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

Comparing Coro/Changes (file contents):
Revision 1.358 by root, Fri May 30 21:34:52 2008 UTC vs.
Revision 1.655 by root, Fri Jun 17 04:49:22 2016 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, andallow to unswap by repeat?
7TODO: swap_sv not undone in pool
8TODO: croak when async_pool tries to run canceled thread?
9
10 - swap_sv swapping order was not symmetrical, causing
11 wrong swaps when swap_sv was used multiple times on the
12 same sv in the same thread.
13 - swap_sv calls can now be undone by calling it again
14 with the same variables.
15 - split Coro::Semaphore::up/adjust into separate xs functions
16 for better error reporting, at slight codesize increase.
17 - (libcoro) arm assembly support.
18
196.49 Sat Oct 17 01:40:12 CEST 2015
20 - throwing an exception to a thread waiting in
21 Coro::Handle using Coro::EV did not stop the watchers,
22 causing the next call to fail (testcase by Martin Pritchard).
23 - bump minimum perl version to 5.10.
24
256.48 Sun Oct 4 19:03:51 CEST 2015
26 - fix memory leak when Coro::Handle uses Coro::EV internally
27 (testcase by Sten Sten).
28 - update code to libev 4 API (internally, Coro still used the
29 version 3 API calls).
30 - remove unused hv_sig.
31
326.47 Sat Jul 11 03:58:20 CEST 2015
33 - swap_svs - NVs can be in the SV head beginning with 5.20.
34
356.46 Tue Jun 30 14:36:30 CEST 2015
36 - restore portability to perls with windows fork emulation
37 (patch by Petr Písař).
38
396.45 Tue Jun 30 01:40:08 CEST 2015
40 - restore portability to perl 5.16 and below.
41
426.44 Tue Jun 30 00:41:54 CEST 2015
43 - avoid segfaulting (or worse) when tracing a canceled thread.
44 - the Event module silently broke it's public hook API, causing
45 failures on perls compiled with -Duselongdouble. Requiring
46 the latest version of Event because I don't know which release
47 changed this (it's not mentioned in the ChangeLog).
48 - allow xs level enterleave hooks via CoroAPI.
49
506.43 Thu Jun 4 15:38:14 CEST 2015
51 - use stability canary.
52 - port to stableperl-5.22.0-1.001.
53 - update libecb.
54
556.42 Wed Feb 11 20:29:52 CET 2015
56 - Coro::SemaphoreSet->try did not actually work (analyzed by
57 SATO Kentaro).
58 - upgrade libecb to be C11 compliant.
59
606.41 Sat Sep 6 22:08:46 CEST 2014
61 - restore portability to perl 5.8.x.
62 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
63 up some memory, but fixes "use feature" and similar modules
64 when used inside a Coro.
65 - allow zero as argument to Coro::Channel to mean the same thing
66 as no argument. this works with older versions as well,
67 but wasn't legal until now.
68 - slightly better c header file detection.
69
706.39 Mon Jun 2 00:00:08 CEST 2014
71 - work around more incompatible changes in 5.20.
72
736.38 Sun Jun 1 21:54:23 CEST 2014
74 - check that perl slots actually have enough space to hold
75 interpreter variables.
76 - untested port to perl 5.19 (64 bit tmps indices) (reported
77 by Andreas König).
78 - croak when cancel is called without a thread context.
79
806.37 Tue Mar 4 13:27:33 CET 2014
81 - *sigh*, removed leftover debugging code from debugging a
82 perl bug, of all things.
83
846.36 Tue Mar 4 07:11:59 CET 2014
85 - semaphores would not clear the destroy hook when interrupted
86 by ->throw, causing segfaults or worse.
87 - ->throw on a thread waiting for a semaphore did not acquire
88 the semaphore, but also didn't wake up other waiters,
89 possibly causing a deadlock.
90 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
91 to report a full stacktrace, which should help find out
92 where the actual call is.
93 - "atomically" destroy data in slf_destroy, because it is
94 the right thing to do, just in case.
95 - disable ecb.h libm dependency, to compile on ancient systems
96 or under adverse conditions.
97
986.33 Mon Nov 18 11:26:27 CET 2013
99 - do not crash when freeing padlists with holes (in 5.18).
100 - tentative SVt_BIND 5.19 port/fix.
101
1026.32 Tue Nov 5 15:35:35 CET 2013
103 - use a new algorithm to derive padlists for perl 5.18. The old
104 one could lead to 0-pointer accesses inside perl (reported
105 by Darin McBride).
106
1076.31 Thu May 9 07:39:48 CEST 2013
108 - Coro::AIO requests would crash if the thread was ready'd
109 while the request was ongoing.
110
1116.29 Wed May 8 02:55:18 CEST 2013
112 - when an on_destroy handler destructs the coro currently being
113 destructed a perl scalar could be accessed after being freed,
114 likely causing a crash.
115
1166.28 Wed Mar 6 06:58:02 CET 2013
117 - clean remnants of existing __DIE__ and __WARN__ handlers so
118 they lose their magic and will not cause segfaults later
119 (testcase by Andrey Sagulin).
120 - improved Coro::State documentation a bit.
121 - Coro::Debug::command now flushes the output.
122 - add hack detection code for x32 abi, because the braindead slugs
123 who designed that made it look exactly like x86_64 without
124 providing proper compile time symbols to test for it. as a result,
125 this detection cannot work reliably.
126 - valgrind stack registering was broken.
127 - do not rely on Time::HiRes anymore in Coro::Debug.
128
1296.23 Fri Dec 7 23:36:37 CET 2012
130 - use experimental fiber implementation on native windows
131 perls.
132 - use sizeof (void *) as multiplication factor for stack sizes,
133 to accomodate the totally braindamaged microsoft 64 bit "os".
134 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
135 activeperl 5.16 crashes when PerlIO_define_layer is called due
136 to some bug in the perl dll, strawberry perl at least passes
137 the testsuite.
138 - implement Coro::Handle->peeraddr/host/port, for slightly
139 improved compatibility with LWP.
140 - implement 5.17 compatibility by almost blindly applying a
141 good-looking patch by Father Chrysostomos.
142 - move stack management functions into libcoro 3.
143 - libcoro version 3 "released".
144 - support magic values as timed_io_once args.
145 - recommend AnyEvent 7+ or EV 4+, also require EV
146 version 4 or newer for Coro::EV.
147
1486.10 Tue Oct 9 01:14:27 CEST 2012
149 - updated ecb.h, it had a typo that caused it to not compile on many
150 big endian systems (reported by many people).
151 - disable memory fences in ecb.h to improve portability.
152
1536.09 Sat Oct 6 23:25:02 CEST 2012
154 - Coro::EV I/O watchers were not interruptible by exceptions
155 (Coro::State::throw) (testcase by sten).
156 - ->throw now puts threads into the ready queue, as this seems to
157 be expected by existing code, and code that doesn't cope with spurious
158 wakeups needs fixing anyway.
159 - use fd -1 in mmap.
160 - cast I32 to int in error message printf.
161 - warn about broken so-called "hardened" kernels.
162
1636.08 Fri Apr 13 12:05:47 CEST 2012
164 - be more aggressive about exiting like perl does - formerly,
165 exiting from the non-main thread would not execute END blocks.
166
1676.07 Fri Nov 11 21:21:48 CET 2011
168 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
169 - update ecb.h.
170
1716.06 Mon Aug 8 23:59:48 CEST 2011
172 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
173 backend, which is an order of magnitude slower. unfortunately, cygwins
174 pthread implementation isn't very complete either, so allocate the stack
175 twice just to be sure.
176 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
177 in its ithreaded perl - best recompile cygwin and use the 'w'indows
178 backend for much better performance. also disable ithreads for
179 even better performance...).
180
1816.05 Thu Aug 4 21:36:36 CEST 2011
182 - blush, condvar values would not be propagated from send to recv anymore
183 (reported by Chip Salzenberg).
184 - use exponential increase for the readline buffer length in
185 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
186
1876.04 Wed Aug 3 17:15:45 CEST 2011
188 - use even more efficient and more compatible condvars for
189 compatibility to AnyEvent 6.x :)
190 - more inconsequential ecb.h updates.
191
1926.03 Wed Aug 3 11:41:30 CEST 2011
193 - change how Coro patches AnyEvent condvars for compatibility to
194 AnyEvent 6.x.
195 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
196 hasn't been included.
197
1986.02 Wed Jul 13 04:35:19 CEST 2011
199 - "improve portability to Gentoo" - gentoo manages to put perl variables
200 in memory areas that are farther than 2gb apart, which the jit couldn't
201 handle and barfed. now it's just a bit slower on gentoo and similar
202 systems.
203
2046.01 Sun Jul 3 12:31:14 CEST 2011
205 - workarounds are good, but the test for whether pthreads are used
206 was not good. this one should be better.
207 - check differently whether gcc generates cfi instructions itself.
208
2096.0 Wed Jun 29 19:43:35 CEST 2011
210 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
211 destroyed and cleaned up automatically (e.g. async { schedule }).
212 - implement a JIT compiler for part of the thread switch code,
213 which gives a 50% speed improvement on threaded perls, and
214 about 4% on non-threaded perls (so threaded perls now finally
215 reach about half the speed of non-threaded perls).
216 - slightly modernise Coro::Intro, add section about rouse functions.
217 - avoid DEFSV and ERRSV, giving another 10% improvement
218 in thread switching.
219 - Coro::State->is_destroyed is now called is_zombie.
220 - implement a Coro->safe_cancel method that might fail, but
221 cancels in a "safer" way if it succeeds.
222 - add preliminary support for DEBUGGING perls.
223 - get rid of two hash-accesses when initialising a new Coro - this
224 speeds up coro creation by almost a factor of two.
225 - croak when a coro that is being cancelled tries to block
226 (e.g. while executing a guard block), instead of crashing or
227 deadlocking.
228 - use a more robust and also faster method to identify Coro::State
229 objects - speeds up everything a bit.
230 - implement Coro->cancel in XS for a 20% speed improvement, and to
231 be able to implement mutual cancellation.
232 - speed up context switches by a percent or two by more efficiently
233 allocating context stack entries.
234 - implement Coro->join and Coro->on_destroy in XS for a speedup and
235 a reduction in memory use.
236 - cancelling a coro while it itself is cancelling another coro is
237 now supported and working, instead of triggering an assertion.
238 - be a bit more crash-resistant when calling (buggy) on_destroy
239 callbacks (best effort).
240 - move on_destroy into the slf_frame, to allow extension slf
241 functions to have destructors.
242 - get rid if coro refcounting - simply crash in other interpreter
243 threads by nulling the pointers on clone.
244 - simplify warn/die hook handling when loading Coro - the convoluted
245 logic seems to be no longer neccessary.
246 - use libecb instead of our own home-grown gcc hacks.
247 - document alternatives to Coro::LWP. Please use them :)
248 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
249 sigaltstack. Really. wine suffers from it, erlang suffers from it,
250 and it's known since at least 2006.
251
2525.372 Wed Feb 23 06:14:30 CET 2011
253 - apparently mingw doesn't provide a working gettimeofday, try to
254 work around that by relying on Time::HiRes (indirectly brought to
255 my attention by Max Maischein).
256 - fix some portability issues when Time::HiRes was used.
257
2585.371 Mon Feb 21 14:36:08 CET 2011
259 - backport to windows process emulation code again.
260
2615.37 Sat Feb 19 07:49:44 CET 2011
262 - add a big "Coro thread life cycle" section to "man Coro".
263 - try a tentative workaround against the breakage that 5.13 has
264 introduced without depreciation period. sigh.
265 - no longer use Time::HiRes if gettimeofday is available, which
266 saves quite a lot of memory.
267
2685.36 Sun Feb 13 05:33:41 CET 2011
269 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
270 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
271 method call.
272 - undocument Coro::Timer::sleep and obsolete whole module.
273 - optimise Coro::Timer::timeout memory and cpu usage.
274 - slightly updated Coro::Intro for recent changes.
275 - do not initialise PL_dirty anymore.
276
2775.25 Thu Nov 11 01:08:39 CET 2010
278 - try a different approach on netbsd - netbsd 5 finally has marginally
279 working pthreads, but still broken ucontext/sigaltstack.
280 - openbsd 4.8 finally got their act together, Coro works out of the box
281 with asm, setjmp and pthreads (no change, just informational).
282
2835.24 Sat Oct 23 11:27:12 CEST 2010
284 - port to the EV 4.0 API.
285 - work around bugs in mingw32, making strawberry perl work
286 out of the box.
287 - correctly modify Coro::AIO function prototypes
288 so that they reflect the "no optional parameters" rule.
289 - "ported" libcoro to C++.
290
2915.23 Mon May 17 18:50:42 CEST 2010
292 - be more resistant to ordering changes when initialising
293 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
294 Waldorf).
295 - document that perl 5.12 deliberately removed support for cloning.
296
2975.22 Wed Apr 14 03:55:35 CEST 2010
298 - correctly return udnef on errors in Coro::Handle::read/write
299 (testcase by Marc Mims).
300 - convert Coro::Util into a "perl compatibility wrapper" - the functions
301 are less useful now, but are drop-in replacements for existing
302 functions, listing better alternatives in the documentation. This also
303 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
304 with Coro::Util::inet_aton.
305 - do not override $Coro::idle unconditionally in Coro.pm, as other
306 modules could have provided their own idle coro already
307 (for exmaple, Coro::AnyEvent).
308 - fix Coro::Util::gethost* functions.
309 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
310
3115.21 Wed Dec 16 07:19:51 CET 2009
312 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
313 together.
314 - add some examples on how to combine other event loops with Coro in
315 Coro::AnyEvent, and how to run it (and not to block). Seems to be
316 the most common source of confusion.
317 - try to catch people naively blocking in an event callback.
318 - work around the perl filehandle bug issue in conjunction with
319 older common::sense (as indirectly pointed out by ZSystem).
320 - clarify the "not from signal handlers" section.
321
3225.2 Sun Oct 4 14:54:24 CEST 2009
323 - Coro::Storable destroyed the prototypes of the functions it wrapped.
324 - export rouse_cb and rouse_wait by default now.
325 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
326 - new method $state->swap_sv.
327 - added section on "windows process emulation" to the manpage,
328 after a not-so-fruitful (nor-friendly) "discussion" with chip
329 salzenberg (discussion implies arguments, but his only arguments
330 were ad-hominems, one wonders why he started it in the first
331 place). I hope this explains it well enough for him to understand,
332 and maybe well enough for others to understand.
333 - use common::sense everywhere now.
334 - idle callbacks are no longer supported, use idle coros instead.
335 - print a thread listing when a deadlock is detected.
336
3375.17 Sat Aug 22 23:09:31 CEST 2009
338 - work around a bug in the perl debugger causing crashes
339 when running under the debugger by marking _pool_handler as nodebug.
340 - speed up Coro::async considerably.
341 - try some hacks to get netbsd to work "more often" - their broken
342 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
343 - convert Coro to AE 5.0 API.
344
3455.162 Tue Jul 28 04:04:03 CEST 2009
346 - perl 5.8.2 is now minimum requirement.
347 - skip t/19_handle.t on broken windows perls.
348
3495.161 Wed Jul 22 04:47:38 CEST 2009
350 - Coro::AnyEvent::poll could have a different prototype when EV was
351 used as backend (analyzed by Tatsuhiko Miyagawa).
352 - Coro::AnyEvent errornously initialised the event loop when loaded,
353 not on demand.
354 - try to workaround rare */t/01_unblock.t failures.
355
3565.16 Tue Jul 21 01:44:37 CEST 2009
357 - Coro::AnyEvent failed to hook into the event loop
358 when no threads had been readied between detecting
359 the event loop and actually running it.
360 - considerably speed up Coro::Select by taking avdantage
361 of AnyEvent > 4.8 and some other optimisations.
362 - implement paragraph readline mode in Coro::Handle
363 (based on patches by Zsbán Ambrus).
364 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
365 and analyzed by Yasuhiro MATSUMOTO).
366 - clarified libcoro license and copyright.
367 - someone stole my EXTRA_META!!!
368 - implement Coro::Select::patch_pp_sselect and it's brother,
369 for hardcode select overriding.
370
3715.151 Mon Jul 6 05:41:57 CEST 2009
372 - backport to windows process emulation code again (patch by
373 Yasuhiro MATSUMOTO).
374 - slightly update Coro::MakeMaker.
375
3765.15 Tue Jun 30 10:28:06 CEST 2009
377 - deprecate Coro::Socket, document how to get ipv6 support via
378 AnyEvent::Socket instead.
379 - implement signal->wait ($cb) interface, similar to semaphores.
380 - work around SvOK not supporting getmagic, so we have to getmagic
381 to test for undef :/ (reported by Matthias Waldorf).
382 - load Coro::AnyEvent in all modules using AnyEvent.
383 - work around perl corrupting our internal data structures,
384 reported by Tokuhiro Matsuno.
385 - enable per-coroutine real and cpu time gathering
386 (Coro::State::enable_times).
387
3885.14 Wed Jun 24 01:37:48 CEST 2009
389 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
390 as this seems to have been a difficult concept (poll, sleep, idle,
391 idle_upto).
392 - add Coro::AnyEvent::readable/writable functions.
393 - clarify Coro::EV/Event/AnyEvent manpages.
394 - free per-thread global scalars in the thread calling ->cancel, to
395 avoid crashes when $_, $@ etc., are magical but some of those
396 had already been freed.
397 - "unexperimentalise" the callback interface for Coro::Semaphore.
398 - speed up ready queue management/context switching by using a linked
399 list instead of an array (~5-10%).
400 - implement "watch" command in Coro::Debug shells.
401 - for fun, implement time-slicing as an example in the manpage.
402 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
403 Coro::Event, use the normal AnyEvent handling instead of dieing
404 (the same is true for Coro::Handle).
405 - properly document Coro::EV::timed_io_once.
406 - avoid unneccessary ->cancel calls in Coro::Handle.
407 - maybe make it work on mingw32 with win32 backend
408 (based on patch by Yasuhiro Matsumoto).
409
4105.132 Fri May 29 09:00:39 CEST 2009
411 - do not keep a reference to the argument itself in
412 Coro::Semaphore::guard, as it could change later.
413 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
414 should support a prepare callback.
415
4165.131 Mon Mar 16 23:20:37 CET 2009
417 - implement and document Coro->suspend, Coro->resume.
418 - fix Coro::Select implementation to not (often) close
419 the passed file descriptors (testcase provided by pippijn).
420
4215.13 Mon Dec 15 21:51:42 CET 2008
422 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
423 - don't set diehook to C<undef> but instead to NULL, to avoid
424 spurious warnings.
425 - fix a lot of bugs in Coro::SemaphoreSet.
426 - Coro::SemaphoreSet will less often create a semaphore needlessly.
427 - add Coro::SemaphoreSet::count and wait methods.
428 - take advantage of the new Guard module.
429 - deprecate Coro::guard.
430 - try to fix the dreaded 01_unblock tests once more. I hate it when
431 testsuites need more fixing than the code they are supposed to test.
432 - croak in more cases when a required callback isn't resolvable.
433 - fix some minor issues in Coro::State->call/eval.
434 - use current coroutine context instead of a temporary one
435 when temporarily switching to another coroutine.
436 - do not call C-level on_destroy handlers during global destruction,
437 to avoid needless segfaults.
438
4395.12 Sun Dec 7 13:30:38 CET 2008
440 - add default config for MirOS, which seems to be a bug-to-bug
441 compatible fork of openbsd ("world domination by releasing
442 openbsd cvs before the openbsd folks do it" or so :).
443 - free_padlist did destroy the names pad, not good, but didn't
444 seem to bother perl - this could fix issues such as eval ""
445 inside a function called from multiple coroutines.
446 - use a different method to detect destruction time.
447 - be more careful when freeing padlists just before global
448 destruction.
449 - fixed and expanded the call/cc example.
450 - renamed _terminate to _coro_run.
451 - new method Coro::Channel->shutdown.
452 - try pthreads on openbsd <4.4 (broken sigaltstack, will
453 pthreads fare better?).
454 - be less picky about destroying "a" running coroutine.
455
4565.11 Tue Nov 25 21:49:05 CET 2008
457 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
458 working code with failed assertions, introducing more bugs than
459 it fixes, requiring elaborate workarounds :(
460
4615.1 Mon Nov 24 08:54:59 CET 2008
462 - wrote a small introductory tutorial - Coro::Intro.
463 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
464 - Coro::Select did errornously dup the file descriptors
465 and didn't work with all AnyEvent backends.
466 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
467 LWP data transfers.
468 - disassociate c contexts from coro objects - this is agruably more
469 correct, but mostly allows sharing of cctxs between coro and state
470 objects, for added memory savings and speed increases.
471 - bumped $Coro::POOL_RSS up to 32kb by default.
472 - no longer set the optype to OP_CUSTOM, as B::* understandably
473 doesn't like this very much (and we *are* a type of entersub).
474 - implement state cloning, just to prove that call/cc can be done.
475 - automatically load Coro::AnyEvent in Coro::Handle.
476 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
477 do not have this method (and don't need it either).
478 - speed up generic anyevent methods in Coro::Handle by using rouse
479 callbacks.
480 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
481 others. It also makes the debugger happier, as you can trace
482 through the idle threads now.
483 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
484 - eg/event was pretty broken.
485 - better 5.8.6 compatibility.
486
4875.0 Thu Nov 20 10:35:05 CET 2008
488 - NEW ARCHITECTURE: use the latest 4.x version if you experience
489 stability issues.
490 - bump API version to 7 - all dependents must be recompiled.
491 - removed timed_* functions - they were not being used anyways
492 and should be replaced by a more generic mechanism -
493 and were annoying to support anyways :)
494 - removed SemaphoreSet's waiter method - use sem method instead.
495 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
496 - async_pool did free a scalar value twice
497 ("Attempt to unreference...").
498 - fix a longstanding bug where calling terminate on a coro that
499 was waiting for a semaphore that was just becoming available
500 would cause a deadlock (semaphore would get into a state where
501 it was available but waiters were still blocked).
502 - calling throw on a coroutine that is waiting for a semaphore will
503 no longer make it acquire the semaphore (and thus leak a count).
504 - perl's process emulation is now not even theoretically supported
505 anymore.
506 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
507 conversion of callback-style to blocking-style.
508 - new methods $coro->schedule_to and ->cede_to, to specifically
509 schedule or cede to a specific coroutine.
510 - new function Coro::Semaphore::wait.
511 - use named constants in Coro::Channel (Richard Hundt).
512 - directly patch the entersub opcode calling SLF functions (cede,
513 transfer and so on). this does speed up context switching, but
514 more importanly, it frees us from the hardcoded behaviour of
515 entersub, so we might actually be able to return something from
516 those functions and atcually create new ones.
517 - take advantage of __builtin_frame_address on gcc.
518 - expose THX in coroapi (not sure whether this was a wise decision,
519 as "threaded" perls are running at half speed anyways).
520 - implement execute_slf (schedule-like-function) interface that makes
521 it possible to implement schedule-like-functions in XS.
522 - use new SLF interface to massively speed up Coro::EV by roughly a
523 factor of two.
524 - used new SLF interface to massively speed up Coro::Semaphore by a
525 factor of three.
526 - used new SLF interface to speed up Coro::AIO by roughly a factor of
527 four and reduce its memory usage considerably.
528 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
529 for a nice speedup and vastly more correct behaviour. Also implement
530 a new method "sem" to get at the underlying semaphore object.
531 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
532 (in comparison) 20-40% speedup.
533 - used new SLF interface to reimplement Coro::Signal gaining
534 some unknown (because I was too lazy), but certain, speedup, and also
535 making signals reliable for the first time.
536 - used new SLF interface and other optimisations to speed up async_pool
537 by a factor of two. It also doesn't rely on perl's exception mechanism
538 to exit anymore. The overhead for terminating an async_pool, coro over
539 a normal async is now very small.
540 - sped up coroutine creation/destruction by 40%.
541 - forgot to include Coro/libcoro/README in the dist for all these years.
542 - work around a freebsd pthreads bug (manual testcancel is required as
543 pthread_cond_wait isn't a cancellation point on freebsd).
544 - use new rouse functions to speed up and simplify Coro::BDB.
545 - make "prefer perl native functions" work with threaded perls.
546 - condense Coro::Debug ps output, hint at v and w flags.
547 - (libcoro) lots of minor cleanups and portability improvements.
548
5494.914 Wed Nov 19 12:54:18 CET 2008
550 - fix a disastrous bug in the readline optimisation
551 introduced in 4.801.
552
5534.913 Sat Nov 15 07:58:28 CET 2008
554 - async_pool did free a scalar value twice
555 ("Attempt to unreference...").
556
5574.912 Thu Nov 13 18:31:23 CET 2008
558 - minor cleanups.
559 - use much larger stacks on linux and perl < 5.8.8.
560 - Coro::Debug::new_unix_server did not unlink the socket
561 when destroyed.
562
5634.911 Tue Nov 11 04:26:17 CET 2008
564 - "port" to threaded perls.
565
5664.91 Mon Nov 10 05:36:38 CET 2008
567 - the ->throw exception object no longer leaks.
568 - creating a new cctx leaked a scopestack entry (memleak).
569 - new coroutines didn't get created with a zero flags field
570 (unknown impact).
571 - calling ->throw on a not-yet-started coroutine should now work
572 instead of being ignored.
573 - ->throw is now supported on Coro::State objects.
574 - clean up cctx creation code a bit.
575 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
576
5774.9 Sat Nov 8 17:45:27 CET 2008
578 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
579 - (libcoro) no longer leak threads in the experimental pthread backend,
580 also speed it up considerably.
581 - (libcoro) do not rely on makecontext passing void *'s unscathed.
582 - fix compiletime dependencies on libcoro in the Makefile.
583 - cctx_count wasn't always updated properly.
584 - Coro::State::cctx_stacksize wasn't applied correctly.
585 - new function Coro::State::cctx_max_idle.
586 - the default max number of idle C contexts is now 4.
587 - (libcoro) try harder to get _setjmp/_longjmp.
588 - (libcoro) cleanup and extend the libcoro API to officially
589 allow the creation of empty source contexts.
590 - very experimental workaround for the totally broken netbsd platform.
591 - tried to hack around openbsd bugs.
592
5934.804 Wed Nov 5 16:36:00 CET 2008
594 - Coro::Debug::new_unix_server would not create a non-blocking listening
595 socket, sometimes causing freezes.
596 - (libcoro) fix misaligned stack points for setjmp and assembly
597 methods, which can cause crashes on x86/x86_64 with a sufficiently
598 aggressive compiler.
599 - new function: Coro::Debug::new_tcp_server.
600 - move ->throw into the Coro class because it only works on coro objects.
601
6024.803 Mon Nov 3 17:16:12 CET 2008
603 - (libcoro) use a global asm statement to become independent of gcc
604 otpimisations for CORO_ASM (thanks to pippijn for the idea).
605 - try to workaround yet another broken bsd, this time dragonfly.
606
6074.802 Thu Oct 30 10:56:12 CET 2008
608 - support -fno-omit-frame-pointer on x86 with the assembly method.
609 - tune 01_unblock.t tests a bit.
610
6114.801 Wed Oct 22 18:33:37 CEST 2008
612 - improve readline speed for very long "lines".
613 - backport to 5.8.8.
614
6154.8 Thu Oct 2 13:34:40 CEST 2008
616 - new function Coro::AIO::aio_wait.
617 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
618 - greatly speed up and reduce memory usage of Coro::AIO requests.
619 - implement some other µ-optimisations.
620
6214.749 Mon Sep 29 14:40:12 CEST 2008
622 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
623
6244.748 Sat Sep 27 14:03:19 CEST 2008
625 - implement, but do not document, PerlIO::cede(granularity).
626 - Coro::Storable forgot to wrap Storable::pstore.
627 - work around the multitude of leaks and memory corruption
628 bugs in PerlIO::via by using our own C-level perliol. As a side
629 effect, Coro::Storable is now much, much, much faster.
630
6314.747 Tue Sep 23 01:59:41 CEST 2008
632 - fix a per-cv memleak (one empty array was leaked per
633 code reference).
634 - avoid a crash in coro->call|rss when the coroutine was already
635 destroyed (most noticably when using Coro::Debug::ps :)
636 - also protect *Storable::FILE.
637 - push up default storable granularity to 20ms.
638
6394.746 Sun Sep 21 03:22:20 CEST 2008
640 - be more insistent on locking Storable against reentrancy
641 in Coro::Storable.
642 - move swap_def?v and throw to Coro::State, as documented.
643
6444.745 Thu Jul 24 00:14:38 CEST 2008
645 - remove debugging code related to MgPV_nolen_const, also try to
646 make it compile with perl 5.8.6 (yes, apple apparently loves
647 outdated software). Reported by John S.
648
6494.744 Tue Jul 8 22:06:35 CEST 2008
650 - correctly provide default DIE/WARN handlers as documented.
651 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
652 implementations inside perl work fine for us.
653
6544.743 Mon Jun 16 00:21:57 CEST 2008
655 - when using Coro::EV without running EV::loop it could
656 result in busy-waiting for events, this has been fixed.
657 - reduce codesize and improve performance by using EV_DEFAULT_UC.
658
6594.742 Sat May 31 14:10:21 CEST 2008
660 - implement a workaround for (some) perl <5.8.8 versions.
661 - require EV 3.3+.
7 662
84.741 Fri May 30 23:33:09 CEST 2008 6634.741 Fri May 30 23:33:09 CEST 2008
9 - tell netbsd how utterly broken their imitation of an OS is 664 - tell netbsd how utterly broken their imitation of an OS is
10 and refuse to build by default if pthreads are in use. 665 and refuse to build by default if pthreads are in use.
11 - switch to "s" method on all bsd's by default, as their ucontext 666 - switch to "s" method on all bsd's by default, as their ucontext
… …
22 - sprinkle "no warnings" freely over everything, also suppress 677 - sprinkle "no warnings" freely over everything, also suppress
23 warnings for some other modules. 678 warnings for some other modules.
24 - fix typo in WSAEWOULDBLOCK. 679 - fix typo in WSAEWOULDBLOCK.
25 680
264.72 Sun May 25 05:14:36 CEST 2008 6814.72 Sun May 25 05:14:36 CEST 2008
27 - tweak META.yaml a bit, unfortunately, there is no documented way 682 - tweak META.yaml a bit, unfortunately, there is no documented way
28 to have optional dependencies with CPAN. doh :( 683 to have optional dependencies with CPAN. doh :(
29 - avoid running some tests on windows because they would fail due to 684 - avoid running some tests on windows because they would fail due to
30 perl bug (broken fork, broken pipes...). 685 perl bug (broken fork, broken pipes...).
31 - work around perl on windows bugs where perl returns undocumented 686 - work around perl on windows bugs where perl returns undocumented
32 error codes for sysread, syswrite etc. by taking advantage 687 error codes for sysread, syswrite etc. by taking advantage
… …
36 - use unix domain sockets in testsuite to work around 691 - use unix domain sockets in testsuite to work around
37 common perl implementation bugs on widows (they are emulated by 692 common perl implementation bugs on widows (they are emulated by
38 tcp sockets on windows. ugh.) 693 tcp sockets on windows. ugh.)
39 694
404.71 Sat May 24 20:01:27 CEST 2008 6954.71 Sat May 24 20:01:27 CEST 2008
41 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 696 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
42 - take advantage of async name resolution of AnyEvent::Util. 697 - take advantage of async name resolution of AnyEvent::Util.
43 - work around brutal inet_aton override in Coro::LWP. 698 - work around brutal inet_aton override in Coro::LWP.
44 - take advantage of the readyhook in Coro::EV, for smoother 699 - take advantage of the readyhook in Coro::EV, for smoother
45 scheduling. 700 scheduling.
46 701
474.7 Sun May 11 00:32:19 CEST 2008 7024.7 Sun May 11 00:32:19 CEST 2008
48 - completely reworked the Coro manpage. 703 - completely reworked the Coro manpage.
49 - added Coro::AnyEvent, generic event loop integration. 704 - added Coro::AnyEvent, generic event loop integration.
50 - implement cancel, ready and kill commands in Coro::Debug. 705 - implement cancel, ready and kill commands in Coro::Debug.
51 - document find_coro in Coro::Debug. 706 - document find_coro in Coro::Debug.
52 - incompatible change: rename has_stack to has_cctx. 707 - incompatible change: rename has_stack to has_cctx.
53 - Coro::AIO and Coro::BDB no longer force event model detection, 708 - Coro::AIO and Coro::BDB no longer force event model detection,
54 use AnyEvent::AIO and AnyEvent::BDB. 709 use AnyEvent::AIO and AnyEvent::BDB.
… …
178 833
1794.11 Thu Oct 11 02:40:24 CEST 2007 8344.11 Thu Oct 11 02:40:24 CEST 2007
180 - port to threaded perls. 835 - port to threaded perls.
181 836
1824.1 Thu Oct 11 02:38:16 CEST 2007 8374.1 Thu Oct 11 02:38:16 CEST 2007
183 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 838 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
184 be local to each coro (see Coro::State). 839 be local to each coro (see Coro::State).
185 - incompatible change: for very deep reasons, cede and cede_notself 840 - incompatible change: for very deep reasons, cede and cede_notself
186 cannot return anything, so nothing will be returned. 841 cannot return anything, so nothing will be returned.
187 - possibly bring back 5.10 compatibility (untested). 842 - possibly bring back 5.10 compatibility (untested).
188 - work around stupid (and wrong) warning on 5.10 :(. 843 - work around stupid (and wrong) warning on 5.10 :(.
… …
260 but stateful semantics. 915 but stateful semantics.
261 - fixed a lot of typos in Coro.pm (patch submitted by David 916 - fixed a lot of typos in Coro.pm (patch submitted by David
262 Steinbrunner, which applied flawlessly). 917 Steinbrunner, which applied flawlessly).
263 918
2643.6 Sat Apr 14 17:13:31 CEST 2007 9193.6 Sat Apr 14 17:13:31 CEST 2007
265 - added some bugfixes to get eg/myhttpd working again. 920 - added some bugfixes to get eg/myhttpd working again.
266 - added Coro::Storable for often-cede'ing freeze/thaw. 921 - added Coro::Storable for often-cede'ing freeze/thaw.
267 - try to do a clean exit when a coroutine calls exit 922 - try to do a clean exit when a coroutine calls exit
268 (EXPERIMENTAL). 923 (EXPERIMENTAL).
269 - got rid of indirect call through _coro_init. 924 - got rid of indirect call through _coro_init.
270 - updated the partly antique examples in eg/ to 925 - updated the partly antique examples in eg/ to

Diff Legend

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