersub opcode calling SLF functions (cede, 573 - directly patch the entersub opcode calling SLF functions (cede,
17 transfer and so on). this does speed up context switching, but 574 transfer and so on). this does speed up context switching, but
18 more importanly, it frees us from the hardcoded behaviour of 575 more importanly, it frees us from the hardcoded behaviour of
19 entersub, so we might actually be able to return something from 576 entersub, so we might actually be able to return something from
20 those functions and atcually create new ones. 577 those functions and atcually create new ones.
21 - take advantage of __builtin_frame_address on gcc. 578 - take advantage of __builtin_frame_address on gcc.
22 - expose THX in coroapi (not sure whether this was a wise decision, 579 - expose THX in coroapi (not sure whether this was a wise decision,
23 as "threaded" perls are running at half speed anyways). 580 as "threaded" perls are running at half speed anyways).
24 - implement execute_slf (schedule-like-function) interface that makes 581 - implement execute_slf (schedule-like-function) interface that makes
25 it possible to implement schedule-like-functions in XS. 582 it possible to implement schedule-like-functions in XS.
583 - use new SLF interface to massively speed up Coro::EV by roughly a
584 factor of two.
585 - used new SLF interface to massively speed up Coro::Semaphore by a
586 factor of three.
587 - used new SLF interface to speed up Coro::AIO by roughly a factor of
588 four and reduce its memory usage considerably.
589 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
590 for a nice speedup and vastly more correct behaviour. Also implement
591 a new method "sem" to get at the underlying semaphore object.
592 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
593 (in comparison) 20-40% speedup.
594 - used new SLF interface to reimplement Coro::Signal gaining
595 some unknown (because I was too lazy), but certain, speedup, and also
596 making signals reliable for the first time.
597 - used new SLF interface and other optimisations to speed up async_pool
598 by a factor of two. It also doesn't rely on perl's exception mechanism
599 to exit anymore. The overhead for terminating an async_pool, coro over
600 a normal async is now very small.
601 - sped up coroutine creation/destruction by 40%.
602 - forgot to include Coro/libcoro/README in the dist for all these years.
603 - work around a freebsd pthreads bug (manual testcancel is required as
604 pthread_cond_wait isn't a cancellation point on freebsd).
605 - use new rouse functions to speed up and simplify Coro::BDB.
606 - make "prefer perl native functions" work with threaded perls.
607 - condense Coro::Debug ps output, hint at v and w flags.
608 - (libcoro) lots of minor cleanups and portability improvements.
609
6104.914 Wed Nov 19 12:54:18 CET 2008
611 - fix a disastrous bug in the readline optimisation
612 introduced in 4.801.
613
6144.913 Sat Nov 15 07:58:28 CET 2008
615 - async_pool did free a scalar value twice
616 ("Attempt to unreference...").
26 617
274.912 Thu Nov 13 18:31:23 CET 2008 6184.912 Thu Nov 13 18:31:23 CET 2008
28 - minor cleanups. 619 - minor cleanups.
29 - use much larger stacks on linux and perl < 5.8.8. 620 - use much larger stacks on linux and perl < 5.8.8.
30 - Coro::Debug::new_unix_server did not unlink the socket 621 - Coro::Debug::new_unix_server did not unlink the socket
… …
43 - ->throw is now supported on Coro::State objects. 634 - ->throw is now supported on Coro::State objects.
44 - clean up cctx creation code a bit. 635 - clean up cctx creation code a bit.
45 - entersub is actually a UNOP, not a LOGOP (not a bugfix). 636 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
46 637
474.9 Sat Nov 8 17:45:27 CET 2008 6384.9 Sat Nov 8 17:45:27 CET 2008
48 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 639 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
49 - (libcoro) no longer leak threads in the experimental pthread backend, 640 - (libcoro) no longer leak threads in the experimental pthread backend,
50 also speed it up considerably. 641 also speed it up considerably.
51 - (libcoro) do not rely on makecontext passing void *'s unscathed. 642 - (libcoro) do not rely on makecontext passing void *'s unscathed.
52 - fix compiletime dependencies on libcoro in the Makefile. 643 - fix compiletime dependencies on libcoro in the Makefile.
53 - cctx_count wasn't always updated properly. 644 - cctx_count wasn't always updated properly.
… …
115 - remove debugging code related to MgPV_nolen_const, also try to 706 - remove debugging code related to MgPV_nolen_const, also try to
116 make it compile with perl 5.8.6 (yes, apple apparently loves 707 make it compile with perl 5.8.6 (yes, apple apparently loves
117 outdated software). Reported by John S. 708 outdated software). Reported by John S.
118 709
1194.744 Tue Jul 8 22:06:35 CEST 2008 7104.744 Tue Jul 8 22:06:35 CEST 2008
120 - correctly provide default DIE/WARN handlers as documented. 711 - correctly provide default DIE/WARN handlers as documented.
121 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 712 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
122 implementations inside perl work fine for us. 713 implementations inside perl work fine for us.
123 714
1244.743 Mon Jun 16 00:21:57 CEST 2008 7154.743 Mon Jun 16 00:21:57 CEST 2008
125 - when using Coro::EV without running EV::loop it could 716 - when using Coro::EV without running EV::loop it could
… …
147 - sprinkle "no warnings" freely over everything, also suppress 738 - sprinkle "no warnings" freely over everything, also suppress
148 warnings for some other modules. 739 warnings for some other modules.
149 - fix typo in WSAEWOULDBLOCK. 740 - fix typo in WSAEWOULDBLOCK.
150 741
1514.72 Sun May 25 05:14:36 CEST 2008 7424.72 Sun May 25 05:14:36 CEST 2008
152 - tweak META.yaml a bit, unfortunately, there is no documented way 743 - tweak META.yaml a bit, unfortunately, there is no documented way
153 to have optional dependencies with CPAN. doh :( 744 to have optional dependencies with CPAN. doh :(
154 - avoid running some tests on windows because they would fail due to 745 - avoid running some tests on windows because they would fail due to
155 perl bug (broken fork, broken pipes...). 746 perl bug (broken fork, broken pipes...).
156 - work around perl on windows bugs where perl returns undocumented 747 - work around perl on windows bugs where perl returns undocumented
157 error codes for sysread, syswrite etc. by taking advantage 748 error codes for sysread, syswrite etc. by taking advantage
… …
161 - use unix domain sockets in testsuite to work around 752 - use unix domain sockets in testsuite to work around
162 common perl implementation bugs on widows (they are emulated by 753 common perl implementation bugs on widows (they are emulated by
163 tcp sockets on windows. ugh.) 754 tcp sockets on windows. ugh.)
164 755
1654.71 Sat May 24 20:01:27 CEST 2008 7564.71 Sat May 24 20:01:27 CEST 2008
166 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 757 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
167 - take advantage of async name resolution of AnyEvent::Util. 758 - take advantage of async name resolution of AnyEvent::Util.
168 - work around brutal inet_aton override in Coro::LWP. 759 - work around brutal inet_aton override in Coro::LWP.
169 - take advantage of the readyhook in Coro::EV, for smoother 760 - take advantage of the readyhook in Coro::EV, for smoother
170 scheduling. 761 scheduling.
171 762
1724.7 Sun May 11 00:32:19 CEST 2008 7634.7 Sun May 11 00:32:19 CEST 2008
173 - completely reworked the Coro manpage. 764 - completely reworked the Coro manpage.
174 - added Coro::AnyEvent, generic event loop integration. 765 - added Coro::AnyEvent, generic event loop integration.
175 - implement cancel, ready and kill commands in Coro::Debug. 766 - implement cancel, ready and kill commands in Coro::Debug.
176 - document find_coro in Coro::Debug. 767 - document find_coro in Coro::Debug.
177 - incompatible change: rename has_stack to has_cctx. 768 - incompatible change: rename has_stack to has_cctx.
178 - Coro::AIO and Coro::BDB no longer force event model detection, 769 - Coro::AIO and Coro::BDB no longer force event model detection,
179 use AnyEvent::AIO and AnyEvent::BDB. 770 use AnyEvent::AIO and AnyEvent::BDB.
… …
303 894
3044.11 Thu Oct 11 02:40:24 CEST 2007 8954.11 Thu Oct 11 02:40:24 CEST 2007
305 - port to threaded perls. 896 - port to threaded perls.
306 897
3074.1 Thu Oct 11 02:38:16 CEST 2007 8984.1 Thu Oct 11 02:38:16 CEST 2007
308 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 899 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
309 be local to each coro (see Coro::State). 900 be local to each coro (see Coro::State).
310 - incompatible change: for very deep reasons, cede and cede_notself 901 - incompatible change: for very deep reasons, cede and cede_notself
311 cannot return anything, so nothing will be returned. 902 cannot return anything, so nothing will be returned.
312 - possibly bring back 5.10 compatibility (untested). 903 - possibly bring back 5.10 compatibility (untested).
313 - work around stupid (and wrong) warning on 5.10 :(. 904 - work around stupid (and wrong) warning on 5.10 :(.
… …
385 but stateful semantics. 976 but stateful semantics.
386 - fixed a lot of typos in Coro.pm (patch submitted by David 977 - fixed a lot of typos in Coro.pm (patch submitted by David
387 Steinbrunner, which applied flawlessly). 978 Steinbrunner, which applied flawlessly).
388 979
3893.6 Sat Apr 14 17:13:31 CEST 2007 9803.6 Sat Apr 14 17:13:31 CEST 2007
390 - added some bugfixes to get eg/myhttpd working again. 981 - added some bugfixes to get eg/myhttpd working again.
391 - added Coro::Storable for often-cede'ing freeze/thaw. 982 - added Coro::Storable for often-cede'ing freeze/thaw.
392 - try to do a clean exit when a coroutine calls exit 983 - try to do a clean exit when a coroutine calls exit
393 (EXPERIMENTAL). 984 (EXPERIMENTAL).
394 - got rid of indirect call through _coro_init. 985 - got rid of indirect call through _coro_init.
395 - updated the partly antique examples in eg/ to 986 - updated the partly antique examples in eg/ to

Diff Legend

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