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

Comparing Coro/Changes (file contents):
Revision 1.403 by root, Tue Nov 11 03:30:21 2008 UTC vs.
Revision 1.688 by root, Sat Jul 25 22:19:30 2020 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: 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
136.57
14 - override AnyEvent condvar methods "properly", as AnyEvent
15 has support for pluggable condvars since 2011...
16 - allow rouse_wait to be called in a different thread.
17 - slightly clarify rouse_cb/rouse_wait documentation.
18 - slightly improve rouse callback performance and memory usage.
19 - try to report correct calller when throwing exception in
20 on $condvar->recv, not a useless line in Coro::AnyEvent.
21
226.56 Thu Feb 20 19:40:07 CET 2020
23 - Coro::EV::time*once functions used ev_once, which is not cancelable.
24 This could cause problems due to I/O watchers not being stopped up
25 on exceptions/cancellation.
26
276.55 Fri Jul 19 17:28:09 CEST 2019
28 - Coro::RWLock woke up only one reader instead of all after a wrlock
29 (analyzed and patch by Maxim Skorobogatov).
30 - slightly improved documentation of Coro::RWLock and RWLock testcase.
31 - quickly (re)-ported to minix 3.3 before minix crashed again.
32
336.54 Sun Dec 16 11:06:07 CET 2018
34 - reinstate Coro::Select::select prototype - it breaks more than it fixes.
35
366.53 Sun Dec 16 10:33:15 CET 2018
37 - the autogenerated prototypes of Coro::AIO now add ";" at the end
38 to avoid generating named unary operators and messing up parsing
39 (mostly affects aio_mlockall 4 | 8 which was parsed as
40 (aio_mlockall 4) | 8).
41 - do not use a prototype for select in Coro::Select, as CORE::select also
42 doesn't have one.
43
446.52 Tue Aug 14 18:49:43 CEST 2018
45 - libcoro update: allow sharing of coro threads using the pthreads
46 backend among different pthreads, as is required by Coro::Multicore.
47 - always allow ->call and ->eval when the coro is the current coroutine. this
48 makes it possible, among other things, to call these on $Coro::main.
49 - minor code cleanups.
50
516.514 Thu Aug 31 18:25:31 CEST 2017
52 - safe_cancel no longer croaks when called on already-destroyed
53 thread, but instead "just works".
54
556.513 Sat Jul 15 01:18:51 CEST 2017
56 - experimental 5.22 comaptibility changes (PadlistNAMES).
57 - removed attributes example and made sure the other examples
58 still work.
59 - fix some statevar types for newer perls, should be of no consequence.
60
616.512 Fri Jul 14 05:03:17 CEST 2017
62 - seems to pass test suite down till 5.8.9, so make this the
63 minimum requirement in Makefile.PL.
64 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
65 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
66 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
67 fortified longjmp variants in some libcs.
68 - re-try JIT allocation without PROT_EXEC, to hopefully improve
69 portability (still doesn't seem to work around broken selinux
70 mmap everywhere).
71 - speling fixes (Xavier Guimard).
72
736.511 Sun Jun 26 23:44:50 CEST 2016
74 - make it compile with threaded perls again.
75 - simplify and speed up __DIE__ and __WARN__ handling.
76
776.51 Sat Jun 25 20:57:20 CEST 2016
78 - 6.5 release didn't compile on most older perls due to
79 a logic error enabling experimental 5.24 code for all versions.
80 - since various broken or deficient patches float around
81 to work around the vtbl API breakage in perl, let's provide
82 our own workaround, which seems to be much less invasive
83 and more compatible than the approaches seen so far. YMMV.
84 - work around assertion failure in perl_destruct on 5.24
85 (analyzed by Dave Mitchell).
86
876.5 Wed Jun 22 22:23:50 CEST 2016
88 - swap_sv swapping order was not symmetrical, causing
89 wrong swaps when swap_sv was used multiple times on the
90 same sv in the same thread.
91 - swap_sv calls can now be undone by calling it again
92 with the same variables.
93 - swap_sv calls will now be undone in async_pool threads.
94 - split Coro::Semaphore::up/adjust into separate xs functions
95 for better error reporting, at slight codesize increase.
96 - (libcoro) arm assembly support, please test and report.
97 - adjust to PL_savestack changes in perl 5.24 (adapted from
98 the debian patch, which unfortunately gets it wrong).
99
1006.49 Sat Oct 17 01:40:12 CEST 2015
101 - throwing an exception to a thread waiting in
102 Coro::Handle using Coro::EV did not stop the watchers,
103 causing the next call to fail (testcase by Martin Pritchard).
104 - bump minimum perl version to 5.10.
105
1066.48 Sun Oct 4 19:03:51 CEST 2015
107 - fix memory leak when Coro::Handle uses Coro::EV internally
108 (testcase by Sten Sten).
109 - update code to libev 4 API (internally, Coro still used the
110 version 3 API calls).
111 - remove unused hv_sig.
112
1136.47 Sat Jul 11 03:58:20 CEST 2015
114 - swap_svs - NVs can be in the SV head beginning with 5.20.
115
1166.46 Tue Jun 30 14:36:30 CEST 2015
117 - restore portability to perls with windows fork emulation
118 (patch by Petr Písař).
119
1206.45 Tue Jun 30 01:40:08 CEST 2015
121 - restore portability to perl 5.16 and below.
122
1236.44 Tue Jun 30 00:41:54 CEST 2015
124 - avoid segfaulting (or worse) when tracing a canceled thread.
125 - the Event module silently broke it's public hook API, causing
126 failures on perls compiled with -Duselongdouble. Requiring
127 the latest version of Event because I don't know which release
128 changed this (it's not mentioned in the ChangeLog).
129 - allow xs level enterleave hooks via CoroAPI.
130
1316.43 Thu Jun 4 15:38:14 CEST 2015
132 - use stability canary.
133 - port to stableperl-5.22.0-1.001.
134 - update libecb.
135
1366.42 Wed Feb 11 20:29:52 CET 2015
137 - Coro::SemaphoreSet->try did not actually work (analyzed by
138 SATO Kentaro).
139 - upgrade libecb to be C11 compliant.
140
1416.41 Sat Sep 6 22:08:46 CEST 2014
142 - restore portability to perl 5.8.x.
143 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
144 up some memory, but fixes "use feature" and similar modules
145 when used inside a Coro.
146 - allow zero as argument to Coro::Channel to mean the same thing
147 as no argument. this works with older versions as well,
148 but wasn't legal until now.
149 - slightly better c header file detection.
150
1516.39 Mon Jun 2 00:00:08 CEST 2014
152 - work around more incompatible changes in 5.20.
153
1546.38 Sun Jun 1 21:54:23 CEST 2014
155 - check that perl slots actually have enough space to hold
156 interpreter variables.
157 - untested port to perl 5.19 (64 bit tmps indices) (reported
158 by Andreas König).
159 - croak when cancel is called without a thread context.
160
1616.37 Tue Mar 4 13:27:33 CET 2014
162 - *sigh*, removed leftover debugging code from debugging a
163 perl bug, of all things.
164
1656.36 Tue Mar 4 07:11:59 CET 2014
166 - semaphores would not clear the destroy hook when interrupted
167 by ->throw, causing segfaults or worse.
168 - ->throw on a thread waiting for a semaphore did not acquire
169 the semaphore, but also didn't wake up other waiters,
170 possibly causing a deadlock.
171 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
172 to report a full stacktrace, which should help find out
173 where the actual call is.
174 - "atomically" destroy data in slf_destroy, because it is
175 the right thing to do, just in case.
176 - disable ecb.h libm dependency, to compile on ancient systems
177 or under adverse conditions.
178
1796.33 Mon Nov 18 11:26:27 CET 2013
180 - do not crash when freeing padlists with holes (in 5.18).
181 - tentative SVt_BIND 5.19 port/fix.
182
1836.32 Tue Nov 5 15:35:35 CET 2013
184 - use a new algorithm to derive padlists for perl 5.18. The old
185 one could lead to 0-pointer accesses inside perl (reported
186 by Darin McBride).
187
1886.31 Thu May 9 07:39:48 CEST 2013
189 - Coro::AIO requests would crash if the thread was ready'd
190 while the request was ongoing.
191
1926.29 Wed May 8 02:55:18 CEST 2013
193 - when an on_destroy handler destructs the coro currently being
194 destructed a perl scalar could be accessed after being freed,
195 likely causing a crash.
196
1976.28 Wed Mar 6 06:58:02 CET 2013
198 - clean remnants of existing __DIE__ and __WARN__ handlers so
199 they lose their magic and will not cause segfaults later
200 (testcase by Andrey Sagulin).
201 - improved Coro::State documentation a bit.
202 - Coro::Debug::command now flushes the output.
203 - add hack detection code for x32 abi, because the braindead slugs
204 who designed that made it look exactly like x86_64 without
205 providing proper compile time symbols to test for it. as a result,
206 this detection cannot work reliably.
207 - valgrind stack registering was broken.
208 - do not rely on Time::HiRes anymore in Coro::Debug.
209
2106.23 Fri Dec 7 23:36:37 CET 2012
211 - use experimental fiber implementation on native windows
212 perls.
213 - use sizeof (void *) as multiplication factor for stack sizes,
214 to accomodate the totally braindamaged microsoft 64 bit "os".
215 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
216 activeperl 5.16 crashes when PerlIO_define_layer is called due
217 to some bug in the perl dll, strawberry perl at least passes
218 the testsuite.
219 - implement Coro::Handle->peeraddr/host/port, for slightly
220 improved compatibility with LWP.
221 - implement 5.17 compatibility by almost blindly applying a
222 good-looking patch by Father Chrysostomos.
223 - move stack management functions into libcoro 3.
224 - libcoro version 3 "released".
225 - support magic values as timed_io_once args.
226 - recommend AnyEvent 7+ or EV 4+, also require EV
227 version 4 or newer for Coro::EV.
228
2296.10 Tue Oct 9 01:14:27 CEST 2012
230 - updated ecb.h, it had a typo that caused it to not compile on many
231 big endian systems (reported by many people).
232 - disable memory fences in ecb.h to improve portability.
233
2346.09 Sat Oct 6 23:25:02 CEST 2012
235 - Coro::EV I/O watchers were not interruptible by exceptions
236 (Coro::State::throw) (testcase by sten).
237 - ->throw now puts threads into the ready queue, as this seems to
238 be expected by existing code, and code that doesn't cope with spurious
239 wakeups needs fixing anyway.
240 - use fd -1 in mmap.
241 - cast I32 to int in error message printf.
242 - warn about broken so-called "hardened" kernels.
243
2446.08 Fri Apr 13 12:05:47 CEST 2012
245 - be more aggressive about exiting like perl does - formerly,
246 exiting from the non-main thread would not execute END blocks.
247
2486.07 Fri Nov 11 21:21:48 CET 2011
249 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
250 - update ecb.h.
251
2526.06 Mon Aug 8 23:59:48 CEST 2011
253 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
254 backend, which is an order of magnitude slower. unfortunately, cygwins
255 pthread implementation isn't very complete either, so allocate the stack
256 twice just to be sure.
257 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
258 in its ithreaded perl - best recompile cygwin and use the 'w'indows
259 backend for much better performance. also disable ithreads for
260 even better performance...).
261
2626.05 Thu Aug 4 21:36:36 CEST 2011
263 - blush, condvar values would not be propagated from send to recv anymore
264 (reported by Chip Salzenberg).
265 - use exponential increase for the readline buffer length in
266 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
267
2686.04 Wed Aug 3 17:15:45 CEST 2011
269 - use even more efficient and more compatible condvars for
270 compatibility to AnyEvent 6.x :)
271 - more inconsequential ecb.h updates.
272
2736.03 Wed Aug 3 11:41:30 CEST 2011
274 - change how Coro patches AnyEvent condvars for compatibility to
275 AnyEvent 6.x.
276 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
277 hasn't been included.
278
2796.02 Wed Jul 13 04:35:19 CEST 2011
280 - "improve portability to Gentoo" - gentoo manages to put perl variables
281 in memory areas that are farther than 2gb apart, which the jit couldn't
282 handle and barfed. now it's just a bit slower on gentoo and similar
283 systems.
284
2856.01 Sun Jul 3 12:31:14 CEST 2011
286 - workarounds are good, but the test for whether pthreads are used
287 was not good. this one should be better.
288 - check differently whether gcc generates cfi instructions itself.
289
2906.0 Wed Jun 29 19:43:35 CEST 2011
291 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
292 destroyed and cleaned up automatically (e.g. async { schedule }).
293 - implement a JIT compiler for part of the thread switch code,
294 which gives a 50% speed improvement on threaded perls, and
295 about 4% on non-threaded perls (so threaded perls now finally
296 reach about half the speed of non-threaded perls).
297 - slightly modernise Coro::Intro, add section about rouse functions.
298 - avoid DEFSV and ERRSV, giving another 10% improvement
299 in thread switching.
300 - Coro::State->is_destroyed is now called is_zombie.
301 - implement a Coro->safe_cancel method that might fail, but
302 cancels in a "safer" way if it succeeds.
303 - add preliminary support for DEBUGGING perls.
304 - get rid of two hash-accesses when initialising a new Coro - this
305 speeds up coro creation by almost a factor of two.
306 - croak when a coro that is being cancelled tries to block
307 (e.g. while executing a guard block), instead of crashing or
308 deadlocking.
309 - use a more robust and also faster method to identify Coro::State
310 objects - speeds up everything a bit.
311 - implement Coro->cancel in XS for a 20% speed improvement, and to
312 be able to implement mutual cancellation.
313 - speed up context switches by a percent or two by more efficiently
314 allocating context stack entries.
315 - implement Coro->join and Coro->on_destroy in XS for a speedup and
316 a reduction in memory use.
317 - cancelling a coro while it itself is cancelling another coro is
318 now supported and working, instead of triggering an assertion.
319 - be a bit more crash-resistant when calling (buggy) on_destroy
320 callbacks (best effort).
321 - move on_destroy into the slf_frame, to allow extension slf
322 functions to have destructors.
323 - get rid if coro refcounting - simply crash in other interpreter
324 threads by nulling the pointers on clone.
325 - simplify warn/die hook handling when loading Coro - the convoluted
326 logic seems to be no longer neccessary.
327 - use libecb instead of our own home-grown gcc hacks.
328 - document alternatives to Coro::LWP. Please use them :)
329 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
330 sigaltstack. Really. wine suffers from it, erlang suffers from it,
331 and it's known since at least 2006.
332
3335.372 Wed Feb 23 06:14:30 CET 2011
334 - apparently mingw doesn't provide a working gettimeofday, try to
335 work around that by relying on Time::HiRes (indirectly brought to
336 my attention by Max Maischein).
337 - fix some portability issues when Time::HiRes was used.
338
3395.371 Mon Feb 21 14:36:08 CET 2011
340 - backport to windows process emulation code again.
341
3425.37 Sat Feb 19 07:49:44 CET 2011
343 - add a big "Coro thread life cycle" section to "man Coro".
344 - try a tentative workaround against the breakage that 5.13 has
345 introduced without depreciation period. sigh.
346 - no longer use Time::HiRes if gettimeofday is available, which
347 saves quite a lot of memory.
348
3495.36 Sun Feb 13 05:33:41 CET 2011
350 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
351 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
352 method call.
353 - undocument Coro::Timer::sleep and obsolete whole module.
354 - optimise Coro::Timer::timeout memory and cpu usage.
355 - slightly updated Coro::Intro for recent changes.
356 - do not initialise PL_dirty anymore.
357
3585.25 Thu Nov 11 01:08:39 CET 2010
359 - try a different approach on netbsd - netbsd 5 finally has marginally
360 working pthreads, but still broken ucontext/sigaltstack.
361 - openbsd 4.8 finally got their act together, Coro works out of the box
362 with asm, setjmp and pthreads (no change, just informational).
363
3645.24 Sat Oct 23 11:27:12 CEST 2010
365 - port to the EV 4.0 API.
366 - work around bugs in mingw32, making strawberry perl work
367 out of the box.
368 - correctly modify Coro::AIO function prototypes
369 so that they reflect the "no optional parameters" rule.
370 - "ported" libcoro to C++.
371
3725.23 Mon May 17 18:50:42 CEST 2010
373 - be more resistant to ordering changes when initialising
374 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
375 Waldorf).
376 - document that perl 5.12 deliberately removed support for cloning.
377
3785.22 Wed Apr 14 03:55:35 CEST 2010
379 - correctly return udnef on errors in Coro::Handle::read/write
380 (testcase by Marc Mims).
381 - convert Coro::Util into a "perl compatibility wrapper" - the functions
382 are less useful now, but are drop-in replacements for existing
383 functions, listing better alternatives in the documentation. This also
384 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
385 with Coro::Util::inet_aton.
386 - do not override $Coro::idle unconditionally in Coro.pm, as other
387 modules could have provided their own idle coro already
388 (for exmaple, Coro::AnyEvent).
389 - fix Coro::Util::gethost* functions.
390 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
391
3925.21 Wed Dec 16 07:19:51 CET 2009
393 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
394 together.
395 - add some examples on how to combine other event loops with Coro in
396 Coro::AnyEvent, and how to run it (and not to block). Seems to be
397 the most common source of confusion.
398 - try to catch people naively blocking in an event callback.
399 - work around the perl filehandle bug issue in conjunction with
400 older common::sense (as indirectly pointed out by ZSystem).
401 - clarify the "not from signal handlers" section.
402
4035.2 Sun Oct 4 14:54:24 CEST 2009
404 - Coro::Storable destroyed the prototypes of the functions it wrapped.
405 - export rouse_cb and rouse_wait by default now.
406 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
407 - new method $state->swap_sv.
408 - added section on "windows process emulation" to the manpage,
409 after a not-so-fruitful (nor-friendly) "discussion" with chip
410 salzenberg (discussion implies arguments, but his only arguments
411 were ad-hominems, one wonders why he started it in the first
412 place). I hope this explains it well enough for him to understand,
413 and maybe well enough for others to understand.
414 - use common::sense everywhere now.
415 - idle callbacks are no longer supported, use idle coros instead.
416 - print a thread listing when a deadlock is detected.
417
4185.17 Sat Aug 22 23:09:31 CEST 2009
419 - work around a bug in the perl debugger causing crashes
420 when running under the debugger by marking _pool_handler as nodebug.
421 - speed up Coro::async considerably.
422 - try some hacks to get netbsd to work "more often" - their broken
423 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
424 - convert Coro to AE 5.0 API.
425
4265.162 Tue Jul 28 04:04:03 CEST 2009
427 - perl 5.8.2 is now minimum requirement.
428 - skip t/19_handle.t on broken windows perls.
429
4305.161 Wed Jul 22 04:47:38 CEST 2009
431 - Coro::AnyEvent::poll could have a different prototype when EV was
432 used as backend (analyzed by Tatsuhiko Miyagawa).
433 - Coro::AnyEvent errornously initialised the event loop when loaded,
434 not on demand.
435 - try to workaround rare */t/01_unblock.t failures.
436
4375.16 Tue Jul 21 01:44:37 CEST 2009
438 - Coro::AnyEvent failed to hook into the event loop
439 when no threads had been readied between detecting
440 the event loop and actually running it.
441 - considerably speed up Coro::Select by taking avdantage
442 of AnyEvent > 4.8 and some other optimisations.
443 - implement paragraph readline mode in Coro::Handle
444 (based on patches by Zsbán Ambrus).
445 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
446 and analyzed by Yasuhiro MATSUMOTO).
447 - clarified libcoro license and copyright.
448 - someone stole my EXTRA_META!!!
449 - implement Coro::Select::patch_pp_sselect and it's brother,
450 for hardcode select overriding.
451
4525.151 Mon Jul 6 05:41:57 CEST 2009
453 - backport to windows process emulation code again (patch by
454 Yasuhiro MATSUMOTO).
455 - slightly update Coro::MakeMaker.
456
4575.15 Tue Jun 30 10:28:06 CEST 2009
458 - deprecate Coro::Socket, document how to get ipv6 support via
459 AnyEvent::Socket instead.
460 - implement signal->wait ($cb) interface, similar to semaphores.
461 - work around SvOK not supporting getmagic, so we have to getmagic
462 to test for undef :/ (reported by Matthias Waldorf).
463 - load Coro::AnyEvent in all modules using AnyEvent.
464 - work around perl corrupting our internal data structures,
465 reported by Tokuhiro Matsuno.
466 - enable per-coroutine real and cpu time gathering
467 (Coro::State::enable_times).
468
4695.14 Wed Jun 24 01:37:48 CEST 2009
470 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
471 as this seems to have been a difficult concept (poll, sleep, idle,
472 idle_upto).
473 - add Coro::AnyEvent::readable/writable functions.
474 - clarify Coro::EV/Event/AnyEvent manpages.
475 - free per-thread global scalars in the thread calling ->cancel, to
476 avoid crashes when $_, $@ etc., are magical but some of those
477 had already been freed.
478 - "unexperimentalise" the callback interface for Coro::Semaphore.
479 - speed up ready queue management/context switching by using a linked
480 list instead of an array (~5-10%).
481 - implement "watch" command in Coro::Debug shells.
482 - for fun, implement time-slicing as an example in the manpage.
483 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
484 Coro::Event, use the normal AnyEvent handling instead of dieing
485 (the same is true for Coro::Handle).
486 - properly document Coro::EV::timed_io_once.
487 - avoid unneccessary ->cancel calls in Coro::Handle.
488 - maybe make it work on mingw32 with win32 backend
489 (based on patch by Yasuhiro Matsumoto).
490
4915.132 Fri May 29 09:00:39 CEST 2009
492 - do not keep a reference to the argument itself in
493 Coro::Semaphore::guard, as it could change later.
494 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
495 should support a prepare callback.
496
4975.131 Mon Mar 16 23:20:37 CET 2009
498 - implement and document Coro->suspend, Coro->resume.
499 - fix Coro::Select implementation to not (often) close
500 the passed file descriptors (testcase provided by pippijn).
501
5025.13 Mon Dec 15 21:51:42 CET 2008
503 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
504 - don't set diehook to C<undef> but instead to NULL, to avoid
505 spurious warnings.
506 - fix a lot of bugs in Coro::SemaphoreSet.
507 - Coro::SemaphoreSet will less often create a semaphore needlessly.
508 - add Coro::SemaphoreSet::count and wait methods.
509 - take advantage of the new Guard module.
510 - deprecate Coro::guard.
511 - try to fix the dreaded 01_unblock tests once more. I hate it when
512 testsuites need more fixing than the code they are supposed to test.
513 - croak in more cases when a required callback isn't resolvable.
514 - fix some minor issues in Coro::State->call/eval.
515 - use current coroutine context instead of a temporary one
516 when temporarily switching to another coroutine.
517 - do not call C-level on_destroy handlers during global destruction,
518 to avoid needless segfaults.
519
5205.12 Sun Dec 7 13:30:38 CET 2008
521 - add default config for MirOS, which seems to be a bug-to-bug
522 compatible fork of openbsd ("world domination by releasing
523 openbsd cvs before the openbsd folks do it" or so :).
524 - free_padlist did destroy the names pad, not good, but didn't
525 seem to bother perl - this could fix issues such as eval ""
526 inside a function called from multiple coroutines.
527 - use a different method to detect destruction time.
528 - be more careful when freeing padlists just before global
529 destruction.
530 - fixed and expanded the call/cc example.
531 - renamed _terminate to _coro_run.
532 - new method Coro::Channel->shutdown.
533 - try pthreads on openbsd <4.4 (broken sigaltstack, will
534 pthreads fare better?).
535 - be less picky about destroying "a" running coroutine.
536
5375.11 Tue Nov 25 21:49:05 CET 2008
538 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
539 working code with failed assertions, introducing more bugs than
540 it fixes, requiring elaborate workarounds :(
541
5425.1 Mon Nov 24 08:54:59 CET 2008
543 - wrote a small introductory tutorial - Coro::Intro.
544 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
545 - Coro::Select did errornously dup the file descriptors
546 and didn't work with all AnyEvent backends.
547 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
548 LWP data transfers.
549 - disassociate c contexts from coro objects - this is agruably more
550 correct, but mostly allows sharing of cctxs between coro and state
551 objects, for added memory savings and speed increases.
552 - bumped $Coro::POOL_RSS up to 32kb by default.
553 - no longer set the optype to OP_CUSTOM, as B::* understandably
554 doesn't like this very much (and we *are* a type of entersub).
555 - implement state cloning, just to prove that call/cc can be done.
556 - automatically load Coro::AnyEvent in Coro::Handle.
557 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
558 do not have this method (and don't need it either).
559 - speed up generic anyevent methods in Coro::Handle by using rouse
560 callbacks.
561 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
562 others. It also makes the debugger happier, as you can trace
563 through the idle threads now.
564 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
565 - eg/event was pretty broken.
566 - better 5.8.6 compatibility.
567
5685.0 Thu Nov 20 10:35:05 CET 2008
569 - NEW ARCHITECTURE: use the latest 4.x version if you experience
570 stability issues.
571 - bump API version to 7 - all dependents must be recompiled.
572 - removed timed_* functions - they were not being used anyways
573 and should be replaced by a more generic mechanism -
574 and were annoying to support anyways :)
575 - removed SemaphoreSet's waiter method - use sem method instead.
576 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
577 - async_pool did free a scalar value twice
578 ("Attempt to unreference...").
579 - fix a longstanding bug where calling terminate on a coro that
580 was waiting for a semaphore that was just becoming available
581 would cause a deadlock (semaphore would get into a state where
582 it was available but waiters were still blocked).
583 - calling throw on a coroutine that is waiting for a semaphore will
584 no longer make it acquire the semaphore (and thus leak a count).
585 - perl's process emulation is now not even theoretically supported
586 anymore.
587 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
588 conversion of callback-style to blocking-style.
589 - new methods $coro->schedule_to and ->cede_to, to specifically
590 schedule or cede to a specific coroutine.
591 - new function Coro::Semaphore::wait.
592 - use named constants in Coro::Channel (Richard Hundt).
593 - directly patch the entersub opcode calling SLF functions (cede,
594 transfer and so on). this does speed up context switching, but
595 more importanly, it frees us from the hardcoded behaviour of
596 entersub, so we might actually be able to return something from
597 those functions and atcually create new ones.
598 - take advantage of __builtin_frame_address on gcc.
599 - expose THX in coroapi (not sure whether this was a wise decision,
600 as "threaded" perls are running at half speed anyways).
601 - implement execute_slf (schedule-like-function) interface that makes
602 it possible to implement schedule-like-functions in XS.
603 - use new SLF interface to massively speed up Coro::EV by roughly a
604 factor of two.
605 - used new SLF interface to massively speed up Coro::Semaphore by a
606 factor of three.
607 - used new SLF interface to speed up Coro::AIO by roughly a factor of
608 four and reduce its memory usage considerably.
609 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
610 for a nice speedup and vastly more correct behaviour. Also implement
611 a new method "sem" to get at the underlying semaphore object.
612 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
613 (in comparison) 20-40% speedup.
614 - used new SLF interface to reimplement Coro::Signal gaining
615 some unknown (because I was too lazy), but certain, speedup, and also
616 making signals reliable for the first time.
617 - used new SLF interface and other optimisations to speed up async_pool
618 by a factor of two. It also doesn't rely on perl's exception mechanism
619 to exit anymore. The overhead for terminating an async_pool, coro over
620 a normal async is now very small.
621 - sped up coroutine creation/destruction by 40%.
622 - forgot to include Coro/libcoro/README in the dist for all these years.
623 - work around a freebsd pthreads bug (manual testcancel is required as
624 pthread_cond_wait isn't a cancellation point on freebsd).
625 - use new rouse functions to speed up and simplify Coro::BDB.
626 - make "prefer perl native functions" work with threaded perls.
627 - condense Coro::Debug ps output, hint at v and w flags.
628 - (libcoro) lots of minor cleanups and portability improvements.
629
6304.914 Wed Nov 19 12:54:18 CET 2008
631 - fix a disastrous bug in the readline optimisation
632 introduced in 4.801.
633
6344.913 Sat Nov 15 07:58:28 CET 2008
635 - async_pool did free a scalar value twice
636 ("Attempt to unreference...").
637
6384.912 Thu Nov 13 18:31:23 CET 2008
639 - minor cleanups.
640 - use much larger stacks on linux and perl < 5.8.8.
641 - Coro::Debug::new_unix_server did not unlink the socket
642 when destroyed.
7 643
84.911 Tue Nov 11 04:26:17 CET 2008 6444.911 Tue Nov 11 04:26:17 CET 2008
9 - "port" to threaded perls. 645 - "port" to threaded perls.
10 646
114.91 Mon Nov 10 05:36:38 CET 2008 6474.91 Mon Nov 10 05:36:38 CET 2008
… …
18 - ->throw is now supported on Coro::State objects. 654 - ->throw is now supported on Coro::State objects.
19 - clean up cctx creation code a bit. 655 - clean up cctx creation code a bit.
20 - entersub is actually a UNOP, not a LOGOP (not a bugfix). 656 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
21 657
224.9 Sat Nov 8 17:45:27 CET 2008 6584.9 Sat Nov 8 17:45:27 CET 2008
23 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 659 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
24 - (libcoro) no longer leak threads in the experimental pthread backend, 660 - (libcoro) no longer leak threads in the experimental pthread backend,
25 also speed it up considerably. 661 also speed it up considerably.
26 - (libcoro) do not rely on makecontext passing void *'s unscathed. 662 - (libcoro) do not rely on makecontext passing void *'s unscathed.
27 - fix compiletime dependencies on libcoro in the Makefile. 663 - fix compiletime dependencies on libcoro in the Makefile.
28 - cctx_count wasn't always updated properly. 664 - cctx_count wasn't always updated properly.
… …
90 - remove debugging code related to MgPV_nolen_const, also try to 726 - remove debugging code related to MgPV_nolen_const, also try to
91 make it compile with perl 5.8.6 (yes, apple apparently loves 727 make it compile with perl 5.8.6 (yes, apple apparently loves
92 outdated software). Reported by John S. 728 outdated software). Reported by John S.
93 729
944.744 Tue Jul 8 22:06:35 CEST 2008 7304.744 Tue Jul 8 22:06:35 CEST 2008
95 - correctly provide default DIE/WARN handlers as documented. 731 - correctly provide default DIE/WARN handlers as documented.
96 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 732 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
97 implementations inside perl work fine for us. 733 implementations inside perl work fine for us.
98 734
994.743 Mon Jun 16 00:21:57 CEST 2008 7354.743 Mon Jun 16 00:21:57 CEST 2008
100 - when using Coro::EV without running EV::loop it could 736 - when using Coro::EV without running EV::loop it could
… …
122 - sprinkle "no warnings" freely over everything, also suppress 758 - sprinkle "no warnings" freely over everything, also suppress
123 warnings for some other modules. 759 warnings for some other modules.
124 - fix typo in WSAEWOULDBLOCK. 760 - fix typo in WSAEWOULDBLOCK.
125 761
1264.72 Sun May 25 05:14:36 CEST 2008 7624.72 Sun May 25 05:14:36 CEST 2008
127 - tweak META.yaml a bit, unfortunately, there is no documented way 763 - tweak META.yaml a bit, unfortunately, there is no documented way
128 to have optional dependencies with CPAN. doh :( 764 to have optional dependencies with CPAN. doh :(
129 - avoid running some tests on windows because they would fail due to 765 - avoid running some tests on windows because they would fail due to
130 perl bug (broken fork, broken pipes...). 766 perl bug (broken fork, broken pipes...).
131 - work around perl on windows bugs where perl returns undocumented 767 - work around perl on windows bugs where perl returns undocumented
132 error codes for sysread, syswrite etc. by taking advantage 768 error codes for sysread, syswrite etc. by taking advantage
… …
136 - use unix domain sockets in testsuite to work around 772 - use unix domain sockets in testsuite to work around
137 common perl implementation bugs on widows (they are emulated by 773 common perl implementation bugs on widows (they are emulated by
138 tcp sockets on windows. ugh.) 774 tcp sockets on windows. ugh.)
139 775
1404.71 Sat May 24 20:01:27 CEST 2008 7764.71 Sat May 24 20:01:27 CEST 2008
141 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 777 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
142 - take advantage of async name resolution of AnyEvent::Util. 778 - take advantage of async name resolution of AnyEvent::Util.
143 - work around brutal inet_aton override in Coro::LWP. 779 - work around brutal inet_aton override in Coro::LWP.
144 - take advantage of the readyhook in Coro::EV, for smoother 780 - take advantage of the readyhook in Coro::EV, for smoother
145 scheduling. 781 scheduling.
146 782
1474.7 Sun May 11 00:32:19 CEST 2008 7834.7 Sun May 11 00:32:19 CEST 2008
148 - completely reworked the Coro manpage. 784 - completely reworked the Coro manpage.
149 - added Coro::AnyEvent, generic event loop integration. 785 - added Coro::AnyEvent, generic event loop integration.
150 - implement cancel, ready and kill commands in Coro::Debug. 786 - implement cancel, ready and kill commands in Coro::Debug.
151 - document find_coro in Coro::Debug. 787 - document find_coro in Coro::Debug.
152 - incompatible change: rename has_stack to has_cctx. 788 - incompatible change: rename has_stack to has_cctx.
153 - Coro::AIO and Coro::BDB no longer force event model detection, 789 - Coro::AIO and Coro::BDB no longer force event model detection,
154 use AnyEvent::AIO and AnyEvent::BDB. 790 use AnyEvent::AIO and AnyEvent::BDB.
… …
278 914
2794.11 Thu Oct 11 02:40:24 CEST 2007 9154.11 Thu Oct 11 02:40:24 CEST 2007
280 - port to threaded perls. 916 - port to threaded perls.
281 917
2824.1 Thu Oct 11 02:38:16 CEST 2007 9184.1 Thu Oct 11 02:38:16 CEST 2007
283 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 919 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
284 be local to each coro (see Coro::State). 920 be local to each coro (see Coro::State).
285 - incompatible change: for very deep reasons, cede and cede_notself 921 - incompatible change: for very deep reasons, cede and cede_notself
286 cannot return anything, so nothing will be returned. 922 cannot return anything, so nothing will be returned.
287 - possibly bring back 5.10 compatibility (untested). 923 - possibly bring back 5.10 compatibility (untested).
288 - work around stupid (and wrong) warning on 5.10 :(. 924 - work around stupid (and wrong) warning on 5.10 :(.
… …
360 but stateful semantics. 996 but stateful semantics.
361 - fixed a lot of typos in Coro.pm (patch submitted by David 997 - fixed a lot of typos in Coro.pm (patch submitted by David
362 Steinbrunner, which applied flawlessly). 998 Steinbrunner, which applied flawlessly).
363 999
3643.6 Sat Apr 14 17:13:31 CEST 2007 10003.6 Sat Apr 14 17:13:31 CEST 2007
365 - added some bugfixes to get eg/myhttpd working again. 1001 - added some bugfixes to get eg/myhttpd working again.
366 - added Coro::Storable for often-cede'ing freeze/thaw. 1002 - added Coro::Storable for often-cede'ing freeze/thaw.
367 - try to do a clean exit when a coroutine calls exit 1003 - try to do a clean exit when a coroutine calls exit
368 (EXPERIMENTAL). 1004 (EXPERIMENTAL).
369 - got rid of indirect call through _coro_init. 1005 - got rid of indirect call through _coro_init.
370 - updated the partly antique examples in eg/ to 1006 - updated the partly antique examples in eg/ to

Diff Legend

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