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

Comparing Coro/Changes (file contents):
Revision 1.357 by root, Thu May 29 18:05:53 2008 UTC vs.
Revision 1.646 by root, Tue Jul 7 00:33:25 2015 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_svs - NVs can be in the SV head beginning with 5.20.
11
126.46 Tue Jun 30 14:36:30 CEST 2015
13 - restore portability to perls with windows fork emulation
14 (patch by Petr Písař).
15
166.45 Tue Jun 30 01:40:08 CEST 2015
17 - restore portability to perl 5.16 and below.
18
196.44 Tue Jun 30 00:41:54 CEST 2015
20 - avoid segfaulting (or worse) when tracing a canceled thread.
21 - the Event module silently broke it's public hook API, causing
22 failures on perls compiled with -Duselongdouble. Requiring
23 the latest version of Event because I don't know which release
24 changed this (it's not mentioned in the ChangeLog).
25 - allow xs level enterleave hooks via CoroAPI.
26
276.43 Thu Jun 4 15:38:14 CEST 2015
28 - use stability canary.
29 - port to stableperl-5.22.0-1.001.
30 - update libecb.
31
326.42 Wed Feb 11 20:29:52 CET 2015
33 - Coro::SemaphoreSet->try did not actually work (analyzed by
34 SATO Kentaro).
35 - upgrade libecb to be C11 compliant.
36
376.41 Sat Sep 6 22:08:46 CEST 2014
38 - restore portability to perl 5.8.x.
39 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
40 up some memory, but fixes "use feature" and similar modules
41 when used inside a Coro.
42 - allow zero as argument to Coro::Channel to mean the same thing
43 as no argument. this works with older versions as well,
44 but wasn't legal until now.
45 - slightly better c header file detection.
46
476.39 Mon Jun 2 00:00:08 CEST 2014
48 - work around more incompatible changes in 5.20.
49
506.38 Sun Jun 1 21:54:23 CEST 2014
51 - check that perl slots actually have enough space to hold
52 interpreter variables.
53 - untested port to perl 5.19 (64 bit tmps indices) (reported
54 by Andreas König).
55 - croak when cancel is called without a thread context.
56
576.37 Tue Mar 4 13:27:33 CET 2014
58 - *sigh*, removed leftover debugging code from debugging a
59 perl bug, of all things.
60
616.36 Tue Mar 4 07:11:59 CET 2014
62 - semaphores would not clear the destroy hook when interrupted
63 by ->throw, causing segfaults or worse.
64 - ->throw on a thread waiting for a semaphore did not acquire
65 the semaphore, but also didn't wake up other waiters,
66 possibly causing a deadlock.
67 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
68 to report a full stacktrace, which should help find out
69 where the actual call is.
70 - "atomically" destroy data in slf_destroy, because it is
71 the right thing to do, just in case.
72 - disable ecb.h libm dependency, to compile on ancient systems
73 or under adverse conditions.
74
756.33 Mon Nov 18 11:26:27 CET 2013
76 - do not crash when freeing padlists with holes (in 5.18).
77 - tentative SVt_BIND 5.19 port/fix.
78
796.32 Tue Nov 5 15:35:35 CET 2013
80 - use a new algorithm to derive padlists for perl 5.18. The old
81 one could lead to 0-pointer accesses inside perl (reported
82 by Darin McBride).
83
846.31 Thu May 9 07:39:48 CEST 2013
85 - Coro::AIO requests would crash if the thread was ready'd
86 while the request was ongoing.
87
886.29 Wed May 8 02:55:18 CEST 2013
89 - when an on_destroy handler destructs the coro currently being
90 destructed a perl scalar could be accessed after being freed,
91 likely causing a crash.
92
936.28 Wed Mar 6 06:58:02 CET 2013
94 - clean remnants of existing __DIE__ and __WARN__ handlers so
95 they lose their magic and will not cause segfaults later
96 (testcase by Andrey Sagulin).
97 - improved Coro::State documentation a bit.
98 - Coro::Debug::command now flushes the output.
99 - add hack detection code for x32 abi, because the braindead slugs
100 who designed that made it look exactly like x86_64 without
101 providing proper compile time symbols to test for it. as a result,
102 this detection cannot work reliably.
103 - valgrind stack registering was broken.
104 - do not rely on Time::HiRes anymore in Coro::Debug.
105
1066.23 Fri Dec 7 23:36:37 CET 2012
107 - use experimental fiber implementation on native windows
108 perls.
109 - use sizeof (void *) as multiplication factor for stack sizes,
110 to accomodate the totally braindamaged microsoft 64 bit "os".
111 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
112 activeperl 5.16 crashes when PerlIO_define_layer is called due
113 to some bug in the perl dll, strawberry perl at least passes
114 the testsuite.
115 - implement Coro::Handle->peeraddr/host/port, for slightly
116 improved compatibility with LWP.
117 - implement 5.17 compatibility by almost blindly applying a
118 good-looking patch by Father Chrysostomos.
119 - move stack management functions into libcoro 3.
120 - libcoro version 3 "released".
121 - support magic values as timed_io_once args.
122 - recommend AnyEvent 7+ or EV 4+, also require EV
123 version 4 or newer for Coro::EV.
124
1256.10 Tue Oct 9 01:14:27 CEST 2012
126 - updated ecb.h, it had a typo that caused it to not compile on many
127 big endian systems (reported by many people).
128 - disable memory fences in ecb.h to improve portability.
129
1306.09 Sat Oct 6 23:25:02 CEST 2012
131 - Coro::EV I/O watchers were not interruptible by exceptions
132 (Coro::State::throw) (testcase by sten).
133 - ->throw now puts threads into the ready queue, as this seems to
134 be expected by existing code, and code that doesn't cope with spurious
135 wakeups needs fixing anyway.
136 - use fd -1 in mmap.
137 - cast I32 to int in error message printf.
138 - warn about broken so-called "hardened" kernels.
139
1406.08 Fri Apr 13 12:05:47 CEST 2012
141 - be more aggressive about exiting like perl does - formerly,
142 exiting from the non-main thread would not execute END blocks.
143
1446.07 Fri Nov 11 21:21:48 CET 2011
145 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
146 - update ecb.h.
147
1486.06 Mon Aug 8 23:59:48 CEST 2011
149 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
150 backend, which is an order of magnitude slower. unfortunately, cygwins
151 pthread implementation isn't very complete either, so allocate the stack
152 twice just to be sure.
153 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
154 in its ithreaded perl - best recompile cygwin and use the 'w'indows
155 backend for much better performance. also disable ithreads for
156 even better performance...).
157
1586.05 Thu Aug 4 21:36:36 CEST 2011
159 - blush, condvar values would not be propagated from send to recv anymore
160 (reported by Chip Salzenberg).
161 - use exponential increase for the readline buffer length in
162 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
163
1646.04 Wed Aug 3 17:15:45 CEST 2011
165 - use even more efficient and more compatible condvars for
166 compatibility to AnyEvent 6.x :)
167 - more inconsequential ecb.h updates.
168
1696.03 Wed Aug 3 11:41:30 CEST 2011
170 - change how Coro patches AnyEvent condvars for compatibility to
171 AnyEvent 6.x.
172 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
173 hasn't been included.
174
1756.02 Wed Jul 13 04:35:19 CEST 2011
176 - "improve portability to Gentoo" - gentoo manages to put perl variables
177 in memory areas that are farther than 2gb apart, which the jit couldn't
178 handle and barfed. now it's just a bit slower on gentoo and similar
179 systems.
180
1816.01 Sun Jul 3 12:31:14 CEST 2011
182 - workarounds are good, but the test for whether pthreads are used
183 was not good. this one should be better.
184 - check differently whether gcc generates cfi instructions itself.
185
1866.0 Wed Jun 29 19:43:35 CEST 2011
187 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
188 destroyed and cleaned up automatically (e.g. async { schedule }).
189 - implement a JIT compiler for part of the thread switch code,
190 which gives a 50% speed improvement on threaded perls, and
191 about 4% on non-threaded perls (so threaded perls now finally
192 reach about half the speed of non-threaded perls).
193 - slightly modernise Coro::Intro, add section about rouse functions.
194 - avoid DEFSV and ERRSV, giving another 10% improvement
195 in thread switching.
196 - Coro::State->is_destroyed is now called is_zombie.
197 - implement a Coro->safe_cancel method that might fail, but
198 cancels in a "safer" way if it succeeds.
199 - add preliminary support for DEBUGGING perls.
200 - get rid of two hash-accesses when initialising a new Coro - this
201 speeds up coro creation by almost a factor of two.
202 - croak when a coro that is being cancelled tries to block
203 (e.g. while executing a guard block), instead of crashing or
204 deadlocking.
205 - use a more robust and also faster method to identify Coro::State
206 objects - speeds up everything a bit.
207 - implement Coro->cancel in XS for a 20% speed improvement, and to
208 be able to implement mutual cancellation.
209 - speed up context switches by a percent or two by more efficiently
210 allocating context stack entries.
211 - implement Coro->join and Coro->on_destroy in XS for a speedup and
212 a reduction in memory use.
213 - cancelling a coro while it itself is cancelling another coro is
214 now supported and working, instead of triggering an assertion.
215 - be a bit more crash-resistant when calling (buggy) on_destroy
216 callbacks (best effort).
217 - move on_destroy into the slf_frame, to allow extension slf
218 functions to have destructors.
219 - get rid if coro refcounting - simply crash in other interpreter
220 threads by nulling the pointers on clone.
221 - simplify warn/die hook handling when loading Coro - the convoluted
222 logic seems to be no longer neccessary.
223 - use libecb instead of our own home-grown gcc hacks.
224 - document alternatives to Coro::LWP. Please use them :)
225 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
226 sigaltstack. Really. wine suffers from it, erlang suffers from it,
227 and it's known since at least 2006.
228
2295.372 Wed Feb 23 06:14:30 CET 2011
230 - apparently mingw doesn't provide a working gettimeofday, try to
231 work around that by relying on Time::HiRes (indirectly brought to
232 my attention by Max Maischein).
233 - fix some portability issues when Time::HiRes was used.
234
2355.371 Mon Feb 21 14:36:08 CET 2011
236 - backport to windows process emulation code again.
237
2385.37 Sat Feb 19 07:49:44 CET 2011
239 - add a big "Coro thread life cycle" section to "man Coro".
240 - try a tentative workaround against the breakage that 5.13 has
241 introduced without depreciation period. sigh.
242 - no longer use Time::HiRes if gettimeofday is available, which
243 saves quite a lot of memory.
244
2455.36 Sun Feb 13 05:33:41 CET 2011
246 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
247 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
248 method call.
249 - undocument Coro::Timer::sleep and obsolete whole module.
250 - optimise Coro::Timer::timeout memory and cpu usage.
251 - slightly updated Coro::Intro for recent changes.
252 - do not initialise PL_dirty anymore.
253
2545.25 Thu Nov 11 01:08:39 CET 2010
255 - try a different approach on netbsd - netbsd 5 finally has marginally
256 working pthreads, but still broken ucontext/sigaltstack.
257 - openbsd 4.8 finally got their act together, Coro works out of the box
258 with asm, setjmp and pthreads (no change, just informational).
259
2605.24 Sat Oct 23 11:27:12 CEST 2010
261 - port to the EV 4.0 API.
262 - work around bugs in mingw32, making strawberry perl work
263 out of the box.
264 - correctly modify Coro::AIO function prototypes
265 so that they reflect the "no optional parameters" rule.
266 - "ported" libcoro to C++.
267
2685.23 Mon May 17 18:50:42 CEST 2010
269 - be more resistant to ordering changes when initialising
270 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
271 Waldorf).
272 - document that perl 5.12 deliberately removed support for cloning.
273
2745.22 Wed Apr 14 03:55:35 CEST 2010
275 - correctly return udnef on errors in Coro::Handle::read/write
276 (testcase by Marc Mims).
277 - convert Coro::Util into a "perl compatibility wrapper" - the functions
278 are less useful now, but are drop-in replacements for existing
279 functions, listing better alternatives in the documentation. This also
280 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
281 with Coro::Util::inet_aton.
282 - do not override $Coro::idle unconditionally in Coro.pm, as other
283 modules could have provided their own idle coro already
284 (for exmaple, Coro::AnyEvent).
285 - fix Coro::Util::gethost* functions.
286 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
287
2885.21 Wed Dec 16 07:19:51 CET 2009
289 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
290 together.
291 - add some examples on how to combine other event loops with Coro in
292 Coro::AnyEvent, and how to run it (and not to block). Seems to be
293 the most common source of confusion.
294 - try to catch people naively blocking in an event callback.
295 - work around the perl filehandle bug issue in conjunction with
296 older common::sense (as indirectly pointed out by ZSystem).
297 - clarify the "not from signal handlers" section.
298
2995.2 Sun Oct 4 14:54:24 CEST 2009
300 - Coro::Storable destroyed the prototypes of the functions it wrapped.
301 - export rouse_cb and rouse_wait by default now.
302 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
303 - new method $state->swap_sv.
304 - added section on "windows process emulation" to the manpage,
305 after a not-so-fruitful (nor-friendly) "discussion" with chip
306 salzenberg (discussion implies arguments, but his only arguments
307 were ad-hominems, one wonders why he started it in the first
308 place). I hope this explains it well enough for him to understand,
309 and maybe well enough for others to understand.
310 - use common::sense everywhere now.
311 - idle callbacks are no longer supported, use idle coros instead.
312 - print a thread listing when a deadlock is detected.
313
3145.17 Sat Aug 22 23:09:31 CEST 2009
315 - work around a bug in the perl debugger causing crashes
316 when running under the debugger by marking _pool_handler as nodebug.
317 - speed up Coro::async considerably.
318 - try some hacks to get netbsd to work "more often" - their broken
319 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
320 - convert Coro to AE 5.0 API.
321
3225.162 Tue Jul 28 04:04:03 CEST 2009
323 - perl 5.8.2 is now minimum requirement.
324 - skip t/19_handle.t on broken windows perls.
325
3265.161 Wed Jul 22 04:47:38 CEST 2009
327 - Coro::AnyEvent::poll could have a different prototype when EV was
328 used as backend (analyzed by Tatsuhiko Miyagawa).
329 - Coro::AnyEvent errornously initialised the event loop when loaded,
330 not on demand.
331 - try to workaround rare */t/01_unblock.t failures.
332
3335.16 Tue Jul 21 01:44:37 CEST 2009
334 - Coro::AnyEvent failed to hook into the event loop
335 when no threads had been readied between detecting
336 the event loop and actually running it.
337 - considerably speed up Coro::Select by taking avdantage
338 of AnyEvent > 4.8 and some other optimisations.
339 - implement paragraph readline mode in Coro::Handle
340 (based on patches by Zsbán Ambrus).
341 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
342 and analyzed by Yasuhiro MATSUMOTO).
343 - clarified libcoro license and copyright.
344 - someone stole my EXTRA_META!!!
345 - implement Coro::Select::patch_pp_sselect and it's brother,
346 for hardcode select overriding.
347
3485.151 Mon Jul 6 05:41:57 CEST 2009
349 - backport to windows process emulation code again (patch by
350 Yasuhiro MATSUMOTO).
351 - slightly update Coro::MakeMaker.
352
3535.15 Tue Jun 30 10:28:06 CEST 2009
354 - deprecate Coro::Socket, document how to get ipv6 support via
355 AnyEvent::Socket instead.
356 - implement signal->wait ($cb) interface, similar to semaphores.
357 - work around SvOK not supporting getmagic, so we have to getmagic
358 to test for undef :/ (reported by Matthias Waldorf).
359 - load Coro::AnyEvent in all modules using AnyEvent.
360 - work around perl corrupting our internal data structures,
361 reported by Tokuhiro Matsuno.
362 - enable per-coroutine real and cpu time gathering
363 (Coro::State::enable_times).
364
3655.14 Wed Jun 24 01:37:48 CEST 2009
366 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
367 as this seems to have been a difficult concept (poll, sleep, idle,
368 idle_upto).
369 - add Coro::AnyEvent::readable/writable functions.
370 - clarify Coro::EV/Event/AnyEvent manpages.
371 - free per-thread global scalars in the thread calling ->cancel, to
372 avoid crashes when $_, $@ etc., are magical but some of those
373 had already been freed.
374 - "unexperimentalise" the callback interface for Coro::Semaphore.
375 - speed up ready queue management/context switching by using a linked
376 list instead of an array (~5-10%).
377 - implement "watch" command in Coro::Debug shells.
378 - for fun, implement time-slicing as an example in the manpage.
379 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
380 Coro::Event, use the normal AnyEvent handling instead of dieing
381 (the same is true for Coro::Handle).
382 - properly document Coro::EV::timed_io_once.
383 - avoid unneccessary ->cancel calls in Coro::Handle.
384 - maybe make it work on mingw32 with win32 backend
385 (based on patch by Yasuhiro Matsumoto).
386
3875.132 Fri May 29 09:00:39 CEST 2009
388 - do not keep a reference to the argument itself in
389 Coro::Semaphore::guard, as it could change later.
390 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
391 should support a prepare callback.
392
3935.131 Mon Mar 16 23:20:37 CET 2009
394 - implement and document Coro->suspend, Coro->resume.
395 - fix Coro::Select implementation to not (often) close
396 the passed file descriptors (testcase provided by pippijn).
397
3985.13 Mon Dec 15 21:51:42 CET 2008
399 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
400 - don't set diehook to C<undef> but instead to NULL, to avoid
401 spurious warnings.
402 - fix a lot of bugs in Coro::SemaphoreSet.
403 - Coro::SemaphoreSet will less often create a semaphore needlessly.
404 - add Coro::SemaphoreSet::count and wait methods.
405 - take advantage of the new Guard module.
406 - deprecate Coro::guard.
407 - try to fix the dreaded 01_unblock tests once more. I hate it when
408 testsuites need more fixing than the code they are supposed to test.
409 - croak in more cases when a required callback isn't resolvable.
410 - fix some minor issues in Coro::State->call/eval.
411 - use current coroutine context instead of a temporary one
412 when temporarily switching to another coroutine.
413 - do not call C-level on_destroy handlers during global destruction,
414 to avoid needless segfaults.
415
4165.12 Sun Dec 7 13:30:38 CET 2008
417 - add default config for MirOS, which seems to be a bug-to-bug
418 compatible fork of openbsd ("world domination by releasing
419 openbsd cvs before the openbsd folks do it" or so :).
420 - free_padlist did destroy the names pad, not good, but didn't
421 seem to bother perl - this could fix issues such as eval ""
422 inside a function called from multiple coroutines.
423 - use a different method to detect destruction time.
424 - be more careful when freeing padlists just before global
425 destruction.
426 - fixed and expanded the call/cc example.
427 - renamed _terminate to _coro_run.
428 - new method Coro::Channel->shutdown.
429 - try pthreads on openbsd <4.4 (broken sigaltstack, will
430 pthreads fare better?).
431 - be less picky about destroying "a" running coroutine.
432
4335.11 Tue Nov 25 21:49:05 CET 2008
434 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
435 working code with failed assertions, introducing more bugs than
436 it fixes, requiring elaborate workarounds :(
437
4385.1 Mon Nov 24 08:54:59 CET 2008
439 - wrote a small introductory tutorial - Coro::Intro.
440 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
441 - Coro::Select did errornously dup the file descriptors
442 and didn't work with all AnyEvent backends.
443 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
444 LWP data transfers.
445 - disassociate c contexts from coro objects - this is agruably more
446 correct, but mostly allows sharing of cctxs between coro and state
447 objects, for added memory savings and speed increases.
448 - bumped $Coro::POOL_RSS up to 32kb by default.
449 - no longer set the optype to OP_CUSTOM, as B::* understandably
450 doesn't like this very much (and we *are* a type of entersub).
451 - implement state cloning, just to prove that call/cc can be done.
452 - automatically load Coro::AnyEvent in Coro::Handle.
453 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
454 do not have this method (and don't need it either).
455 - speed up generic anyevent methods in Coro::Handle by using rouse
456 callbacks.
457 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
458 others. It also makes the debugger happier, as you can trace
459 through the idle threads now.
460 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
461 - eg/event was pretty broken.
462 - better 5.8.6 compatibility.
463
4645.0 Thu Nov 20 10:35:05 CET 2008
465 - NEW ARCHITECTURE: use the latest 4.x version if you experience
466 stability issues.
467 - bump API version to 7 - all dependents must be recompiled.
468 - removed timed_* functions - they were not being used anyways
469 and should be replaced by a more generic mechanism -
470 and were annoying to support anyways :)
471 - removed SemaphoreSet's waiter method - use sem method instead.
472 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
473 - async_pool did free a scalar value twice
474 ("Attempt to unreference...").
475 - fix a longstanding bug where calling terminate on a coro that
476 was waiting for a semaphore that was just becoming available
477 would cause a deadlock (semaphore would get into a state where
478 it was available but waiters were still blocked).
479 - calling throw on a coroutine that is waiting for a semaphore will
480 no longer make it acquire the semaphore (and thus leak a count).
481 - perl's process emulation is now not even theoretically supported
482 anymore.
483 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
484 conversion of callback-style to blocking-style.
485 - new methods $coro->schedule_to and ->cede_to, to specifically
486 schedule or cede to a specific coroutine.
487 - new function Coro::Semaphore::wait.
488 - use named constants in Coro::Channel (Richard Hundt).
489 - directly patch the entersub opcode calling SLF functions (cede,
490 transfer and so on). this does speed up context switching, but
491 more importanly, it frees us from the hardcoded behaviour of
492 entersub, so we might actually be able to return something from
493 those functions and atcually create new ones.
494 - take advantage of __builtin_frame_address on gcc.
495 - expose THX in coroapi (not sure whether this was a wise decision,
496 as "threaded" perls are running at half speed anyways).
497 - implement execute_slf (schedule-like-function) interface that makes
498 it possible to implement schedule-like-functions in XS.
499 - use new SLF interface to massively speed up Coro::EV by roughly a
500 factor of two.
501 - used new SLF interface to massively speed up Coro::Semaphore by a
502 factor of three.
503 - used new SLF interface to speed up Coro::AIO by roughly a factor of
504 four and reduce its memory usage considerably.
505 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
506 for a nice speedup and vastly more correct behaviour. Also implement
507 a new method "sem" to get at the underlying semaphore object.
508 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
509 (in comparison) 20-40% speedup.
510 - used new SLF interface to reimplement Coro::Signal gaining
511 some unknown (because I was too lazy), but certain, speedup, and also
512 making signals reliable for the first time.
513 - used new SLF interface and other optimisations to speed up async_pool
514 by a factor of two. It also doesn't rely on perl's exception mechanism
515 to exit anymore. The overhead for terminating an async_pool, coro over
516 a normal async is now very small.
517 - sped up coroutine creation/destruction by 40%.
518 - forgot to include Coro/libcoro/README in the dist for all these years.
519 - work around a freebsd pthreads bug (manual testcancel is required as
520 pthread_cond_wait isn't a cancellation point on freebsd).
521 - use new rouse functions to speed up and simplify Coro::BDB.
522 - make "prefer perl native functions" work with threaded perls.
523 - condense Coro::Debug ps output, hint at v and w flags.
524 - (libcoro) lots of minor cleanups and portability improvements.
525
5264.914 Wed Nov 19 12:54:18 CET 2008
527 - fix a disastrous bug in the readline optimisation
528 introduced in 4.801.
529
5304.913 Sat Nov 15 07:58:28 CET 2008
531 - async_pool did free a scalar value twice
532 ("Attempt to unreference...").
533
5344.912 Thu Nov 13 18:31:23 CET 2008
535 - minor cleanups.
536 - use much larger stacks on linux and perl < 5.8.8.
537 - Coro::Debug::new_unix_server did not unlink the socket
538 when destroyed.
539
5404.911 Tue Nov 11 04:26:17 CET 2008
541 - "port" to threaded perls.
542
5434.91 Mon Nov 10 05:36:38 CET 2008
544 - the ->throw exception object no longer leaks.
545 - creating a new cctx leaked a scopestack entry (memleak).
546 - new coroutines didn't get created with a zero flags field
547 (unknown impact).
548 - calling ->throw on a not-yet-started coroutine should now work
549 instead of being ignored.
550 - ->throw is now supported on Coro::State objects.
551 - clean up cctx creation code a bit.
552 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
553
5544.9 Sat Nov 8 17:45:27 CET 2008
555 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
556 - (libcoro) no longer leak threads in the experimental pthread backend,
557 also speed it up considerably.
558 - (libcoro) do not rely on makecontext passing void *'s unscathed.
559 - fix compiletime dependencies on libcoro in the Makefile.
560 - cctx_count wasn't always updated properly.
561 - Coro::State::cctx_stacksize wasn't applied correctly.
562 - new function Coro::State::cctx_max_idle.
563 - the default max number of idle C contexts is now 4.
564 - (libcoro) try harder to get _setjmp/_longjmp.
565 - (libcoro) cleanup and extend the libcoro API to officially
566 allow the creation of empty source contexts.
567 - very experimental workaround for the totally broken netbsd platform.
568 - tried to hack around openbsd bugs.
569
5704.804 Wed Nov 5 16:36:00 CET 2008
571 - Coro::Debug::new_unix_server would not create a non-blocking listening
572 socket, sometimes causing freezes.
573 - (libcoro) fix misaligned stack points for setjmp and assembly
574 methods, which can cause crashes on x86/x86_64 with a sufficiently
575 aggressive compiler.
576 - new function: Coro::Debug::new_tcp_server.
577 - move ->throw into the Coro class because it only works on coro objects.
578
5794.803 Mon Nov 3 17:16:12 CET 2008
580 - (libcoro) use a global asm statement to become independent of gcc
581 otpimisations for CORO_ASM (thanks to pippijn for the idea).
582 - try to workaround yet another broken bsd, this time dragonfly.
583
5844.802 Thu Oct 30 10:56:12 CET 2008
585 - support -fno-omit-frame-pointer on x86 with the assembly method.
586 - tune 01_unblock.t tests a bit.
587
5884.801 Wed Oct 22 18:33:37 CEST 2008
589 - improve readline speed for very long "lines".
590 - backport to 5.8.8.
591
5924.8 Thu Oct 2 13:34:40 CEST 2008
593 - new function Coro::AIO::aio_wait.
594 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
595 - greatly speed up and reduce memory usage of Coro::AIO requests.
596 - implement some other µ-optimisations.
597
5984.749 Mon Sep 29 14:40:12 CEST 2008
599 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
600
6014.748 Sat Sep 27 14:03:19 CEST 2008
602 - implement, but do not document, PerlIO::cede(granularity).
603 - Coro::Storable forgot to wrap Storable::pstore.
604 - work around the multitude of leaks and memory corruption
605 bugs in PerlIO::via by using our own C-level perliol. As a side
606 effect, Coro::Storable is now much, much, much faster.
607
6084.747 Tue Sep 23 01:59:41 CEST 2008
609 - fix a per-cv memleak (one empty array was leaked per
610 code reference).
611 - avoid a crash in coro->call|rss when the coroutine was already
612 destroyed (most noticably when using Coro::Debug::ps :)
613 - also protect *Storable::FILE.
614 - push up default storable granularity to 20ms.
615
6164.746 Sun Sep 21 03:22:20 CEST 2008
617 - be more insistent on locking Storable against reentrancy
618 in Coro::Storable.
619 - move swap_def?v and throw to Coro::State, as documented.
620
6214.745 Thu Jul 24 00:14:38 CEST 2008
622 - remove debugging code related to MgPV_nolen_const, also try to
623 make it compile with perl 5.8.6 (yes, apple apparently loves
624 outdated software). Reported by John S.
625
6264.744 Tue Jul 8 22:06:35 CEST 2008
627 - correctly provide default DIE/WARN handlers as documented.
628 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
629 implementations inside perl work fine for us.
630
6314.743 Mon Jun 16 00:21:57 CEST 2008
632 - when using Coro::EV without running EV::loop it could
633 result in busy-waiting for events, this has been fixed.
634 - reduce codesize and improve performance by using EV_DEFAULT_UC.
635
6364.742 Sat May 31 14:10:21 CEST 2008
637 - implement a workaround for (some) perl <5.8.8 versions.
638 - require EV 3.3+.
639
6404.741 Fri May 30 23:33:09 CEST 2008
641 - tell netbsd how utterly broken their imitation of an OS is
642 and refuse to build by default if pthreads are in use.
643 - switch to "s" method on all bsd's by default, as their ucontext
644 stuff seems just too broken.
645 - fix a bug in Coro::Select.
7 646
84.74 Thu May 29 20:05:31 CEST 2008 6474.74 Thu May 29 20:05:31 CEST 2008
9 - do not test Coro::LWP for lack of dependencies. 648 - do not test Coro::LWP for lack of dependencies.
10 - do not test Coro::Select for lack of working perls. 649 - do not test Coro::Select for lack of working perls.
11 650
… …
15 - sprinkle "no warnings" freely over everything, also suppress 654 - sprinkle "no warnings" freely over everything, also suppress
16 warnings for some other modules. 655 warnings for some other modules.
17 - fix typo in WSAEWOULDBLOCK. 656 - fix typo in WSAEWOULDBLOCK.
18 657
194.72 Sun May 25 05:14:36 CEST 2008 6584.72 Sun May 25 05:14:36 CEST 2008
20 - tweak META.yaml a bit, unfortunately, there is no documented way 659 - tweak META.yaml a bit, unfortunately, there is no documented way
21 to have optional dependencies with CPAN. doh :( 660 to have optional dependencies with CPAN. doh :(
22 - avoid running some tests on windows because they would fail due to 661 - avoid running some tests on windows because they would fail due to
23 perl bug (broken fork, broken pipes...). 662 perl bug (broken fork, broken pipes...).
24 - work around perl on windows bugs where perl returns undocumented 663 - work around perl on windows bugs where perl returns undocumented
25 error codes for sysread, syswrite etc. by taking advantage 664 error codes for sysread, syswrite etc. by taking advantage
… …
29 - use unix domain sockets in testsuite to work around 668 - use unix domain sockets in testsuite to work around
30 common perl implementation bugs on widows (they are emulated by 669 common perl implementation bugs on widows (they are emulated by
31 tcp sockets on windows. ugh.) 670 tcp sockets on windows. ugh.)
32 671
334.71 Sat May 24 20:01:27 CEST 2008 6724.71 Sat May 24 20:01:27 CEST 2008
34 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 673 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
35 - take advantage of async name resolution of AnyEvent::Util. 674 - take advantage of async name resolution of AnyEvent::Util.
36 - work around brutal inet_aton override in Coro::LWP. 675 - work around brutal inet_aton override in Coro::LWP.
37 - take advantage of the readyhook in Coro::EV, for smoother 676 - take advantage of the readyhook in Coro::EV, for smoother
38 scheduling. 677 scheduling.
39 678
404.7 Sun May 11 00:32:19 CEST 2008 6794.7 Sun May 11 00:32:19 CEST 2008
41 - completely reworked the Coro manpage. 680 - completely reworked the Coro manpage.
42 - added Coro::AnyEvent, generic event loop integration. 681 - added Coro::AnyEvent, generic event loop integration.
43 - implement cancel, ready and kill commands in Coro::Debug. 682 - implement cancel, ready and kill commands in Coro::Debug.
44 - document find_coro in Coro::Debug. 683 - document find_coro in Coro::Debug.
45 - incompatible change: rename has_stack to has_cctx. 684 - incompatible change: rename has_stack to has_cctx.
46 - Coro::AIO and Coro::BDB no longer force event model detection, 685 - Coro::AIO and Coro::BDB no longer force event model detection,
47 use AnyEvent::AIO and AnyEvent::BDB. 686 use AnyEvent::AIO and AnyEvent::BDB.
… …
171 810
1724.11 Thu Oct 11 02:40:24 CEST 2007 8114.11 Thu Oct 11 02:40:24 CEST 2007
173 - port to threaded perls. 812 - port to threaded perls.
174 813
1754.1 Thu Oct 11 02:38:16 CEST 2007 8144.1 Thu Oct 11 02:38:16 CEST 2007
176 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 815 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
177 be local to each coro (see Coro::State). 816 be local to each coro (see Coro::State).
178 - incompatible change: for very deep reasons, cede and cede_notself 817 - incompatible change: for very deep reasons, cede and cede_notself
179 cannot return anything, so nothing will be returned. 818 cannot return anything, so nothing will be returned.
180 - possibly bring back 5.10 compatibility (untested). 819 - possibly bring back 5.10 compatibility (untested).
181 - work around stupid (and wrong) warning on 5.10 :(. 820 - work around stupid (and wrong) warning on 5.10 :(.
… …
253 but stateful semantics. 892 but stateful semantics.
254 - fixed a lot of typos in Coro.pm (patch submitted by David 893 - fixed a lot of typos in Coro.pm (patch submitted by David
255 Steinbrunner, which applied flawlessly). 894 Steinbrunner, which applied flawlessly).
256 895
2573.6 Sat Apr 14 17:13:31 CEST 2007 8963.6 Sat Apr 14 17:13:31 CEST 2007
258 - added some bugfixes to get eg/myhttpd working again. 897 - added some bugfixes to get eg/myhttpd working again.
259 - added Coro::Storable for often-cede'ing freeze/thaw. 898 - added Coro::Storable for often-cede'ing freeze/thaw.
260 - try to do a clean exit when a coroutine calls exit 899 - try to do a clean exit when a coroutine calls exit
261 (EXPERIMENTAL). 900 (EXPERIMENTAL).
262 - got rid of indirect call through _coro_init. 901 - got rid of indirect call through _coro_init.
263 - updated the partly antique examples in eg/ to 902 - updated the partly antique examples in eg/ to

Diff Legend

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