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

Comparing Coro/Changes (file contents):
Revision 1.439 by root, Wed Nov 19 10:44:41 2008 UTC vs.
Revision 1.686 by root, Mon Mar 16 11:12:52 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?
7 10
8TODO: testcancel 11TODO: State.xs could short-circuit some macros such as CORO_EXECUTE_SLF
9TODO: guard as SLF - to avoid deadlocks. 12TODO: use frame->destroy to improve Coro::EV
10 - avoid deadlock in Coro::Channel if maxsize == 1 (Richard Hundt). 136.57
14 - allow rouse_wait to be called in a different thread.
15 - slightly clarify rouse_cb/rouse_wait documentation.
16 - slightly improve rouse callback performance and memory usage.
11 17
125.0 186.56 Thu Feb 20 19:40:07 CET 2020
19 - Coro::EV::time*once functions used ev_once, which is not cancelable.
20 This could cause problems due to I/O watchers not being stopped up
21 on exceptions/cancellation.
22
236.55 Fri Jul 19 17:28:09 CEST 2019
24 - Coro::RWLock woke up only one reader instead of all after a wrlock
25 (analyzed and patch by Maxim Skorobogatov).
26 - slightly improved documentation of Coro::RWLock and RWLock testcase.
27 - quickly (re)-ported to minix 3.3 before minix crashed again.
28
296.54 Sun Dec 16 11:06:07 CET 2018
30 - reinstate Coro::Select::select prototype - it breaks more than it fixes.
31
326.53 Sun Dec 16 10:33:15 CET 2018
33 - the autogenerated prototypes of Coro::AIO now add ";" at the end
34 to avoid generating named unary operators and messing up parsing
35 (mostly affects aio_mlockall 4 | 8 which was parsed as
36 (aio_mlockall 4) | 8).
37 - do not use a prototype for select in Coro::Select, as CORE::select also
38 doesn't have one.
39
406.52 Tue Aug 14 18:49:43 CEST 2018
41 - libcoro update: allow sharing of coro threads using the pthreads
42 backend among different pthreads, as is required by Coro::Multicore.
43 - always allow ->call and ->eval when the coro is the current coroutine. this
44 makes it possible, among other things, to call these on $Coro::main.
45 - minor code cleanups.
46
476.514 Thu Aug 31 18:25:31 CEST 2017
48 - safe_cancel no longer croaks when called on already-destroyed
49 thread, but instead "just works".
50
516.513 Sat Jul 15 01:18:51 CEST 2017
52 - experimental 5.22 comaptibility changes (PadlistNAMES).
53 - removed attributes example and made sure the other examples
54 still work.
55 - fix some statevar types for newer perls, should be of no consequence.
56
576.512 Fri Jul 14 05:03:17 CEST 2017
58 - seems to pass test suite down till 5.8.9, so make this the
59 minimum requirement in Makefile.PL.
60 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
61 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
62 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
63 fortified longjmp variants in some libcs.
64 - re-try JIT allocation without PROT_EXEC, to hopefully improve
65 portability (still doesn't seem to work around broken selinux
66 mmap everywhere).
67 - speling fixes (Xavier Guimard).
68
696.511 Sun Jun 26 23:44:50 CEST 2016
70 - make it compile with threaded perls again.
71 - simplify and speed up __DIE__ and __WARN__ handling.
72
736.51 Sat Jun 25 20:57:20 CEST 2016
74 - 6.5 release didn't compile on most older perls due to
75 a logic error enabling experimental 5.24 code for all versions.
76 - since various broken or deficient patches float around
77 to work around the vtbl API breakage in perl, let's provide
78 our own workaround, which seems to be much less invasive
79 and more compatible than the approaches seen so far. YMMV.
80 - work around assertion failure in perl_destruct on 5.24
81 (analyzed by Dave Mitchell).
82
836.5 Wed Jun 22 22:23:50 CEST 2016
84 - swap_sv swapping order was not symmetrical, causing
85 wrong swaps when swap_sv was used multiple times on the
86 same sv in the same thread.
87 - swap_sv calls can now be undone by calling it again
88 with the same variables.
89 - swap_sv calls will now be undone in async_pool threads.
90 - split Coro::Semaphore::up/adjust into separate xs functions
91 for better error reporting, at slight codesize increase.
92 - (libcoro) arm assembly support, please test and report.
93 - adjust to PL_savestack changes in perl 5.24 (adapted from
94 the debian patch, which unfortunately gets it wrong).
95
966.49 Sat Oct 17 01:40:12 CEST 2015
97 - throwing an exception to a thread waiting in
98 Coro::Handle using Coro::EV did not stop the watchers,
99 causing the next call to fail (testcase by Martin Pritchard).
100 - bump minimum perl version to 5.10.
101
1026.48 Sun Oct 4 19:03:51 CEST 2015
103 - fix memory leak when Coro::Handle uses Coro::EV internally
104 (testcase by Sten Sten).
105 - update code to libev 4 API (internally, Coro still used the
106 version 3 API calls).
107 - remove unused hv_sig.
108
1096.47 Sat Jul 11 03:58:20 CEST 2015
110 - swap_svs - NVs can be in the SV head beginning with 5.20.
111
1126.46 Tue Jun 30 14:36:30 CEST 2015
113 - restore portability to perls with windows fork emulation
114 (patch by Petr Písař).
115
1166.45 Tue Jun 30 01:40:08 CEST 2015
117 - restore portability to perl 5.16 and below.
118
1196.44 Tue Jun 30 00:41:54 CEST 2015
120 - avoid segfaulting (or worse) when tracing a canceled thread.
121 - the Event module silently broke it's public hook API, causing
122 failures on perls compiled with -Duselongdouble. Requiring
123 the latest version of Event because I don't know which release
124 changed this (it's not mentioned in the ChangeLog).
125 - allow xs level enterleave hooks via CoroAPI.
126
1276.43 Thu Jun 4 15:38:14 CEST 2015
128 - use stability canary.
129 - port to stableperl-5.22.0-1.001.
130 - update libecb.
131
1326.42 Wed Feb 11 20:29:52 CET 2015
133 - Coro::SemaphoreSet->try did not actually work (analyzed by
134 SATO Kentaro).
135 - upgrade libecb to be C11 compliant.
136
1376.41 Sat Sep 6 22:08:46 CEST 2014
138 - restore portability to perl 5.8.x.
139 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
140 up some memory, but fixes "use feature" and similar modules
141 when used inside a Coro.
142 - allow zero as argument to Coro::Channel to mean the same thing
143 as no argument. this works with older versions as well,
144 but wasn't legal until now.
145 - slightly better c header file detection.
146
1476.39 Mon Jun 2 00:00:08 CEST 2014
148 - work around more incompatible changes in 5.20.
149
1506.38 Sun Jun 1 21:54:23 CEST 2014
151 - check that perl slots actually have enough space to hold
152 interpreter variables.
153 - untested port to perl 5.19 (64 bit tmps indices) (reported
154 by Andreas König).
155 - croak when cancel is called without a thread context.
156
1576.37 Tue Mar 4 13:27:33 CET 2014
158 - *sigh*, removed leftover debugging code from debugging a
159 perl bug, of all things.
160
1616.36 Tue Mar 4 07:11:59 CET 2014
162 - semaphores would not clear the destroy hook when interrupted
163 by ->throw, causing segfaults or worse.
164 - ->throw on a thread waiting for a semaphore did not acquire
165 the semaphore, but also didn't wake up other waiters,
166 possibly causing a deadlock.
167 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
168 to report a full stacktrace, which should help find out
169 where the actual call is.
170 - "atomically" destroy data in slf_destroy, because it is
171 the right thing to do, just in case.
172 - disable ecb.h libm dependency, to compile on ancient systems
173 or under adverse conditions.
174
1756.33 Mon Nov 18 11:26:27 CET 2013
176 - do not crash when freeing padlists with holes (in 5.18).
177 - tentative SVt_BIND 5.19 port/fix.
178
1796.32 Tue Nov 5 15:35:35 CET 2013
180 - use a new algorithm to derive padlists for perl 5.18. The old
181 one could lead to 0-pointer accesses inside perl (reported
182 by Darin McBride).
183
1846.31 Thu May 9 07:39:48 CEST 2013
185 - Coro::AIO requests would crash if the thread was ready'd
186 while the request was ongoing.
187
1886.29 Wed May 8 02:55:18 CEST 2013
189 - when an on_destroy handler destructs the coro currently being
190 destructed a perl scalar could be accessed after being freed,
191 likely causing a crash.
192
1936.28 Wed Mar 6 06:58:02 CET 2013
194 - clean remnants of existing __DIE__ and __WARN__ handlers so
195 they lose their magic and will not cause segfaults later
196 (testcase by Andrey Sagulin).
197 - improved Coro::State documentation a bit.
198 - Coro::Debug::command now flushes the output.
199 - add hack detection code for x32 abi, because the braindead slugs
200 who designed that made it look exactly like x86_64 without
201 providing proper compile time symbols to test for it. as a result,
202 this detection cannot work reliably.
203 - valgrind stack registering was broken.
204 - do not rely on Time::HiRes anymore in Coro::Debug.
205
2066.23 Fri Dec 7 23:36:37 CET 2012
207 - use experimental fiber implementation on native windows
208 perls.
209 - use sizeof (void *) as multiplication factor for stack sizes,
210 to accomodate the totally braindamaged microsoft 64 bit "os".
211 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
212 activeperl 5.16 crashes when PerlIO_define_layer is called due
213 to some bug in the perl dll, strawberry perl at least passes
214 the testsuite.
215 - implement Coro::Handle->peeraddr/host/port, for slightly
216 improved compatibility with LWP.
217 - implement 5.17 compatibility by almost blindly applying a
218 good-looking patch by Father Chrysostomos.
219 - move stack management functions into libcoro 3.
220 - libcoro version 3 "released".
221 - support magic values as timed_io_once args.
222 - recommend AnyEvent 7+ or EV 4+, also require EV
223 version 4 or newer for Coro::EV.
224
2256.10 Tue Oct 9 01:14:27 CEST 2012
226 - updated ecb.h, it had a typo that caused it to not compile on many
227 big endian systems (reported by many people).
228 - disable memory fences in ecb.h to improve portability.
229
2306.09 Sat Oct 6 23:25:02 CEST 2012
231 - Coro::EV I/O watchers were not interruptible by exceptions
232 (Coro::State::throw) (testcase by sten).
233 - ->throw now puts threads into the ready queue, as this seems to
234 be expected by existing code, and code that doesn't cope with spurious
235 wakeups needs fixing anyway.
236 - use fd -1 in mmap.
237 - cast I32 to int in error message printf.
238 - warn about broken so-called "hardened" kernels.
239
2406.08 Fri Apr 13 12:05:47 CEST 2012
241 - be more aggressive about exiting like perl does - formerly,
242 exiting from the non-main thread would not execute END blocks.
243
2446.07 Fri Nov 11 21:21:48 CET 2011
245 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
246 - update ecb.h.
247
2486.06 Mon Aug 8 23:59:48 CEST 2011
249 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
250 backend, which is an order of magnitude slower. unfortunately, cygwins
251 pthread implementation isn't very complete either, so allocate the stack
252 twice just to be sure.
253 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
254 in its ithreaded perl - best recompile cygwin and use the 'w'indows
255 backend for much better performance. also disable ithreads for
256 even better performance...).
257
2586.05 Thu Aug 4 21:36:36 CEST 2011
259 - blush, condvar values would not be propagated from send to recv anymore
260 (reported by Chip Salzenberg).
261 - use exponential increase for the readline buffer length in
262 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
263
2646.04 Wed Aug 3 17:15:45 CEST 2011
265 - use even more efficient and more compatible condvars for
266 compatibility to AnyEvent 6.x :)
267 - more inconsequential ecb.h updates.
268
2696.03 Wed Aug 3 11:41:30 CEST 2011
270 - change how Coro patches AnyEvent condvars for compatibility to
271 AnyEvent 6.x.
272 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
273 hasn't been included.
274
2756.02 Wed Jul 13 04:35:19 CEST 2011
276 - "improve portability to Gentoo" - gentoo manages to put perl variables
277 in memory areas that are farther than 2gb apart, which the jit couldn't
278 handle and barfed. now it's just a bit slower on gentoo and similar
279 systems.
280
2816.01 Sun Jul 3 12:31:14 CEST 2011
282 - workarounds are good, but the test for whether pthreads are used
283 was not good. this one should be better.
284 - check differently whether gcc generates cfi instructions itself.
285
2866.0 Wed Jun 29 19:43:35 CEST 2011
287 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
288 destroyed and cleaned up automatically (e.g. async { schedule }).
289 - implement a JIT compiler for part of the thread switch code,
290 which gives a 50% speed improvement on threaded perls, and
291 about 4% on non-threaded perls (so threaded perls now finally
292 reach about half the speed of non-threaded perls).
293 - slightly modernise Coro::Intro, add section about rouse functions.
294 - avoid DEFSV and ERRSV, giving another 10% improvement
295 in thread switching.
296 - Coro::State->is_destroyed is now called is_zombie.
297 - implement a Coro->safe_cancel method that might fail, but
298 cancels in a "safer" way if it succeeds.
299 - add preliminary support for DEBUGGING perls.
300 - get rid of two hash-accesses when initialising a new Coro - this
301 speeds up coro creation by almost a factor of two.
302 - croak when a coro that is being cancelled tries to block
303 (e.g. while executing a guard block), instead of crashing or
304 deadlocking.
305 - use a more robust and also faster method to identify Coro::State
306 objects - speeds up everything a bit.
307 - implement Coro->cancel in XS for a 20% speed improvement, and to
308 be able to implement mutual cancellation.
309 - speed up context switches by a percent or two by more efficiently
310 allocating context stack entries.
311 - implement Coro->join and Coro->on_destroy in XS for a speedup and
312 a reduction in memory use.
313 - cancelling a coro while it itself is cancelling another coro is
314 now supported and working, instead of triggering an assertion.
315 - be a bit more crash-resistant when calling (buggy) on_destroy
316 callbacks (best effort).
317 - move on_destroy into the slf_frame, to allow extension slf
318 functions to have destructors.
319 - get rid if coro refcounting - simply crash in other interpreter
320 threads by nulling the pointers on clone.
321 - simplify warn/die hook handling when loading Coro - the convoluted
322 logic seems to be no longer neccessary.
323 - use libecb instead of our own home-grown gcc hacks.
324 - document alternatives to Coro::LWP. Please use them :)
325 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
326 sigaltstack. Really. wine suffers from it, erlang suffers from it,
327 and it's known since at least 2006.
328
3295.372 Wed Feb 23 06:14:30 CET 2011
330 - apparently mingw doesn't provide a working gettimeofday, try to
331 work around that by relying on Time::HiRes (indirectly brought to
332 my attention by Max Maischein).
333 - fix some portability issues when Time::HiRes was used.
334
3355.371 Mon Feb 21 14:36:08 CET 2011
336 - backport to windows process emulation code again.
337
3385.37 Sat Feb 19 07:49:44 CET 2011
339 - add a big "Coro thread life cycle" section to "man Coro".
340 - try a tentative workaround against the breakage that 5.13 has
341 introduced without depreciation period. sigh.
342 - no longer use Time::HiRes if gettimeofday is available, which
343 saves quite a lot of memory.
344
3455.36 Sun Feb 13 05:33:41 CET 2011
346 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
347 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
348 method call.
349 - undocument Coro::Timer::sleep and obsolete whole module.
350 - optimise Coro::Timer::timeout memory and cpu usage.
351 - slightly updated Coro::Intro for recent changes.
352 - do not initialise PL_dirty anymore.
353
3545.25 Thu Nov 11 01:08:39 CET 2010
355 - try a different approach on netbsd - netbsd 5 finally has marginally
356 working pthreads, but still broken ucontext/sigaltstack.
357 - openbsd 4.8 finally got their act together, Coro works out of the box
358 with asm, setjmp and pthreads (no change, just informational).
359
3605.24 Sat Oct 23 11:27:12 CEST 2010
361 - port to the EV 4.0 API.
362 - work around bugs in mingw32, making strawberry perl work
363 out of the box.
364 - correctly modify Coro::AIO function prototypes
365 so that they reflect the "no optional parameters" rule.
366 - "ported" libcoro to C++.
367
3685.23 Mon May 17 18:50:42 CEST 2010
369 - be more resistant to ordering changes when initialising
370 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
371 Waldorf).
372 - document that perl 5.12 deliberately removed support for cloning.
373
3745.22 Wed Apr 14 03:55:35 CEST 2010
375 - correctly return udnef on errors in Coro::Handle::read/write
376 (testcase by Marc Mims).
377 - convert Coro::Util into a "perl compatibility wrapper" - the functions
378 are less useful now, but are drop-in replacements for existing
379 functions, listing better alternatives in the documentation. This also
380 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
381 with Coro::Util::inet_aton.
382 - do not override $Coro::idle unconditionally in Coro.pm, as other
383 modules could have provided their own idle coro already
384 (for exmaple, Coro::AnyEvent).
385 - fix Coro::Util::gethost* functions.
386 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
387
3885.21 Wed Dec 16 07:19:51 CET 2009
389 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
390 together.
391 - add some examples on how to combine other event loops with Coro in
392 Coro::AnyEvent, and how to run it (and not to block). Seems to be
393 the most common source of confusion.
394 - try to catch people naively blocking in an event callback.
395 - work around the perl filehandle bug issue in conjunction with
396 older common::sense (as indirectly pointed out by ZSystem).
397 - clarify the "not from signal handlers" section.
398
3995.2 Sun Oct 4 14:54:24 CEST 2009
400 - Coro::Storable destroyed the prototypes of the functions it wrapped.
401 - export rouse_cb and rouse_wait by default now.
402 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
403 - new method $state->swap_sv.
404 - added section on "windows process emulation" to the manpage,
405 after a not-so-fruitful (nor-friendly) "discussion" with chip
406 salzenberg (discussion implies arguments, but his only arguments
407 were ad-hominems, one wonders why he started it in the first
408 place). I hope this explains it well enough for him to understand,
409 and maybe well enough for others to understand.
410 - use common::sense everywhere now.
411 - idle callbacks are no longer supported, use idle coros instead.
412 - print a thread listing when a deadlock is detected.
413
4145.17 Sat Aug 22 23:09:31 CEST 2009
415 - work around a bug in the perl debugger causing crashes
416 when running under the debugger by marking _pool_handler as nodebug.
417 - speed up Coro::async considerably.
418 - try some hacks to get netbsd to work "more often" - their broken
419 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
420 - convert Coro to AE 5.0 API.
421
4225.162 Tue Jul 28 04:04:03 CEST 2009
423 - perl 5.8.2 is now minimum requirement.
424 - skip t/19_handle.t on broken windows perls.
425
4265.161 Wed Jul 22 04:47:38 CEST 2009
427 - Coro::AnyEvent::poll could have a different prototype when EV was
428 used as backend (analyzed by Tatsuhiko Miyagawa).
429 - Coro::AnyEvent errornously initialised the event loop when loaded,
430 not on demand.
431 - try to workaround rare */t/01_unblock.t failures.
432
4335.16 Tue Jul 21 01:44:37 CEST 2009
434 - Coro::AnyEvent failed to hook into the event loop
435 when no threads had been readied between detecting
436 the event loop and actually running it.
437 - considerably speed up Coro::Select by taking avdantage
438 of AnyEvent > 4.8 and some other optimisations.
439 - implement paragraph readline mode in Coro::Handle
440 (based on patches by Zsbán Ambrus).
441 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
442 and analyzed by Yasuhiro MATSUMOTO).
443 - clarified libcoro license and copyright.
444 - someone stole my EXTRA_META!!!
445 - implement Coro::Select::patch_pp_sselect and it's brother,
446 for hardcode select overriding.
447
4485.151 Mon Jul 6 05:41:57 CEST 2009
449 - backport to windows process emulation code again (patch by
450 Yasuhiro MATSUMOTO).
451 - slightly update Coro::MakeMaker.
452
4535.15 Tue Jun 30 10:28:06 CEST 2009
454 - deprecate Coro::Socket, document how to get ipv6 support via
455 AnyEvent::Socket instead.
456 - implement signal->wait ($cb) interface, similar to semaphores.
457 - work around SvOK not supporting getmagic, so we have to getmagic
458 to test for undef :/ (reported by Matthias Waldorf).
459 - load Coro::AnyEvent in all modules using AnyEvent.
460 - work around perl corrupting our internal data structures,
461 reported by Tokuhiro Matsuno.
462 - enable per-coroutine real and cpu time gathering
463 (Coro::State::enable_times).
464
4655.14 Wed Jun 24 01:37:48 CEST 2009
466 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
467 as this seems to have been a difficult concept (poll, sleep, idle,
468 idle_upto).
469 - add Coro::AnyEvent::readable/writable functions.
470 - clarify Coro::EV/Event/AnyEvent manpages.
471 - free per-thread global scalars in the thread calling ->cancel, to
472 avoid crashes when $_, $@ etc., are magical but some of those
473 had already been freed.
474 - "unexperimentalise" the callback interface for Coro::Semaphore.
475 - speed up ready queue management/context switching by using a linked
476 list instead of an array (~5-10%).
477 - implement "watch" command in Coro::Debug shells.
478 - for fun, implement time-slicing as an example in the manpage.
479 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
480 Coro::Event, use the normal AnyEvent handling instead of dieing
481 (the same is true for Coro::Handle).
482 - properly document Coro::EV::timed_io_once.
483 - avoid unneccessary ->cancel calls in Coro::Handle.
484 - maybe make it work on mingw32 with win32 backend
485 (based on patch by Yasuhiro Matsumoto).
486
4875.132 Fri May 29 09:00:39 CEST 2009
488 - do not keep a reference to the argument itself in
489 Coro::Semaphore::guard, as it could change later.
490 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
491 should support a prepare callback.
492
4935.131 Mon Mar 16 23:20:37 CET 2009
494 - implement and document Coro->suspend, Coro->resume.
495 - fix Coro::Select implementation to not (often) close
496 the passed file descriptors (testcase provided by pippijn).
497
4985.13 Mon Dec 15 21:51:42 CET 2008
499 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
500 - don't set diehook to C<undef> but instead to NULL, to avoid
501 spurious warnings.
502 - fix a lot of bugs in Coro::SemaphoreSet.
503 - Coro::SemaphoreSet will less often create a semaphore needlessly.
504 - add Coro::SemaphoreSet::count and wait methods.
505 - take advantage of the new Guard module.
506 - deprecate Coro::guard.
507 - try to fix the dreaded 01_unblock tests once more. I hate it when
508 testsuites need more fixing than the code they are supposed to test.
509 - croak in more cases when a required callback isn't resolvable.
510 - fix some minor issues in Coro::State->call/eval.
511 - use current coroutine context instead of a temporary one
512 when temporarily switching to another coroutine.
513 - do not call C-level on_destroy handlers during global destruction,
514 to avoid needless segfaults.
515
5165.12 Sun Dec 7 13:30:38 CET 2008
517 - add default config for MirOS, which seems to be a bug-to-bug
518 compatible fork of openbsd ("world domination by releasing
519 openbsd cvs before the openbsd folks do it" or so :).
520 - free_padlist did destroy the names pad, not good, but didn't
521 seem to bother perl - this could fix issues such as eval ""
522 inside a function called from multiple coroutines.
523 - use a different method to detect destruction time.
524 - be more careful when freeing padlists just before global
525 destruction.
526 - fixed and expanded the call/cc example.
527 - renamed _terminate to _coro_run.
528 - new method Coro::Channel->shutdown.
529 - try pthreads on openbsd <4.4 (broken sigaltstack, will
530 pthreads fare better?).
531 - be less picky about destroying "a" running coroutine.
532
5335.11 Tue Nov 25 21:49:05 CET 2008
534 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
535 working code with failed assertions, introducing more bugs than
536 it fixes, requiring elaborate workarounds :(
537
5385.1 Mon Nov 24 08:54:59 CET 2008
539 - wrote a small introductory tutorial - Coro::Intro.
540 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
541 - Coro::Select did errornously dup the file descriptors
542 and didn't work with all AnyEvent backends.
543 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
544 LWP data transfers.
545 - disassociate c contexts from coro objects - this is agruably more
546 correct, but mostly allows sharing of cctxs between coro and state
547 objects, for added memory savings and speed increases.
548 - bumped $Coro::POOL_RSS up to 32kb by default.
549 - no longer set the optype to OP_CUSTOM, as B::* understandably
550 doesn't like this very much (and we *are* a type of entersub).
551 - implement state cloning, just to prove that call/cc can be done.
552 - automatically load Coro::AnyEvent in Coro::Handle.
553 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
554 do not have this method (and don't need it either).
555 - speed up generic anyevent methods in Coro::Handle by using rouse
556 callbacks.
557 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
558 others. It also makes the debugger happier, as you can trace
559 through the idle threads now.
560 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
561 - eg/event was pretty broken.
562 - better 5.8.6 compatibility.
563
5645.0 Thu Nov 20 10:35:05 CET 2008
13 - NEW ARCHITECTURE: use the latest 4.x version if you experience 565 - NEW ARCHITECTURE: use the latest 4.x version if you experience
14 stability issues. 566 stability issues.
15 - bump API version to 7 - all dependents must be recompiled. 567 - bump API version to 7 - all dependents must be recompiled.
16 - removed timed_* functions - they were not being used anyways 568 - removed timed_* functions - they were not being used anyways
17 and should be replaced by a more generic mechanism - 569 and should be replaced by a more generic mechanism -
18 and were annoying to support anyways :) 570 and were annoying to support anyways :)
19 - removed SemaphoreSet's waiter method - use sem instead. 571 - removed SemaphoreSet's waiter method - use sem method instead.
20 - Coro::Semaphore->adjust didn't correctly wake up enough waiters. 572 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
21 - async_pool did free a scalar value twice 573 - async_pool did free a scalar value twice
22 ("Attempt to unreference..."). 574 ("Attempt to unreference...").
23 - fix a longstanding bug where calling terminate on a coro that 575 - fix a longstanding bug where calling terminate on a coro that
24 was waiting for a semaphore that was just becoming available 576 was waiting for a semaphore that was just becoming available
… …
26 it was available but waiters were still blocked). 578 it was available but waiters were still blocked).
27 - calling throw on a coroutine that is waiting for a semaphore will 579 - calling throw on a coroutine that is waiting for a semaphore will
28 no longer make it acquire the semaphore (and thus leak a count). 580 no longer make it acquire the semaphore (and thus leak a count).
29 - perl's process emulation is now not even theoretically supported 581 - perl's process emulation is now not even theoretically supported
30 anymore. 582 anymore.
31 - new functions Coro::rouse_cb and Coro::rouse_wait. 583 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
584 conversion of callback-style to blocking-style.
585 - new methods $coro->schedule_to and ->cede_to, to specifically
586 schedule or cede to a specific coroutine.
32 - new function Coro::Semaphore::wait. 587 - new function Coro::Semaphore::wait.
33 - use named constants in Coro::Channel (Richard Hundt). 588 - use named constants in Coro::Channel (Richard Hundt).
34 - directly patch the entersub opcode calling SLF functions (cede, 589 - directly patch the entersub opcode calling SLF functions (cede,
35 transfer and so on). this does speed up context switching, but 590 transfer and so on). this does speed up context switching, but
36 more importanly, it frees us from the hardcoded behaviour of 591 more importanly, it frees us from the hardcoded behaviour of
… …
53 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate 608 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
54 (in comparison) 20-40% speedup. 609 (in comparison) 20-40% speedup.
55 - used new SLF interface to reimplement Coro::Signal gaining 610 - used new SLF interface to reimplement Coro::Signal gaining
56 some unknown (because I was too lazy), but certain, speedup, and also 611 some unknown (because I was too lazy), but certain, speedup, and also
57 making signals reliable for the first time. 612 making signals reliable for the first time.
613 - used new SLF interface and other optimisations to speed up async_pool
614 by a factor of two. It also doesn't rely on perl's exception mechanism
615 to exit anymore. The overhead for terminating an async_pool, coro over
616 a normal async is now very small.
617 - sped up coroutine creation/destruction by 40%.
58 - forgot to include Coro/libcoro/README in the dist for all these years. 618 - forgot to include Coro/libcoro/README in the dist for all these years.
59 - work around a freebsd pthreads bug (manual testcancel is required as 619 - work around a freebsd pthreads bug (manual testcancel is required as
60 pthread_cond_wait isn't a cancellation point on freebsd). 620 pthread_cond_wait isn't a cancellation point on freebsd).
61 - use new rouse functions to speed up and simplify Coro::BDB. 621 - use new rouse functions to speed up and simplify Coro::BDB.
62 - make "prefer perl native functions" work with threaded perls. 622 - make "prefer perl native functions" work with threaded perls.
623 - condense Coro::Debug ps output, hint at v and w flags.
63 - (libcoro) lots of minor cleanups and portability improvements. 624 - (libcoro) lots of minor cleanups and portability improvements.
625
6264.914 Wed Nov 19 12:54:18 CET 2008
627 - fix a disastrous bug in the readline optimisation
628 introduced in 4.801.
64 629
654.913 Sat Nov 15 07:58:28 CET 2008 6304.913 Sat Nov 15 07:58:28 CET 2008
66 - async_pool did free a scalar value twice 631 - async_pool did free a scalar value twice
67 ("Attempt to unreference..."). 632 ("Attempt to unreference...").
68 633
… …
85 - ->throw is now supported on Coro::State objects. 650 - ->throw is now supported on Coro::State objects.
86 - clean up cctx creation code a bit. 651 - clean up cctx creation code a bit.
87 - entersub is actually a UNOP, not a LOGOP (not a bugfix). 652 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
88 653
894.9 Sat Nov 8 17:45:27 CET 2008 6544.9 Sat Nov 8 17:45:27 CET 2008
90 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 655 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
91 - (libcoro) no longer leak threads in the experimental pthread backend, 656 - (libcoro) no longer leak threads in the experimental pthread backend,
92 also speed it up considerably. 657 also speed it up considerably.
93 - (libcoro) do not rely on makecontext passing void *'s unscathed. 658 - (libcoro) do not rely on makecontext passing void *'s unscathed.
94 - fix compiletime dependencies on libcoro in the Makefile. 659 - fix compiletime dependencies on libcoro in the Makefile.
95 - cctx_count wasn't always updated properly. 660 - cctx_count wasn't always updated properly.
… …
157 - remove debugging code related to MgPV_nolen_const, also try to 722 - remove debugging code related to MgPV_nolen_const, also try to
158 make it compile with perl 5.8.6 (yes, apple apparently loves 723 make it compile with perl 5.8.6 (yes, apple apparently loves
159 outdated software). Reported by John S. 724 outdated software). Reported by John S.
160 725
1614.744 Tue Jul 8 22:06:35 CEST 2008 7264.744 Tue Jul 8 22:06:35 CEST 2008
162 - correctly provide default DIE/WARN handlers as documented. 727 - correctly provide default DIE/WARN handlers as documented.
163 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 728 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
164 implementations inside perl work fine for us. 729 implementations inside perl work fine for us.
165 730
1664.743 Mon Jun 16 00:21:57 CEST 2008 7314.743 Mon Jun 16 00:21:57 CEST 2008
167 - when using Coro::EV without running EV::loop it could 732 - when using Coro::EV without running EV::loop it could
… …
189 - sprinkle "no warnings" freely over everything, also suppress 754 - sprinkle "no warnings" freely over everything, also suppress
190 warnings for some other modules. 755 warnings for some other modules.
191 - fix typo in WSAEWOULDBLOCK. 756 - fix typo in WSAEWOULDBLOCK.
192 757
1934.72 Sun May 25 05:14:36 CEST 2008 7584.72 Sun May 25 05:14:36 CEST 2008
194 - tweak META.yaml a bit, unfortunately, there is no documented way 759 - tweak META.yaml a bit, unfortunately, there is no documented way
195 to have optional dependencies with CPAN. doh :( 760 to have optional dependencies with CPAN. doh :(
196 - avoid running some tests on windows because they would fail due to 761 - avoid running some tests on windows because they would fail due to
197 perl bug (broken fork, broken pipes...). 762 perl bug (broken fork, broken pipes...).
198 - work around perl on windows bugs where perl returns undocumented 763 - work around perl on windows bugs where perl returns undocumented
199 error codes for sysread, syswrite etc. by taking advantage 764 error codes for sysread, syswrite etc. by taking advantage
… …
203 - use unix domain sockets in testsuite to work around 768 - use unix domain sockets in testsuite to work around
204 common perl implementation bugs on widows (they are emulated by 769 common perl implementation bugs on widows (they are emulated by
205 tcp sockets on windows. ugh.) 770 tcp sockets on windows. ugh.)
206 771
2074.71 Sat May 24 20:01:27 CEST 2008 7724.71 Sat May 24 20:01:27 CEST 2008
208 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 773 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
209 - take advantage of async name resolution of AnyEvent::Util. 774 - take advantage of async name resolution of AnyEvent::Util.
210 - work around brutal inet_aton override in Coro::LWP. 775 - work around brutal inet_aton override in Coro::LWP.
211 - take advantage of the readyhook in Coro::EV, for smoother 776 - take advantage of the readyhook in Coro::EV, for smoother
212 scheduling. 777 scheduling.
213 778
2144.7 Sun May 11 00:32:19 CEST 2008 7794.7 Sun May 11 00:32:19 CEST 2008
215 - completely reworked the Coro manpage. 780 - completely reworked the Coro manpage.
216 - added Coro::AnyEvent, generic event loop integration. 781 - added Coro::AnyEvent, generic event loop integration.
217 - implement cancel, ready and kill commands in Coro::Debug. 782 - implement cancel, ready and kill commands in Coro::Debug.
218 - document find_coro in Coro::Debug. 783 - document find_coro in Coro::Debug.
219 - incompatible change: rename has_stack to has_cctx. 784 - incompatible change: rename has_stack to has_cctx.
220 - Coro::AIO and Coro::BDB no longer force event model detection, 785 - Coro::AIO and Coro::BDB no longer force event model detection,
221 use AnyEvent::AIO and AnyEvent::BDB. 786 use AnyEvent::AIO and AnyEvent::BDB.
… …
345 910
3464.11 Thu Oct 11 02:40:24 CEST 2007 9114.11 Thu Oct 11 02:40:24 CEST 2007
347 - port to threaded perls. 912 - port to threaded perls.
348 913
3494.1 Thu Oct 11 02:38:16 CEST 2007 9144.1 Thu Oct 11 02:38:16 CEST 2007
350 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 915 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
351 be local to each coro (see Coro::State). 916 be local to each coro (see Coro::State).
352 - incompatible change: for very deep reasons, cede and cede_notself 917 - incompatible change: for very deep reasons, cede and cede_notself
353 cannot return anything, so nothing will be returned. 918 cannot return anything, so nothing will be returned.
354 - possibly bring back 5.10 compatibility (untested). 919 - possibly bring back 5.10 compatibility (untested).
355 - work around stupid (and wrong) warning on 5.10 :(. 920 - work around stupid (and wrong) warning on 5.10 :(.
… …
427 but stateful semantics. 992 but stateful semantics.
428 - fixed a lot of typos in Coro.pm (patch submitted by David 993 - fixed a lot of typos in Coro.pm (patch submitted by David
429 Steinbrunner, which applied flawlessly). 994 Steinbrunner, which applied flawlessly).
430 995
4313.6 Sat Apr 14 17:13:31 CEST 2007 9963.6 Sat Apr 14 17:13:31 CEST 2007
432 - added some bugfixes to get eg/myhttpd working again. 997 - added some bugfixes to get eg/myhttpd working again.
433 - added Coro::Storable for often-cede'ing freeze/thaw. 998 - added Coro::Storable for often-cede'ing freeze/thaw.
434 - try to do a clean exit when a coroutine calls exit 999 - try to do a clean exit when a coroutine calls exit
435 (EXPERIMENTAL). 1000 (EXPERIMENTAL).
436 - got rid of indirect call through _coro_init. 1001 - got rid of indirect call through _coro_init.
437 - updated the partly antique examples in eg/ to 1002 - updated the partly antique examples in eg/ to

Diff Legend

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