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

Comparing Coro/Changes (file contents):
Revision 1.333 by root, Sun Apr 6 19:23:49 2008 UTC vs.
Revision 1.668 by root, Fri Jul 14 03:03:56 2017 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO. 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: maybe implement a default message channel, very much like Erlang's 4TODO: channel->maxsize(newsize)?
5 actor model (which is cool in a lot of important aspects (failures!), 5TODO: __GCC_HAVE_DWARF2_CFI_ASM
6 but very lacking in others (higher level ipc)). 6TODO: swap_sv, maybe add scope_swap_sv?
7TODO: croak when async_pool tries to run canceled thread?
8
96.512 Fri Jul 14 05:03:17 CEST 2017
10 - seems to pass test suite down till 5.8.9, so make this the
11 minimum requirement in Makefile.PL.
12 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
13 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
14 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
15 fortified longjmp variants in some libcs.
16 - re-try JIT allocation without PROT_EXEC, to hopefully improve
17 portability (still doesn't seem to work around broken selinux
18 mmap everywhere).
19 - speling fixes (Xavier Guimard).
20
216.511 Sun Jun 26 23:44:50 CEST 2016
22 - make it compile with threaded perls again.
23 - simplify and speed up __DIE__ and __WARN__ handling.
24
256.51 Sat Jun 25 20:57:20 CEST 2016
26 - 6.5 release didn't compile on most older perls due to
27 a logic error enabling experimental 5.24 code for all versions.
28 - since various broken or deficient patches float around
29 to work around the vtbl API breakage in perl, let's provide
30 our own workaround, which seems to be much less invasive
31 and more compatible than the approaches seen so far. YMMV.
32 - work around assertion failure in perl_destruct on 5.24
33 (analyzed by Dave Mitchell).
34
356.5 Wed Jun 22 22:23:50 CEST 2016
36 - swap_sv swapping order was not symmetrical, causing
37 wrong swaps when swap_sv was used multiple times on the
38 same sv in the same thread.
39 - swap_sv calls can now be undone by calling it again
40 with the same variables.
41 - swap_sv calls will now be undone in async_pool threads.
42 - split Coro::Semaphore::up/adjust into separate xs functions
43 for better error reporting, at slight codesize increase.
44 - (libcoro) arm assembly support, please test and report.
45 - adjust to PL_savestack changes in perl 5.24 (adapted from
46 the debian patch, which unfortunately gets it wrong).
47
486.49 Sat Oct 17 01:40:12 CEST 2015
49 - throwing an exception to a thread waiting in
50 Coro::Handle using Coro::EV did not stop the watchers,
51 causing the next call to fail (testcase by Martin Pritchard).
52 - bump minimum perl version to 5.10.
53
546.48 Sun Oct 4 19:03:51 CEST 2015
55 - fix memory leak when Coro::Handle uses Coro::EV internally
56 (testcase by Sten Sten).
57 - update code to libev 4 API (internally, Coro still used the
58 version 3 API calls).
59 - remove unused hv_sig.
60
616.47 Sat Jul 11 03:58:20 CEST 2015
62 - swap_svs - NVs can be in the SV head beginning with 5.20.
63
646.46 Tue Jun 30 14:36:30 CEST 2015
65 - restore portability to perls with windows fork emulation
66 (patch by Petr Písař).
67
686.45 Tue Jun 30 01:40:08 CEST 2015
69 - restore portability to perl 5.16 and below.
70
716.44 Tue Jun 30 00:41:54 CEST 2015
72 - avoid segfaulting (or worse) when tracing a canceled thread.
73 - the Event module silently broke it's public hook API, causing
74 failures on perls compiled with -Duselongdouble. Requiring
75 the latest version of Event because I don't know which release
76 changed this (it's not mentioned in the ChangeLog).
77 - allow xs level enterleave hooks via CoroAPI.
78
796.43 Thu Jun 4 15:38:14 CEST 2015
80 - use stability canary.
81 - port to stableperl-5.22.0-1.001.
82 - update libecb.
83
846.42 Wed Feb 11 20:29:52 CET 2015
85 - Coro::SemaphoreSet->try did not actually work (analyzed by
86 SATO Kentaro).
87 - upgrade libecb to be C11 compliant.
88
896.41 Sat Sep 6 22:08:46 CEST 2014
90 - restore portability to perl 5.8.x.
91 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
92 up some memory, but fixes "use feature" and similar modules
93 when used inside a Coro.
94 - allow zero as argument to Coro::Channel to mean the same thing
95 as no argument. this works with older versions as well,
96 but wasn't legal until now.
97 - slightly better c header file detection.
98
996.39 Mon Jun 2 00:00:08 CEST 2014
100 - work around more incompatible changes in 5.20.
101
1026.38 Sun Jun 1 21:54:23 CEST 2014
103 - check that perl slots actually have enough space to hold
104 interpreter variables.
105 - untested port to perl 5.19 (64 bit tmps indices) (reported
106 by Andreas König).
107 - croak when cancel is called without a thread context.
108
1096.37 Tue Mar 4 13:27:33 CET 2014
110 - *sigh*, removed leftover debugging code from debugging a
111 perl bug, of all things.
112
1136.36 Tue Mar 4 07:11:59 CET 2014
114 - semaphores would not clear the destroy hook when interrupted
115 by ->throw, causing segfaults or worse.
116 - ->throw on a thread waiting for a semaphore did not acquire
117 the semaphore, but also didn't wake up other waiters,
118 possibly causing a deadlock.
119 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
120 to report a full stacktrace, which should help find out
121 where the actual call is.
122 - "atomically" destroy data in slf_destroy, because it is
123 the right thing to do, just in case.
124 - disable ecb.h libm dependency, to compile on ancient systems
125 or under adverse conditions.
126
1276.33 Mon Nov 18 11:26:27 CET 2013
128 - do not crash when freeing padlists with holes (in 5.18).
129 - tentative SVt_BIND 5.19 port/fix.
130
1316.32 Tue Nov 5 15:35:35 CET 2013
132 - use a new algorithm to derive padlists for perl 5.18. The old
133 one could lead to 0-pointer accesses inside perl (reported
134 by Darin McBride).
135
1366.31 Thu May 9 07:39:48 CEST 2013
137 - Coro::AIO requests would crash if the thread was ready'd
138 while the request was ongoing.
139
1406.29 Wed May 8 02:55:18 CEST 2013
141 - when an on_destroy handler destructs the coro currently being
142 destructed a perl scalar could be accessed after being freed,
143 likely causing a crash.
144
1456.28 Wed Mar 6 06:58:02 CET 2013
146 - clean remnants of existing __DIE__ and __WARN__ handlers so
147 they lose their magic and will not cause segfaults later
148 (testcase by Andrey Sagulin).
149 - improved Coro::State documentation a bit.
150 - Coro::Debug::command now flushes the output.
151 - add hack detection code for x32 abi, because the braindead slugs
152 who designed that made it look exactly like x86_64 without
153 providing proper compile time symbols to test for it. as a result,
154 this detection cannot work reliably.
155 - valgrind stack registering was broken.
156 - do not rely on Time::HiRes anymore in Coro::Debug.
157
1586.23 Fri Dec 7 23:36:37 CET 2012
159 - use experimental fiber implementation on native windows
160 perls.
161 - use sizeof (void *) as multiplication factor for stack sizes,
162 to accomodate the totally braindamaged microsoft 64 bit "os".
163 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
164 activeperl 5.16 crashes when PerlIO_define_layer is called due
165 to some bug in the perl dll, strawberry perl at least passes
166 the testsuite.
167 - implement Coro::Handle->peeraddr/host/port, for slightly
168 improved compatibility with LWP.
169 - implement 5.17 compatibility by almost blindly applying a
170 good-looking patch by Father Chrysostomos.
171 - move stack management functions into libcoro 3.
172 - libcoro version 3 "released".
173 - support magic values as timed_io_once args.
174 - recommend AnyEvent 7+ or EV 4+, also require EV
175 version 4 or newer for Coro::EV.
176
1776.10 Tue Oct 9 01:14:27 CEST 2012
178 - updated ecb.h, it had a typo that caused it to not compile on many
179 big endian systems (reported by many people).
180 - disable memory fences in ecb.h to improve portability.
181
1826.09 Sat Oct 6 23:25:02 CEST 2012
183 - Coro::EV I/O watchers were not interruptible by exceptions
184 (Coro::State::throw) (testcase by sten).
185 - ->throw now puts threads into the ready queue, as this seems to
186 be expected by existing code, and code that doesn't cope with spurious
187 wakeups needs fixing anyway.
188 - use fd -1 in mmap.
189 - cast I32 to int in error message printf.
190 - warn about broken so-called "hardened" kernels.
191
1926.08 Fri Apr 13 12:05:47 CEST 2012
193 - be more aggressive about exiting like perl does - formerly,
194 exiting from the non-main thread would not execute END blocks.
195
1966.07 Fri Nov 11 21:21:48 CET 2011
197 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
198 - update ecb.h.
199
2006.06 Mon Aug 8 23:59:48 CEST 2011
201 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
202 backend, which is an order of magnitude slower. unfortunately, cygwins
203 pthread implementation isn't very complete either, so allocate the stack
204 twice just to be sure.
205 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
206 in its ithreaded perl - best recompile cygwin and use the 'w'indows
207 backend for much better performance. also disable ithreads for
208 even better performance...).
209
2106.05 Thu Aug 4 21:36:36 CEST 2011
211 - blush, condvar values would not be propagated from send to recv anymore
212 (reported by Chip Salzenberg).
213 - use exponential increase for the readline buffer length in
214 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
215
2166.04 Wed Aug 3 17:15:45 CEST 2011
217 - use even more efficient and more compatible condvars for
218 compatibility to AnyEvent 6.x :)
219 - more inconsequential ecb.h updates.
220
2216.03 Wed Aug 3 11:41:30 CEST 2011
222 - change how Coro patches AnyEvent condvars for compatibility to
223 AnyEvent 6.x.
224 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
225 hasn't been included.
226
2276.02 Wed Jul 13 04:35:19 CEST 2011
228 - "improve portability to Gentoo" - gentoo manages to put perl variables
229 in memory areas that are farther than 2gb apart, which the jit couldn't
230 handle and barfed. now it's just a bit slower on gentoo and similar
231 systems.
232
2336.01 Sun Jul 3 12:31:14 CEST 2011
234 - workarounds are good, but the test for whether pthreads are used
235 was not good. this one should be better.
236 - check differently whether gcc generates cfi instructions itself.
237
2386.0 Wed Jun 29 19:43:35 CEST 2011
239 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
240 destroyed and cleaned up automatically (e.g. async { schedule }).
241 - implement a JIT compiler for part of the thread switch code,
242 which gives a 50% speed improvement on threaded perls, and
243 about 4% on non-threaded perls (so threaded perls now finally
244 reach about half the speed of non-threaded perls).
245 - slightly modernise Coro::Intro, add section about rouse functions.
246 - avoid DEFSV and ERRSV, giving another 10% improvement
247 in thread switching.
248 - Coro::State->is_destroyed is now called is_zombie.
249 - implement a Coro->safe_cancel method that might fail, but
250 cancels in a "safer" way if it succeeds.
251 - add preliminary support for DEBUGGING perls.
252 - get rid of two hash-accesses when initialising a new Coro - this
253 speeds up coro creation by almost a factor of two.
254 - croak when a coro that is being cancelled tries to block
255 (e.g. while executing a guard block), instead of crashing or
256 deadlocking.
257 - use a more robust and also faster method to identify Coro::State
258 objects - speeds up everything a bit.
259 - implement Coro->cancel in XS for a 20% speed improvement, and to
260 be able to implement mutual cancellation.
261 - speed up context switches by a percent or two by more efficiently
262 allocating context stack entries.
263 - implement Coro->join and Coro->on_destroy in XS for a speedup and
264 a reduction in memory use.
265 - cancelling a coro while it itself is cancelling another coro is
266 now supported and working, instead of triggering an assertion.
267 - be a bit more crash-resistant when calling (buggy) on_destroy
268 callbacks (best effort).
269 - move on_destroy into the slf_frame, to allow extension slf
270 functions to have destructors.
271 - get rid if coro refcounting - simply crash in other interpreter
272 threads by nulling the pointers on clone.
273 - simplify warn/die hook handling when loading Coro - the convoluted
274 logic seems to be no longer neccessary.
275 - use libecb instead of our own home-grown gcc hacks.
276 - document alternatives to Coro::LWP. Please use them :)
277 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
278 sigaltstack. Really. wine suffers from it, erlang suffers from it,
279 and it's known since at least 2006.
280
2815.372 Wed Feb 23 06:14:30 CET 2011
282 - apparently mingw doesn't provide a working gettimeofday, try to
283 work around that by relying on Time::HiRes (indirectly brought to
284 my attention by Max Maischein).
285 - fix some portability issues when Time::HiRes was used.
286
2875.371 Mon Feb 21 14:36:08 CET 2011
288 - backport to windows process emulation code again.
289
2905.37 Sat Feb 19 07:49:44 CET 2011
291 - add a big "Coro thread life cycle" section to "man Coro".
292 - try a tentative workaround against the breakage that 5.13 has
293 introduced without depreciation period. sigh.
294 - no longer use Time::HiRes if gettimeofday is available, which
295 saves quite a lot of memory.
296
2975.36 Sun Feb 13 05:33:41 CET 2011
298 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
299 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
300 method call.
301 - undocument Coro::Timer::sleep and obsolete whole module.
302 - optimise Coro::Timer::timeout memory and cpu usage.
303 - slightly updated Coro::Intro for recent changes.
304 - do not initialise PL_dirty anymore.
305
3065.25 Thu Nov 11 01:08:39 CET 2010
307 - try a different approach on netbsd - netbsd 5 finally has marginally
308 working pthreads, but still broken ucontext/sigaltstack.
309 - openbsd 4.8 finally got their act together, Coro works out of the box
310 with asm, setjmp and pthreads (no change, just informational).
311
3125.24 Sat Oct 23 11:27:12 CEST 2010
313 - port to the EV 4.0 API.
314 - work around bugs in mingw32, making strawberry perl work
315 out of the box.
316 - correctly modify Coro::AIO function prototypes
317 so that they reflect the "no optional parameters" rule.
318 - "ported" libcoro to C++.
319
3205.23 Mon May 17 18:50:42 CEST 2010
321 - be more resistant to ordering changes when initialising
322 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
323 Waldorf).
324 - document that perl 5.12 deliberately removed support for cloning.
325
3265.22 Wed Apr 14 03:55:35 CEST 2010
327 - correctly return udnef on errors in Coro::Handle::read/write
328 (testcase by Marc Mims).
329 - convert Coro::Util into a "perl compatibility wrapper" - the functions
330 are less useful now, but are drop-in replacements for existing
331 functions, listing better alternatives in the documentation. This also
332 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
333 with Coro::Util::inet_aton.
334 - do not override $Coro::idle unconditionally in Coro.pm, as other
335 modules could have provided their own idle coro already
336 (for exmaple, Coro::AnyEvent).
337 - fix Coro::Util::gethost* functions.
338 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
339
3405.21 Wed Dec 16 07:19:51 CET 2009
341 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
342 together.
343 - add some examples on how to combine other event loops with Coro in
344 Coro::AnyEvent, and how to run it (and not to block). Seems to be
345 the most common source of confusion.
346 - try to catch people naively blocking in an event callback.
347 - work around the perl filehandle bug issue in conjunction with
348 older common::sense (as indirectly pointed out by ZSystem).
349 - clarify the "not from signal handlers" section.
350
3515.2 Sun Oct 4 14:54:24 CEST 2009
352 - Coro::Storable destroyed the prototypes of the functions it wrapped.
353 - export rouse_cb and rouse_wait by default now.
354 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
355 - new method $state->swap_sv.
356 - added section on "windows process emulation" to the manpage,
357 after a not-so-fruitful (nor-friendly) "discussion" with chip
358 salzenberg (discussion implies arguments, but his only arguments
359 were ad-hominems, one wonders why he started it in the first
360 place). I hope this explains it well enough for him to understand,
361 and maybe well enough for others to understand.
362 - use common::sense everywhere now.
363 - idle callbacks are no longer supported, use idle coros instead.
364 - print a thread listing when a deadlock is detected.
365
3665.17 Sat Aug 22 23:09:31 CEST 2009
367 - work around a bug in the perl debugger causing crashes
368 when running under the debugger by marking _pool_handler as nodebug.
369 - speed up Coro::async considerably.
370 - try some hacks to get netbsd to work "more often" - their broken
371 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
372 - convert Coro to AE 5.0 API.
373
3745.162 Tue Jul 28 04:04:03 CEST 2009
375 - perl 5.8.2 is now minimum requirement.
376 - skip t/19_handle.t on broken windows perls.
377
3785.161 Wed Jul 22 04:47:38 CEST 2009
379 - Coro::AnyEvent::poll could have a different prototype when EV was
380 used as backend (analyzed by Tatsuhiko Miyagawa).
381 - Coro::AnyEvent errornously initialised the event loop when loaded,
382 not on demand.
383 - try to workaround rare */t/01_unblock.t failures.
384
3855.16 Tue Jul 21 01:44:37 CEST 2009
386 - Coro::AnyEvent failed to hook into the event loop
387 when no threads had been readied between detecting
388 the event loop and actually running it.
389 - considerably speed up Coro::Select by taking avdantage
390 of AnyEvent > 4.8 and some other optimisations.
391 - implement paragraph readline mode in Coro::Handle
392 (based on patches by Zsbán Ambrus).
393 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
394 and analyzed by Yasuhiro MATSUMOTO).
395 - clarified libcoro license and copyright.
396 - someone stole my EXTRA_META!!!
397 - implement Coro::Select::patch_pp_sselect and it's brother,
398 for hardcode select overriding.
399
4005.151 Mon Jul 6 05:41:57 CEST 2009
401 - backport to windows process emulation code again (patch by
402 Yasuhiro MATSUMOTO).
403 - slightly update Coro::MakeMaker.
404
4055.15 Tue Jun 30 10:28:06 CEST 2009
406 - deprecate Coro::Socket, document how to get ipv6 support via
407 AnyEvent::Socket instead.
408 - implement signal->wait ($cb) interface, similar to semaphores.
409 - work around SvOK not supporting getmagic, so we have to getmagic
410 to test for undef :/ (reported by Matthias Waldorf).
411 - load Coro::AnyEvent in all modules using AnyEvent.
412 - work around perl corrupting our internal data structures,
413 reported by Tokuhiro Matsuno.
414 - enable per-coroutine real and cpu time gathering
415 (Coro::State::enable_times).
416
4175.14 Wed Jun 24 01:37:48 CEST 2009
418 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
419 as this seems to have been a difficult concept (poll, sleep, idle,
420 idle_upto).
421 - add Coro::AnyEvent::readable/writable functions.
422 - clarify Coro::EV/Event/AnyEvent manpages.
423 - free per-thread global scalars in the thread calling ->cancel, to
424 avoid crashes when $_, $@ etc., are magical but some of those
425 had already been freed.
426 - "unexperimentalise" the callback interface for Coro::Semaphore.
427 - speed up ready queue management/context switching by using a linked
428 list instead of an array (~5-10%).
429 - implement "watch" command in Coro::Debug shells.
430 - for fun, implement time-slicing as an example in the manpage.
431 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
432 Coro::Event, use the normal AnyEvent handling instead of dieing
433 (the same is true for Coro::Handle).
434 - properly document Coro::EV::timed_io_once.
435 - avoid unneccessary ->cancel calls in Coro::Handle.
436 - maybe make it work on mingw32 with win32 backend
437 (based on patch by Yasuhiro Matsumoto).
438
4395.132 Fri May 29 09:00:39 CEST 2009
440 - do not keep a reference to the argument itself in
441 Coro::Semaphore::guard, as it could change later.
442 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
443 should support a prepare callback.
444
4455.131 Mon Mar 16 23:20:37 CET 2009
446 - implement and document Coro->suspend, Coro->resume.
447 - fix Coro::Select implementation to not (often) close
448 the passed file descriptors (testcase provided by pippijn).
449
4505.13 Mon Dec 15 21:51:42 CET 2008
451 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
452 - don't set diehook to C<undef> but instead to NULL, to avoid
453 spurious warnings.
454 - fix a lot of bugs in Coro::SemaphoreSet.
455 - Coro::SemaphoreSet will less often create a semaphore needlessly.
456 - add Coro::SemaphoreSet::count and wait methods.
457 - take advantage of the new Guard module.
458 - deprecate Coro::guard.
459 - try to fix the dreaded 01_unblock tests once more. I hate it when
460 testsuites need more fixing than the code they are supposed to test.
461 - croak in more cases when a required callback isn't resolvable.
462 - fix some minor issues in Coro::State->call/eval.
463 - use current coroutine context instead of a temporary one
464 when temporarily switching to another coroutine.
465 - do not call C-level on_destroy handlers during global destruction,
466 to avoid needless segfaults.
467
4685.12 Sun Dec 7 13:30:38 CET 2008
469 - add default config for MirOS, which seems to be a bug-to-bug
470 compatible fork of openbsd ("world domination by releasing
471 openbsd cvs before the openbsd folks do it" or so :).
472 - free_padlist did destroy the names pad, not good, but didn't
473 seem to bother perl - this could fix issues such as eval ""
474 inside a function called from multiple coroutines.
475 - use a different method to detect destruction time.
476 - be more careful when freeing padlists just before global
477 destruction.
478 - fixed and expanded the call/cc example.
479 - renamed _terminate to _coro_run.
480 - new method Coro::Channel->shutdown.
481 - try pthreads on openbsd <4.4 (broken sigaltstack, will
482 pthreads fare better?).
483 - be less picky about destroying "a" running coroutine.
484
4855.11 Tue Nov 25 21:49:05 CET 2008
486 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
487 working code with failed assertions, introducing more bugs than
488 it fixes, requiring elaborate workarounds :(
489
4905.1 Mon Nov 24 08:54:59 CET 2008
491 - wrote a small introductory tutorial - Coro::Intro.
492 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
493 - Coro::Select did errornously dup the file descriptors
494 and didn't work with all AnyEvent backends.
495 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
496 LWP data transfers.
497 - disassociate c contexts from coro objects - this is agruably more
498 correct, but mostly allows sharing of cctxs between coro and state
499 objects, for added memory savings and speed increases.
500 - bumped $Coro::POOL_RSS up to 32kb by default.
501 - no longer set the optype to OP_CUSTOM, as B::* understandably
502 doesn't like this very much (and we *are* a type of entersub).
503 - implement state cloning, just to prove that call/cc can be done.
504 - automatically load Coro::AnyEvent in Coro::Handle.
505 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
506 do not have this method (and don't need it either).
507 - speed up generic anyevent methods in Coro::Handle by using rouse
508 callbacks.
509 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
510 others. It also makes the debugger happier, as you can trace
511 through the idle threads now.
512 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
513 - eg/event was pretty broken.
514 - better 5.8.6 compatibility.
515
5165.0 Thu Nov 20 10:35:05 CET 2008
517 - NEW ARCHITECTURE: use the latest 4.x version if you experience
518 stability issues.
519 - bump API version to 7 - all dependents must be recompiled.
520 - removed timed_* functions - they were not being used anyways
521 and should be replaced by a more generic mechanism -
522 and were annoying to support anyways :)
523 - removed SemaphoreSet's waiter method - use sem method instead.
524 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
525 - async_pool did free a scalar value twice
526 ("Attempt to unreference...").
527 - fix a longstanding bug where calling terminate on a coro that
528 was waiting for a semaphore that was just becoming available
529 would cause a deadlock (semaphore would get into a state where
530 it was available but waiters were still blocked).
531 - calling throw on a coroutine that is waiting for a semaphore will
532 no longer make it acquire the semaphore (and thus leak a count).
533 - perl's process emulation is now not even theoretically supported
534 anymore.
535 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
536 conversion of callback-style to blocking-style.
537 - new methods $coro->schedule_to and ->cede_to, to specifically
538 schedule or cede to a specific coroutine.
539 - new function Coro::Semaphore::wait.
540 - use named constants in Coro::Channel (Richard Hundt).
541 - directly patch the entersub opcode calling SLF functions (cede,
542 transfer and so on). this does speed up context switching, but
543 more importanly, it frees us from the hardcoded behaviour of
544 entersub, so we might actually be able to return something from
545 those functions and atcually create new ones.
546 - take advantage of __builtin_frame_address on gcc.
547 - expose THX in coroapi (not sure whether this was a wise decision,
548 as "threaded" perls are running at half speed anyways).
549 - implement execute_slf (schedule-like-function) interface that makes
550 it possible to implement schedule-like-functions in XS.
551 - use new SLF interface to massively speed up Coro::EV by roughly a
552 factor of two.
553 - used new SLF interface to massively speed up Coro::Semaphore by a
554 factor of three.
555 - used new SLF interface to speed up Coro::AIO by roughly a factor of
556 four and reduce its memory usage considerably.
557 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
558 for a nice speedup and vastly more correct behaviour. Also implement
559 a new method "sem" to get at the underlying semaphore object.
560 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
561 (in comparison) 20-40% speedup.
562 - used new SLF interface to reimplement Coro::Signal gaining
563 some unknown (because I was too lazy), but certain, speedup, and also
564 making signals reliable for the first time.
565 - used new SLF interface and other optimisations to speed up async_pool
566 by a factor of two. It also doesn't rely on perl's exception mechanism
567 to exit anymore. The overhead for terminating an async_pool, coro over
568 a normal async is now very small.
569 - sped up coroutine creation/destruction by 40%.
570 - forgot to include Coro/libcoro/README in the dist for all these years.
571 - work around a freebsd pthreads bug (manual testcancel is required as
572 pthread_cond_wait isn't a cancellation point on freebsd).
573 - use new rouse functions to speed up and simplify Coro::BDB.
574 - make "prefer perl native functions" work with threaded perls.
575 - condense Coro::Debug ps output, hint at v and w flags.
576 - (libcoro) lots of minor cleanups and portability improvements.
577
5784.914 Wed Nov 19 12:54:18 CET 2008
579 - fix a disastrous bug in the readline optimisation
580 introduced in 4.801.
581
5824.913 Sat Nov 15 07:58:28 CET 2008
583 - async_pool did free a scalar value twice
584 ("Attempt to unreference...").
585
5864.912 Thu Nov 13 18:31:23 CET 2008
587 - minor cleanups.
588 - use much larger stacks on linux and perl < 5.8.8.
589 - Coro::Debug::new_unix_server did not unlink the socket
590 when destroyed.
591
5924.911 Tue Nov 11 04:26:17 CET 2008
593 - "port" to threaded perls.
594
5954.91 Mon Nov 10 05:36:38 CET 2008
596 - the ->throw exception object no longer leaks.
597 - creating a new cctx leaked a scopestack entry (memleak).
598 - new coroutines didn't get created with a zero flags field
599 (unknown impact).
600 - calling ->throw on a not-yet-started coroutine should now work
601 instead of being ignored.
602 - ->throw is now supported on Coro::State objects.
603 - clean up cctx creation code a bit.
604 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
605
6064.9 Sat Nov 8 17:45:27 CET 2008
607 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
608 - (libcoro) no longer leak threads in the experimental pthread backend,
609 also speed it up considerably.
610 - (libcoro) do not rely on makecontext passing void *'s unscathed.
611 - fix compiletime dependencies on libcoro in the Makefile.
612 - cctx_count wasn't always updated properly.
613 - Coro::State::cctx_stacksize wasn't applied correctly.
614 - new function Coro::State::cctx_max_idle.
615 - the default max number of idle C contexts is now 4.
616 - (libcoro) try harder to get _setjmp/_longjmp.
617 - (libcoro) cleanup and extend the libcoro API to officially
618 allow the creation of empty source contexts.
619 - very experimental workaround for the totally broken netbsd platform.
620 - tried to hack around openbsd bugs.
621
6224.804 Wed Nov 5 16:36:00 CET 2008
623 - Coro::Debug::new_unix_server would not create a non-blocking listening
624 socket, sometimes causing freezes.
625 - (libcoro) fix misaligned stack points for setjmp and assembly
626 methods, which can cause crashes on x86/x86_64 with a sufficiently
627 aggressive compiler.
628 - new function: Coro::Debug::new_tcp_server.
629 - move ->throw into the Coro class because it only works on coro objects.
630
6314.803 Mon Nov 3 17:16:12 CET 2008
632 - (libcoro) use a global asm statement to become independent of gcc
633 otpimisations for CORO_ASM (thanks to pippijn for the idea).
634 - try to workaround yet another broken bsd, this time dragonfly.
635
6364.802 Thu Oct 30 10:56:12 CET 2008
637 - support -fno-omit-frame-pointer on x86 with the assembly method.
638 - tune 01_unblock.t tests a bit.
639
6404.801 Wed Oct 22 18:33:37 CEST 2008
641 - improve readline speed for very long "lines".
642 - backport to 5.8.8.
643
6444.8 Thu Oct 2 13:34:40 CEST 2008
645 - new function Coro::AIO::aio_wait.
646 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
647 - greatly speed up and reduce memory usage of Coro::AIO requests.
648 - implement some other µ-optimisations.
649
6504.749 Mon Sep 29 14:40:12 CEST 2008
651 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
652
6534.748 Sat Sep 27 14:03:19 CEST 2008
654 - implement, but do not document, PerlIO::cede(granularity).
655 - Coro::Storable forgot to wrap Storable::pstore.
656 - work around the multitude of leaks and memory corruption
657 bugs in PerlIO::via by using our own C-level perliol. As a side
658 effect, Coro::Storable is now much, much, much faster.
659
6604.747 Tue Sep 23 01:59:41 CEST 2008
661 - fix a per-cv memleak (one empty array was leaked per
662 code reference).
663 - avoid a crash in coro->call|rss when the coroutine was already
664 destroyed (most noticably when using Coro::Debug::ps :)
665 - also protect *Storable::FILE.
666 - push up default storable granularity to 20ms.
667
6684.746 Sun Sep 21 03:22:20 CEST 2008
669 - be more insistent on locking Storable against reentrancy
670 in Coro::Storable.
671 - move swap_def?v and throw to Coro::State, as documented.
672
6734.745 Thu Jul 24 00:14:38 CEST 2008
674 - remove debugging code related to MgPV_nolen_const, also try to
675 make it compile with perl 5.8.6 (yes, apple apparently loves
676 outdated software). Reported by John S.
677
6784.744 Tue Jul 8 22:06:35 CEST 2008
679 - correctly provide default DIE/WARN handlers as documented.
680 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
681 implementations inside perl work fine for us.
682
6834.743 Mon Jun 16 00:21:57 CEST 2008
684 - when using Coro::EV without running EV::loop it could
685 result in busy-waiting for events, this has been fixed.
686 - reduce codesize and improve performance by using EV_DEFAULT_UC.
687
6884.742 Sat May 31 14:10:21 CEST 2008
689 - implement a workaround for (some) perl <5.8.8 versions.
690 - require EV 3.3+.
691
6924.741 Fri May 30 23:33:09 CEST 2008
693 - tell netbsd how utterly broken their imitation of an OS is
694 and refuse to build by default if pthreads are in use.
695 - switch to "s" method on all bsd's by default, as their ucontext
696 stuff seems just too broken.
697 - fix a bug in Coro::Select.
698
6994.74 Thu May 29 20:05:31 CEST 2008
700 - do not test Coro::LWP for lack of dependencies.
701 - do not test Coro::Select for lack of working perls.
702
7034.73 Thu May 29 2008
704 - fix a bug in Coro::EV which would cause it to block despite
705 there being runnable coroutines.
706 - sprinkle "no warnings" freely over everything, also suppress
707 warnings for some other modules.
708 - fix typo in WSAEWOULDBLOCK.
709
7104.72 Sun May 25 05:14:36 CEST 2008
711 - tweak META.yaml a bit, unfortunately, there is no documented way
712 to have optional dependencies with CPAN. doh :(
713 - avoid running some tests on windows because they would fail due to
714 perl bug (broken fork, broken pipes...).
715 - work around perl on windows bugs where perl returns undocumented
716 error codes for sysread, syswrite etc. by taking advantage
717 of AnyEvent's workaround for that problem.
718 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
719 a common perl implementation bug on windows.
720 - use unix domain sockets in testsuite to work around
721 common perl implementation bugs on widows (they are emulated by
722 tcp sockets on windows. ugh.)
723
7244.71 Sat May 24 20:01:27 CEST 2008
725 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
726 - take advantage of async name resolution of AnyEvent::Util.
727 - work around brutal inet_aton override in Coro::LWP.
728 - take advantage of the readyhook in Coro::EV, for smoother
729 scheduling.
730
7314.7 Sun May 11 00:32:19 CEST 2008
732 - completely reworked the Coro manpage.
733 - added Coro::AnyEvent, generic event loop integration.
734 - implement cancel, ready and kill commands in Coro::Debug.
735 - document find_coro in Coro::Debug.
736 - incompatible change: rename has_stack to has_cctx.
737 - Coro::AIO and Coro::BDB no longer force event model detection,
738 use AnyEvent::AIO and AnyEvent::BDB.
739
7404.6 Sat Apr 26 10:05:14 CEST 2008
741 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
742 by the Coro module. It was a mistake to have it in the first place.
743 - (experimental) support for activestate perl 5.10 (method "w").
744 - (experimental) support for strawberry perl (method "a").
745 - coro_sigelem_set did not return a value although it had to,
746 actual impact unknown.
747
7484.51 Mon Apr 14 13:28:27 CEST 2008
749 - make it compile again on 5.8.
750
7514.50 Thu Apr 10 09:43:17 CEST 2008
752 - I did it twice! (see 4.49).
7 753
84.49 Sun Apr 6 21:23:31 CEST 2008 7544.49 Sun Apr 6 21:23:31 CEST 2008
9 - grr, instead of compiling the recent changes on 5.10 only they 755 - grr, instead of compiling the recent changes on 5.10 only they
10 were compiled on 5.8 only. 756 were compiled on 5.8 only.
11 757
… …
116 862
1174.11 Thu Oct 11 02:40:24 CEST 2007 8634.11 Thu Oct 11 02:40:24 CEST 2007
118 - port to threaded perls. 864 - port to threaded perls.
119 865
1204.1 Thu Oct 11 02:38:16 CEST 2007 8664.1 Thu Oct 11 02:38:16 CEST 2007
121 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 867 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
122 be local to each coro (see Coro::State). 868 be local to each coro (see Coro::State).
123 - incompatible change: for very deep reasons, cede and cede_notself 869 - incompatible change: for very deep reasons, cede and cede_notself
124 cannot return anything, so nothing will be returned. 870 cannot return anything, so nothing will be returned.
125 - possibly bring back 5.10 compatibility (untested). 871 - possibly bring back 5.10 compatibility (untested).
126 - work around stupid (and wrong) warning on 5.10 :(. 872 - work around stupid (and wrong) warning on 5.10 :(.
… …
198 but stateful semantics. 944 but stateful semantics.
199 - fixed a lot of typos in Coro.pm (patch submitted by David 945 - fixed a lot of typos in Coro.pm (patch submitted by David
200 Steinbrunner, which applied flawlessly). 946 Steinbrunner, which applied flawlessly).
201 947
2023.6 Sat Apr 14 17:13:31 CEST 2007 9483.6 Sat Apr 14 17:13:31 CEST 2007
203 - added some bugfixes to get eg/myhttpd working again. 949 - added some bugfixes to get eg/myhttpd working again.
204 - added Coro::Storable for often-cede'ing freeze/thaw. 950 - added Coro::Storable for often-cede'ing freeze/thaw.
205 - try to do a clean exit when a coroutine calls exit 951 - try to do a clean exit when a coroutine calls exit
206 (EXPERIMENTAL). 952 (EXPERIMENTAL).
207 - got rid of indirect call through _coro_init. 953 - got rid of indirect call through _coro_init.
208 - updated the partly antique examples in eg/ to 954 - updated the partly antique examples in eg/ to

Diff Legend

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