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 1.651 by root, Fri Oct 16 08:11:18 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
10TODO: exceptions do not reset EV readable watchers
11TODO: http://pastebin.com/UauGDx7D
12 - bump minimum perl version to 5.10.
13
146.48 Sun Oct 4 19:03:51 CEST 2015
15 - fix memory leak when Coro::Handle uses Coro::EV internally
16 (testcase by Sten Sten).
17 - update code to libev 4 API (internally, Coro still used the
18 version 3 API calls).
19 - remove unused hv_sig.
20
216.47 Sat Jul 11 03:58:20 CEST 2015
22 - swap_svs - NVs can be in the SV head beginning with 5.20.
23
246.46 Tue Jun 30 14:36:30 CEST 2015
25 - restore portability to perls with windows fork emulation
26 (patch by Petr Písař).
27
286.45 Tue Jun 30 01:40:08 CEST 2015
29 - restore portability to perl 5.16 and below.
30
316.44 Tue Jun 30 00:41:54 CEST 2015
32 - avoid segfaulting (or worse) when tracing a canceled thread.
33 - the Event module silently broke it's public hook API, causing
34 failures on perls compiled with -Duselongdouble. Requiring
35 the latest version of Event because I don't know which release
36 changed this (it's not mentioned in the ChangeLog).
37 - allow xs level enterleave hooks via CoroAPI.
38
396.43 Thu Jun 4 15:38:14 CEST 2015
40 - use stability canary.
41 - port to stableperl-5.22.0-1.001.
42 - update libecb.
43
446.42 Wed Feb 11 20:29:52 CET 2015
45 - Coro::SemaphoreSet->try did not actually work (analyzed by
46 SATO Kentaro).
47 - upgrade libecb to be C11 compliant.
48
496.41 Sat Sep 6 22:08:46 CEST 2014
50 - restore portability to perl 5.8.x.
51 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
52 up some memory, but fixes "use feature" and similar modules
53 when used inside a Coro.
54 - allow zero as argument to Coro::Channel to mean the same thing
55 as no argument. this works with older versions as well,
56 but wasn't legal until now.
57 - slightly better c header file detection.
58
596.39 Mon Jun 2 00:00:08 CEST 2014
60 - work around more incompatible changes in 5.20.
61
626.38 Sun Jun 1 21:54:23 CEST 2014
63 - check that perl slots actually have enough space to hold
64 interpreter variables.
65 - untested port to perl 5.19 (64 bit tmps indices) (reported
66 by Andreas König).
67 - croak when cancel is called without a thread context.
68
696.37 Tue Mar 4 13:27:33 CET 2014
70 - *sigh*, removed leftover debugging code from debugging a
71 perl bug, of all things.
72
736.36 Tue Mar 4 07:11:59 CET 2014
74 - semaphores would not clear the destroy hook when interrupted
75 by ->throw, causing segfaults or worse.
76 - ->throw on a thread waiting for a semaphore did not acquire
77 the semaphore, but also didn't wake up other waiters,
78 possibly causing a deadlock.
79 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
80 to report a full stacktrace, which should help find out
81 where the actual call is.
82 - "atomically" destroy data in slf_destroy, because it is
83 the right thing to do, just in case.
84 - disable ecb.h libm dependency, to compile on ancient systems
85 or under adverse conditions.
86
876.33 Mon Nov 18 11:26:27 CET 2013
88 - do not crash when freeing padlists with holes (in 5.18).
89 - tentative SVt_BIND 5.19 port/fix.
90
916.32 Tue Nov 5 15:35:35 CET 2013
92 - use a new algorithm to derive padlists for perl 5.18. The old
93 one could lead to 0-pointer accesses inside perl (reported
94 by Darin McBride).
95
966.31 Thu May 9 07:39:48 CEST 2013
97 - Coro::AIO requests would crash if the thread was ready'd
98 while the request was ongoing.
99
1006.29 Wed May 8 02:55:18 CEST 2013
101 - when an on_destroy handler destructs the coro currently being
102 destructed a perl scalar could be accessed after being freed,
103 likely causing a crash.
104
1056.28 Wed Mar 6 06:58:02 CET 2013
106 - clean remnants of existing __DIE__ and __WARN__ handlers so
107 they lose their magic and will not cause segfaults later
108 (testcase by Andrey Sagulin).
109 - improved Coro::State documentation a bit.
110 - Coro::Debug::command now flushes the output.
111 - add hack detection code for x32 abi, because the braindead slugs
112 who designed that made it look exactly like x86_64 without
113 providing proper compile time symbols to test for it. as a result,
114 this detection cannot work reliably.
115 - valgrind stack registering was broken.
116 - do not rely on Time::HiRes anymore in Coro::Debug.
117
1186.23 Fri Dec 7 23:36:37 CET 2012
119 - use experimental fiber implementation on native windows
120 perls.
121 - use sizeof (void *) as multiplication factor for stack sizes,
122 to accomodate the totally braindamaged microsoft 64 bit "os".
123 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
124 activeperl 5.16 crashes when PerlIO_define_layer is called due
125 to some bug in the perl dll, strawberry perl at least passes
126 the testsuite.
127 - implement Coro::Handle->peeraddr/host/port, for slightly
128 improved compatibility with LWP.
129 - implement 5.17 compatibility by almost blindly applying a
130 good-looking patch by Father Chrysostomos.
131 - move stack management functions into libcoro 3.
132 - libcoro version 3 "released".
133 - support magic values as timed_io_once args.
134 - recommend AnyEvent 7+ or EV 4+, also require EV
135 version 4 or newer for Coro::EV.
136
1376.10 Tue Oct 9 01:14:27 CEST 2012
138 - updated ecb.h, it had a typo that caused it to not compile on many
139 big endian systems (reported by many people).
140 - disable memory fences in ecb.h to improve portability.
141
1426.09 Sat Oct 6 23:25:02 CEST 2012
143 - Coro::EV I/O watchers were not interruptible by exceptions
144 (Coro::State::throw) (testcase by sten).
145 - ->throw now puts threads into the ready queue, as this seems to
146 be expected by existing code, and code that doesn't cope with spurious
147 wakeups needs fixing anyway.
148 - use fd -1 in mmap.
149 - cast I32 to int in error message printf.
150 - warn about broken so-called "hardened" kernels.
151
1526.08 Fri Apr 13 12:05:47 CEST 2012
153 - be more aggressive about exiting like perl does - formerly,
154 exiting from the non-main thread would not execute END blocks.
155
1566.07 Fri Nov 11 21:21:48 CET 2011
157 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
158 - update ecb.h.
159
1606.06 Mon Aug 8 23:59:48 CEST 2011
161 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
162 backend, which is an order of magnitude slower. unfortunately, cygwins
163 pthread implementation isn't very complete either, so allocate the stack
164 twice just to be sure.
165 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
166 in its ithreaded perl - best recompile cygwin and use the 'w'indows
167 backend for much better performance. also disable ithreads for
168 even better performance...).
169
1706.05 Thu Aug 4 21:36:36 CEST 2011
171 - blush, condvar values would not be propagated from send to recv anymore
172 (reported by Chip Salzenberg).
173 - use exponential increase for the readline buffer length in
174 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
175
1766.04 Wed Aug 3 17:15:45 CEST 2011
177 - use even more efficient and more compatible condvars for
178 compatibility to AnyEvent 6.x :)
179 - more inconsequential ecb.h updates.
180
1816.03 Wed Aug 3 11:41:30 CEST 2011
182 - change how Coro patches AnyEvent condvars for compatibility to
183 AnyEvent 6.x.
184 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
185 hasn't been included.
186
1876.02 Wed Jul 13 04:35:19 CEST 2011
188 - "improve portability to Gentoo" - gentoo manages to put perl variables
189 in memory areas that are farther than 2gb apart, which the jit couldn't
190 handle and barfed. now it's just a bit slower on gentoo and similar
191 systems.
192
1936.01 Sun Jul 3 12:31:14 CEST 2011
194 - workarounds are good, but the test for whether pthreads are used
195 was not good. this one should be better.
196 - check differently whether gcc generates cfi instructions itself.
197
1986.0 Wed Jun 29 19:43:35 CEST 2011
199 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
200 destroyed and cleaned up automatically (e.g. async { schedule }).
201 - implement a JIT compiler for part of the thread switch code,
202 which gives a 50% speed improvement on threaded perls, and
203 about 4% on non-threaded perls (so threaded perls now finally
204 reach about half the speed of non-threaded perls).
205 - slightly modernise Coro::Intro, add section about rouse functions.
206 - avoid DEFSV and ERRSV, giving another 10% improvement
207 in thread switching.
208 - Coro::State->is_destroyed is now called is_zombie.
209 - implement a Coro->safe_cancel method that might fail, but
210 cancels in a "safer" way if it succeeds.
211 - add preliminary support for DEBUGGING perls.
212 - get rid of two hash-accesses when initialising a new Coro - this
213 speeds up coro creation by almost a factor of two.
214 - croak when a coro that is being cancelled tries to block
215 (e.g. while executing a guard block), instead of crashing or
216 deadlocking.
217 - use a more robust and also faster method to identify Coro::State
218 objects - speeds up everything a bit.
219 - implement Coro->cancel in XS for a 20% speed improvement, and to
220 be able to implement mutual cancellation.
221 - speed up context switches by a percent or two by more efficiently
222 allocating context stack entries.
223 - implement Coro->join and Coro->on_destroy in XS for a speedup and
224 a reduction in memory use.
225 - cancelling a coro while it itself is cancelling another coro is
226 now supported and working, instead of triggering an assertion.
227 - be a bit more crash-resistant when calling (buggy) on_destroy
228 callbacks (best effort).
229 - move on_destroy into the slf_frame, to allow extension slf
230 functions to have destructors.
231 - get rid if coro refcounting - simply crash in other interpreter
232 threads by nulling the pointers on clone.
233 - simplify warn/die hook handling when loading Coro - the convoluted
234 logic seems to be no longer neccessary.
235 - use libecb instead of our own home-grown gcc hacks.
236 - document alternatives to Coro::LWP. Please use them :)
237 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
238 sigaltstack. Really. wine suffers from it, erlang suffers from it,
239 and it's known since at least 2006.
240
2415.372 Wed Feb 23 06:14:30 CET 2011
242 - apparently mingw doesn't provide a working gettimeofday, try to
243 work around that by relying on Time::HiRes (indirectly brought to
244 my attention by Max Maischein).
245 - fix some portability issues when Time::HiRes was used.
246
2475.371 Mon Feb 21 14:36:08 CET 2011
248 - backport to windows process emulation code again.
249
2505.37 Sat Feb 19 07:49:44 CET 2011
251 - add a big "Coro thread life cycle" section to "man Coro".
252 - try a tentative workaround against the breakage that 5.13 has
253 introduced without depreciation period. sigh.
254 - no longer use Time::HiRes if gettimeofday is available, which
255 saves quite a lot of memory.
256
2575.36 Sun Feb 13 05:33:41 CET 2011
258 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
259 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
260 method call.
261 - undocument Coro::Timer::sleep and obsolete whole module.
262 - optimise Coro::Timer::timeout memory and cpu usage.
263 - slightly updated Coro::Intro for recent changes.
264 - do not initialise PL_dirty anymore.
265
2665.25 Thu Nov 11 01:08:39 CET 2010
267 - try a different approach on netbsd - netbsd 5 finally has marginally
268 working pthreads, but still broken ucontext/sigaltstack.
269 - openbsd 4.8 finally got their act together, Coro works out of the box
270 with asm, setjmp and pthreads (no change, just informational).
271
2725.24 Sat Oct 23 11:27:12 CEST 2010
273 - port to the EV 4.0 API.
274 - work around bugs in mingw32, making strawberry perl work
275 out of the box.
276 - correctly modify Coro::AIO function prototypes
277 so that they reflect the "no optional parameters" rule.
278 - "ported" libcoro to C++.
279
2805.23 Mon May 17 18:50:42 CEST 2010
281 - be more resistant to ordering changes when initialising
282 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
283 Waldorf).
284 - document that perl 5.12 deliberately removed support for cloning.
285
2865.22 Wed Apr 14 03:55:35 CEST 2010
287 - correctly return udnef on errors in Coro::Handle::read/write
288 (testcase by Marc Mims).
289 - convert Coro::Util into a "perl compatibility wrapper" - the functions
290 are less useful now, but are drop-in replacements for existing
291 functions, listing better alternatives in the documentation. This also
292 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
293 with Coro::Util::inet_aton.
294 - do not override $Coro::idle unconditionally in Coro.pm, as other
295 modules could have provided their own idle coro already
296 (for exmaple, Coro::AnyEvent).
297 - fix Coro::Util::gethost* functions.
298 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
299
3005.21 Wed Dec 16 07:19:51 CET 2009
301 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
302 together.
303 - add some examples on how to combine other event loops with Coro in
304 Coro::AnyEvent, and how to run it (and not to block). Seems to be
305 the most common source of confusion.
306 - try to catch people naively blocking in an event callback.
307 - work around the perl filehandle bug issue in conjunction with
308 older common::sense (as indirectly pointed out by ZSystem).
309 - clarify the "not from signal handlers" section.
310
3115.2 Sun Oct 4 14:54:24 CEST 2009
312 - Coro::Storable destroyed the prototypes of the functions it wrapped.
313 - export rouse_cb and rouse_wait by default now.
314 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
315 - new method $state->swap_sv.
316 - added section on "windows process emulation" to the manpage,
317 after a not-so-fruitful (nor-friendly) "discussion" with chip
318 salzenberg (discussion implies arguments, but his only arguments
319 were ad-hominems, one wonders why he started it in the first
320 place). I hope this explains it well enough for him to understand,
321 and maybe well enough for others to understand.
322 - use common::sense everywhere now.
323 - idle callbacks are no longer supported, use idle coros instead.
324 - print a thread listing when a deadlock is detected.
325
3265.17 Sat Aug 22 23:09:31 CEST 2009
327 - work around a bug in the perl debugger causing crashes
328 when running under the debugger by marking _pool_handler as nodebug.
329 - speed up Coro::async considerably.
330 - try some hacks to get netbsd to work "more often" - their broken
331 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
332 - convert Coro to AE 5.0 API.
333
3345.162 Tue Jul 28 04:04:03 CEST 2009
335 - perl 5.8.2 is now minimum requirement.
336 - skip t/19_handle.t on broken windows perls.
337
3385.161 Wed Jul 22 04:47:38 CEST 2009
339 - Coro::AnyEvent::poll could have a different prototype when EV was
340 used as backend (analyzed by Tatsuhiko Miyagawa).
341 - Coro::AnyEvent errornously initialised the event loop when loaded,
342 not on demand.
343 - try to workaround rare */t/01_unblock.t failures.
344
3455.16 Tue Jul 21 01:44:37 CEST 2009
346 - Coro::AnyEvent failed to hook into the event loop
347 when no threads had been readied between detecting
348 the event loop and actually running it.
349 - considerably speed up Coro::Select by taking avdantage
350 of AnyEvent > 4.8 and some other optimisations.
351 - implement paragraph readline mode in Coro::Handle
352 (based on patches by Zsbán Ambrus).
353 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
354 and analyzed by Yasuhiro MATSUMOTO).
355 - clarified libcoro license and copyright.
356 - someone stole my EXTRA_META!!!
357 - implement Coro::Select::patch_pp_sselect and it's brother,
358 for hardcode select overriding.
359
3605.151 Mon Jul 6 05:41:57 CEST 2009
361 - backport to windows process emulation code again (patch by
362 Yasuhiro MATSUMOTO).
363 - slightly update Coro::MakeMaker.
364
3655.15 Tue Jun 30 10:28:06 CEST 2009
366 - deprecate Coro::Socket, document how to get ipv6 support via
367 AnyEvent::Socket instead.
368 - implement signal->wait ($cb) interface, similar to semaphores.
369 - work around SvOK not supporting getmagic, so we have to getmagic
370 to test for undef :/ (reported by Matthias Waldorf).
371 - load Coro::AnyEvent in all modules using AnyEvent.
372 - work around perl corrupting our internal data structures,
373 reported by Tokuhiro Matsuno.
374 - enable per-coroutine real and cpu time gathering
375 (Coro::State::enable_times).
376
3775.14 Wed Jun 24 01:37:48 CEST 2009
378 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
379 as this seems to have been a difficult concept (poll, sleep, idle,
380 idle_upto).
381 - add Coro::AnyEvent::readable/writable functions.
382 - clarify Coro::EV/Event/AnyEvent manpages.
383 - free per-thread global scalars in the thread calling ->cancel, to
384 avoid crashes when $_, $@ etc., are magical but some of those
385 had already been freed.
386 - "unexperimentalise" the callback interface for Coro::Semaphore.
387 - speed up ready queue management/context switching by using a linked
388 list instead of an array (~5-10%).
389 - implement "watch" command in Coro::Debug shells.
390 - for fun, implement time-slicing as an example in the manpage.
391 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
392 Coro::Event, use the normal AnyEvent handling instead of dieing
393 (the same is true for Coro::Handle).
394 - properly document Coro::EV::timed_io_once.
395 - avoid unneccessary ->cancel calls in Coro::Handle.
396 - maybe make it work on mingw32 with win32 backend
397 (based on patch by Yasuhiro Matsumoto).
398
3995.132 Fri May 29 09:00:39 CEST 2009
400 - do not keep a reference to the argument itself in
401 Coro::Semaphore::guard, as it could change later.
402 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
403 should support a prepare callback.
404
4055.131 Mon Mar 16 23:20:37 CET 2009
406 - implement and document Coro->suspend, Coro->resume.
407 - fix Coro::Select implementation to not (often) close
408 the passed file descriptors (testcase provided by pippijn).
409
4105.13 Mon Dec 15 21:51:42 CET 2008
411 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
412 - don't set diehook to C<undef> but instead to NULL, to avoid
413 spurious warnings.
414 - fix a lot of bugs in Coro::SemaphoreSet.
415 - Coro::SemaphoreSet will less often create a semaphore needlessly.
416 - add Coro::SemaphoreSet::count and wait methods.
417 - take advantage of the new Guard module.
418 - deprecate Coro::guard.
419 - try to fix the dreaded 01_unblock tests once more. I hate it when
420 testsuites need more fixing than the code they are supposed to test.
421 - croak in more cases when a required callback isn't resolvable.
422 - fix some minor issues in Coro::State->call/eval.
423 - use current coroutine context instead of a temporary one
424 when temporarily switching to another coroutine.
425 - do not call C-level on_destroy handlers during global destruction,
426 to avoid needless segfaults.
427
4285.12 Sun Dec 7 13:30:38 CET 2008
429 - add default config for MirOS, which seems to be a bug-to-bug
430 compatible fork of openbsd ("world domination by releasing
431 openbsd cvs before the openbsd folks do it" or so :).
432 - free_padlist did destroy the names pad, not good, but didn't
433 seem to bother perl - this could fix issues such as eval ""
434 inside a function called from multiple coroutines.
435 - use a different method to detect destruction time.
436 - be more careful when freeing padlists just before global
437 destruction.
438 - fixed and expanded the call/cc example.
439 - renamed _terminate to _coro_run.
440 - new method Coro::Channel->shutdown.
441 - try pthreads on openbsd <4.4 (broken sigaltstack, will
442 pthreads fare better?).
443 - be less picky about destroying "a" running coroutine.
444
4455.11 Tue Nov 25 21:49:05 CET 2008
446 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
447 working code with failed assertions, introducing more bugs than
448 it fixes, requiring elaborate workarounds :(
449
4505.1 Mon Nov 24 08:54:59 CET 2008
451 - wrote a small introductory tutorial - Coro::Intro.
452 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
453 - Coro::Select did errornously dup the file descriptors
454 and didn't work with all AnyEvent backends.
455 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
456 LWP data transfers.
457 - disassociate c contexts from coro objects - this is agruably more
458 correct, but mostly allows sharing of cctxs between coro and state
459 objects, for added memory savings and speed increases.
460 - bumped $Coro::POOL_RSS up to 32kb by default.
461 - no longer set the optype to OP_CUSTOM, as B::* understandably
462 doesn't like this very much (and we *are* a type of entersub).
463 - implement state cloning, just to prove that call/cc can be done.
464 - automatically load Coro::AnyEvent in Coro::Handle.
465 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
466 do not have this method (and don't need it either).
467 - speed up generic anyevent methods in Coro::Handle by using rouse
468 callbacks.
469 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
470 others. It also makes the debugger happier, as you can trace
471 through the idle threads now.
472 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
473 - eg/event was pretty broken.
474 - better 5.8.6 compatibility.
475
4765.0 Thu Nov 20 10:35:05 CET 2008
477 - NEW ARCHITECTURE: use the latest 4.x version if you experience
478 stability issues.
479 - bump API version to 7 - all dependents must be recompiled.
480 - removed timed_* functions - they were not being used anyways
481 and should be replaced by a more generic mechanism -
482 and were annoying to support anyways :)
483 - removed SemaphoreSet's waiter method - use sem method instead.
484 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
485 - async_pool did free a scalar value twice
486 ("Attempt to unreference...").
487 - fix a longstanding bug where calling terminate on a coro that
488 was waiting for a semaphore that was just becoming available
489 would cause a deadlock (semaphore would get into a state where
490 it was available but waiters were still blocked).
491 - calling throw on a coroutine that is waiting for a semaphore will
492 no longer make it acquire the semaphore (and thus leak a count).
493 - perl's process emulation is now not even theoretically supported
494 anymore.
495 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
496 conversion of callback-style to blocking-style.
497 - new methods $coro->schedule_to and ->cede_to, to specifically
498 schedule or cede to a specific coroutine.
499 - new function Coro::Semaphore::wait.
500 - use named constants in Coro::Channel (Richard Hundt).
501 - directly patch the entersub opcode calling SLF functions (cede,
502 transfer and so on). this does speed up context switching, but
503 more importanly, it frees us from the hardcoded behaviour of
504 entersub, so we might actually be able to return something from
505 those functions and atcually create new ones.
506 - take advantage of __builtin_frame_address on gcc.
507 - expose THX in coroapi (not sure whether this was a wise decision,
508 as "threaded" perls are running at half speed anyways).
509 - implement execute_slf (schedule-like-function) interface that makes
510 it possible to implement schedule-like-functions in XS.
511 - use new SLF interface to massively speed up Coro::EV by roughly a
512 factor of two.
513 - used new SLF interface to massively speed up Coro::Semaphore by a
514 factor of three.
515 - used new SLF interface to speed up Coro::AIO by roughly a factor of
516 four and reduce its memory usage considerably.
517 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
518 for a nice speedup and vastly more correct behaviour. Also implement
519 a new method "sem" to get at the underlying semaphore object.
520 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
521 (in comparison) 20-40% speedup.
522 - used new SLF interface to reimplement Coro::Signal gaining
523 some unknown (because I was too lazy), but certain, speedup, and also
524 making signals reliable for the first time.
525 - used new SLF interface and other optimisations to speed up async_pool
526 by a factor of two. It also doesn't rely on perl's exception mechanism
527 to exit anymore. The overhead for terminating an async_pool, coro over
528 a normal async is now very small.
529 - sped up coroutine creation/destruction by 40%.
530 - forgot to include Coro/libcoro/README in the dist for all these years.
531 - work around a freebsd pthreads bug (manual testcancel is required as
532 pthread_cond_wait isn't a cancellation point on freebsd).
533 - use new rouse functions to speed up and simplify Coro::BDB.
534 - make "prefer perl native functions" work with threaded perls.
535 - condense Coro::Debug ps output, hint at v and w flags.
536 - (libcoro) lots of minor cleanups and portability improvements.
537
5384.914 Wed Nov 19 12:54:18 CET 2008
539 - fix a disastrous bug in the readline optimisation
540 introduced in 4.801.
541
5424.913 Sat Nov 15 07:58:28 CET 2008
543 - async_pool did free a scalar value twice
544 ("Attempt to unreference...").
545
5464.912 Thu Nov 13 18:31:23 CET 2008
547 - minor cleanups.
548 - use much larger stacks on linux and perl < 5.8.8.
549 - Coro::Debug::new_unix_server did not unlink the socket
550 when destroyed.
551
5524.911 Tue Nov 11 04:26:17 CET 2008
553 - "port" to threaded perls.
554
5554.91 Mon Nov 10 05:36:38 CET 2008
556 - the ->throw exception object no longer leaks.
557 - creating a new cctx leaked a scopestack entry (memleak).
558 - new coroutines didn't get created with a zero flags field
559 (unknown impact).
560 - calling ->throw on a not-yet-started coroutine should now work
561 instead of being ignored.
562 - ->throw is now supported on Coro::State objects.
563 - clean up cctx creation code a bit.
564 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
565
5664.9 Sat Nov 8 17:45:27 CET 2008
567 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
568 - (libcoro) no longer leak threads in the experimental pthread backend,
569 also speed it up considerably.
570 - (libcoro) do not rely on makecontext passing void *'s unscathed.
571 - fix compiletime dependencies on libcoro in the Makefile.
572 - cctx_count wasn't always updated properly.
573 - Coro::State::cctx_stacksize wasn't applied correctly.
574 - new function Coro::State::cctx_max_idle.
575 - the default max number of idle C contexts is now 4.
576 - (libcoro) try harder to get _setjmp/_longjmp.
577 - (libcoro) cleanup and extend the libcoro API to officially
578 allow the creation of empty source contexts.
579 - very experimental workaround for the totally broken netbsd platform.
580 - tried to hack around openbsd bugs.
581
5824.804 Wed Nov 5 16:36:00 CET 2008
583 - Coro::Debug::new_unix_server would not create a non-blocking listening
584 socket, sometimes causing freezes.
585 - (libcoro) fix misaligned stack points for setjmp and assembly
586 methods, which can cause crashes on x86/x86_64 with a sufficiently
587 aggressive compiler.
588 - new function: Coro::Debug::new_tcp_server.
589 - move ->throw into the Coro class because it only works on coro objects.
590
5914.803 Mon Nov 3 17:16:12 CET 2008
592 - (libcoro) use a global asm statement to become independent of gcc
593 otpimisations for CORO_ASM (thanks to pippijn for the idea).
594 - try to workaround yet another broken bsd, this time dragonfly.
595
5964.802 Thu Oct 30 10:56:12 CET 2008
597 - support -fno-omit-frame-pointer on x86 with the assembly method.
598 - tune 01_unblock.t tests a bit.
599
6004.801 Wed Oct 22 18:33:37 CEST 2008
601 - improve readline speed for very long "lines".
602 - backport to 5.8.8.
603
6044.8 Thu Oct 2 13:34:40 CEST 2008
605 - new function Coro::AIO::aio_wait.
606 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
607 - greatly speed up and reduce memory usage of Coro::AIO requests.
608 - implement some other µ-optimisations.
609
6104.749 Mon Sep 29 14:40:12 CEST 2008
611 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
612
6134.748 Sat Sep 27 14:03:19 CEST 2008
614 - implement, but do not document, PerlIO::cede(granularity).
615 - Coro::Storable forgot to wrap Storable::pstore.
616 - work around the multitude of leaks and memory corruption
617 bugs in PerlIO::via by using our own C-level perliol. As a side
618 effect, Coro::Storable is now much, much, much faster.
619
6204.747 Tue Sep 23 01:59:41 CEST 2008
621 - fix a per-cv memleak (one empty array was leaked per
622 code reference).
623 - avoid a crash in coro->call|rss when the coroutine was already
624 destroyed (most noticably when using Coro::Debug::ps :)
625 - also protect *Storable::FILE.
626 - push up default storable granularity to 20ms.
627
6284.746 Sun Sep 21 03:22:20 CEST 2008
629 - be more insistent on locking Storable against reentrancy
630 in Coro::Storable.
631 - move swap_def?v and throw to Coro::State, as documented.
632
6334.745 Thu Jul 24 00:14:38 CEST 2008
634 - remove debugging code related to MgPV_nolen_const, also try to
635 make it compile with perl 5.8.6 (yes, apple apparently loves
636 outdated software). Reported by John S.
637
6384.744 Tue Jul 8 22:06:35 CEST 2008
639 - correctly provide default DIE/WARN handlers as documented.
640 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
641 implementations inside perl work fine for us.
642
6434.743 Mon Jun 16 00:21:57 CEST 2008
644 - when using Coro::EV without running EV::loop it could
645 result in busy-waiting for events, this has been fixed.
646 - reduce codesize and improve performance by using EV_DEFAULT_UC.
647
6484.742 Sat May 31 14:10:21 CEST 2008
649 - implement a workaround for (some) perl <5.8.8 versions.
650 - require EV 3.3+.
651
6524.741 Fri May 30 23:33:09 CEST 2008
653 - tell netbsd how utterly broken their imitation of an OS is
654 and refuse to build by default if pthreads are in use.
655 - switch to "s" method on all bsd's by default, as their ucontext
656 stuff seems just too broken.
657 - fix a bug in Coro::Select.
7 658
84.74 Thu May 29 20:05:31 CEST 2008 6594.74 Thu May 29 20:05:31 CEST 2008
9 - do not test Coro::LWP for lack of dependencies. 660 - do not test Coro::LWP for lack of dependencies.
10 - do not test Coro::Select for lack of working perls. 661 - do not test Coro::Select for lack of working perls.
11 662
… …
15 - sprinkle "no warnings" freely over everything, also suppress 666 - sprinkle "no warnings" freely over everything, also suppress
16 warnings for some other modules. 667 warnings for some other modules.
17 - fix typo in WSAEWOULDBLOCK. 668 - fix typo in WSAEWOULDBLOCK.
18 669
194.72 Sun May 25 05:14:36 CEST 2008 6704.72 Sun May 25 05:14:36 CEST 2008
20 - tweak META.yaml a bit, unfortunately, there is no documented way 671 - tweak META.yaml a bit, unfortunately, there is no documented way
21 to have optional dependencies with CPAN. doh :( 672 to have optional dependencies with CPAN. doh :(
22 - avoid running some tests on windows because they would fail due to 673 - avoid running some tests on windows because they would fail due to
23 perl bug (broken fork, broken pipes...). 674 perl bug (broken fork, broken pipes...).
24 - work around perl on windows bugs where perl returns undocumented 675 - work around perl on windows bugs where perl returns undocumented
25 error codes for sysread, syswrite etc. by taking advantage 676 error codes for sysread, syswrite etc. by taking advantage
… …
29 - use unix domain sockets in testsuite to work around 680 - use unix domain sockets in testsuite to work around
30 common perl implementation bugs on widows (they are emulated by 681 common perl implementation bugs on widows (they are emulated by
31 tcp sockets on windows. ugh.) 682 tcp sockets on windows. ugh.)
32 683
334.71 Sat May 24 20:01:27 CEST 2008 6844.71 Sat May 24 20:01:27 CEST 2008
34 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 685 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
35 - take advantage of async name resolution of AnyEvent::Util. 686 - take advantage of async name resolution of AnyEvent::Util.
36 - work around brutal inet_aton override in Coro::LWP. 687 - work around brutal inet_aton override in Coro::LWP.
37 - take advantage of the readyhook in Coro::EV, for smoother 688 - take advantage of the readyhook in Coro::EV, for smoother
38 scheduling. 689 scheduling.
39 690
404.7 Sun May 11 00:32:19 CEST 2008 6914.7 Sun May 11 00:32:19 CEST 2008
41 - completely reworked the Coro manpage. 692 - completely reworked the Coro manpage.
42 - added Coro::AnyEvent, generic event loop integration. 693 - added Coro::AnyEvent, generic event loop integration.
43 - implement cancel, ready and kill commands in Coro::Debug. 694 - implement cancel, ready and kill commands in Coro::Debug.
44 - document find_coro in Coro::Debug. 695 - document find_coro in Coro::Debug.
45 - incompatible change: rename has_stack to has_cctx. 696 - incompatible change: rename has_stack to has_cctx.
46 - Coro::AIO and Coro::BDB no longer force event model detection, 697 - Coro::AIO and Coro::BDB no longer force event model detection,
47 use AnyEvent::AIO and AnyEvent::BDB. 698 use AnyEvent::AIO and AnyEvent::BDB.
… …
171 822
1724.11 Thu Oct 11 02:40:24 CEST 2007 8234.11 Thu Oct 11 02:40:24 CEST 2007
173 - port to threaded perls. 824 - port to threaded perls.
174 825
1754.1 Thu Oct 11 02:38:16 CEST 2007 8264.1 Thu Oct 11 02:38:16 CEST 2007
176 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 827 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
177 be local to each coro (see Coro::State). 828 be local to each coro (see Coro::State).
178 - incompatible change: for very deep reasons, cede and cede_notself 829 - incompatible change: for very deep reasons, cede and cede_notself
179 cannot return anything, so nothing will be returned. 830 cannot return anything, so nothing will be returned.
180 - possibly bring back 5.10 compatibility (untested). 831 - possibly bring back 5.10 compatibility (untested).
181 - work around stupid (and wrong) warning on 5.10 :(. 832 - work around stupid (and wrong) warning on 5.10 :(.
… …
253 but stateful semantics. 904 but stateful semantics.
254 - fixed a lot of typos in Coro.pm (patch submitted by David 905 - fixed a lot of typos in Coro.pm (patch submitted by David
255 Steinbrunner, which applied flawlessly). 906 Steinbrunner, which applied flawlessly).
256 907
2573.6 Sat Apr 14 17:13:31 CEST 2007 9083.6 Sat Apr 14 17:13:31 CEST 2007
258 - added some bugfixes to get eg/myhttpd working again. 909 - added some bugfixes to get eg/myhttpd working again.
259 - added Coro::Storable for often-cede'ing freeze/thaw. 910 - added Coro::Storable for often-cede'ing freeze/thaw.
260 - try to do a clean exit when a coroutine calls exit 911 - try to do a clean exit when a coroutine calls exit
261 (EXPERIMENTAL). 912 (EXPERIMENTAL).
262 - got rid of indirect call through _coro_init. 913 - got rid of indirect call through _coro_init.
263 - updated the partly antique examples in eg/ to 914 - updated the partly antique examples in eg/ to

Diff Legend

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