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

Comparing Coro/Changes (file contents):
Revision 1.351 by root, Sun May 25 01:32:15 2008 UTC vs.
Revision 1.666 by root, Fri Jul 14 02:55:09 2017 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO. 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: maybe implement a default message channel, very much like Erlang's 4TODO: channel->maxsize(newsize)?
5 actor model (which is cool in a lot of important aspects (failures!), 5TODO: __GCC_HAVE_DWARF2_CFI_ASM
6 but very lacking in others (higher level ipc)). 6TODO: swap_sv, maybe add scope_swap_sv?
7TODO: croak when async_pool tries to run canceled thread?
7 8
9 - seems to pass test suite down till 5.8.9, so make this the
10 minimum requirement in Makefile.PL.
11 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
12 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
13 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
14 fortified longjmp variants in some libcs.
15 - re-try JIT allocation without PROT_EXEC, to hopefully improve
16 portability (still doesn't seem to work around broken selinux
17 mmap everywhere).
18
196.511 Sun Jun 26 23:44:50 CEST 2016
20 - make it compile with threaded perls again.
21 - simplify and speed up __DIE__ and __WARN__ handling.
22
236.51 Sat Jun 25 20:57:20 CEST 2016
24 - 6.5 release didn't compile on most older perls due to
25 a logic error enabling experimental 5.24 code for all versions.
26 - since various broken or deficient patches float around
27 to work around the vtbl API breakage in perl, let's provide
28 our own workaround, which seems to be much less invasive
29 and more compatible than the approaches seen so far. YMMV.
30 - work around assertion failure in perl_destruct on 5.24
31 (analyzed by Dave Mitchell).
32
336.5 Wed Jun 22 22:23:50 CEST 2016
34 - swap_sv swapping order was not symmetrical, causing
35 wrong swaps when swap_sv was used multiple times on the
36 same sv in the same thread.
37 - swap_sv calls can now be undone by calling it again
38 with the same variables.
39 - swap_sv calls will now be undone in async_pool threads.
40 - split Coro::Semaphore::up/adjust into separate xs functions
41 for better error reporting, at slight codesize increase.
42 - (libcoro) arm assembly support, please test and report.
43 - adjust to PL_savestack changes in perl 5.24 (adapted from
44 the debian patch, which unfortunately gets it wrong).
45
466.49 Sat Oct 17 01:40:12 CEST 2015
47 - throwing an exception to a thread waiting in
48 Coro::Handle using Coro::EV did not stop the watchers,
49 causing the next call to fail (testcase by Martin Pritchard).
50 - bump minimum perl version to 5.10.
51
526.48 Sun Oct 4 19:03:51 CEST 2015
53 - fix memory leak when Coro::Handle uses Coro::EV internally
54 (testcase by Sten Sten).
55 - update code to libev 4 API (internally, Coro still used the
56 version 3 API calls).
57 - remove unused hv_sig.
58
596.47 Sat Jul 11 03:58:20 CEST 2015
60 - swap_svs - NVs can be in the SV head beginning with 5.20.
61
626.46 Tue Jun 30 14:36:30 CEST 2015
63 - restore portability to perls with windows fork emulation
64 (patch by Petr Písař).
65
666.45 Tue Jun 30 01:40:08 CEST 2015
67 - restore portability to perl 5.16 and below.
68
696.44 Tue Jun 30 00:41:54 CEST 2015
70 - avoid segfaulting (or worse) when tracing a canceled thread.
71 - the Event module silently broke it's public hook API, causing
72 failures on perls compiled with -Duselongdouble. Requiring
73 the latest version of Event because I don't know which release
74 changed this (it's not mentioned in the ChangeLog).
75 - allow xs level enterleave hooks via CoroAPI.
76
776.43 Thu Jun 4 15:38:14 CEST 2015
78 - use stability canary.
79 - port to stableperl-5.22.0-1.001.
80 - update libecb.
81
826.42 Wed Feb 11 20:29:52 CET 2015
83 - Coro::SemaphoreSet->try did not actually work (analyzed by
84 SATO Kentaro).
85 - upgrade libecb to be C11 compliant.
86
876.41 Sat Sep 6 22:08:46 CEST 2014
88 - restore portability to perl 5.8.x.
89 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
90 up some memory, but fixes "use feature" and similar modules
91 when used inside a Coro.
92 - allow zero as argument to Coro::Channel to mean the same thing
93 as no argument. this works with older versions as well,
94 but wasn't legal until now.
95 - slightly better c header file detection.
96
976.39 Mon Jun 2 00:00:08 CEST 2014
98 - work around more incompatible changes in 5.20.
99
1006.38 Sun Jun 1 21:54:23 CEST 2014
101 - check that perl slots actually have enough space to hold
102 interpreter variables.
103 - untested port to perl 5.19 (64 bit tmps indices) (reported
104 by Andreas König).
105 - croak when cancel is called without a thread context.
106
1076.37 Tue Mar 4 13:27:33 CET 2014
108 - *sigh*, removed leftover debugging code from debugging a
109 perl bug, of all things.
110
1116.36 Tue Mar 4 07:11:59 CET 2014
112 - semaphores would not clear the destroy hook when interrupted
113 by ->throw, causing segfaults or worse.
114 - ->throw on a thread waiting for a semaphore did not acquire
115 the semaphore, but also didn't wake up other waiters,
116 possibly causing a deadlock.
117 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
118 to report a full stacktrace, which should help find out
119 where the actual call is.
120 - "atomically" destroy data in slf_destroy, because it is
121 the right thing to do, just in case.
122 - disable ecb.h libm dependency, to compile on ancient systems
123 or under adverse conditions.
124
1256.33 Mon Nov 18 11:26:27 CET 2013
126 - do not crash when freeing padlists with holes (in 5.18).
127 - tentative SVt_BIND 5.19 port/fix.
128
1296.32 Tue Nov 5 15:35:35 CET 2013
130 - use a new algorithm to derive padlists for perl 5.18. The old
131 one could lead to 0-pointer accesses inside perl (reported
132 by Darin McBride).
133
1346.31 Thu May 9 07:39:48 CEST 2013
135 - Coro::AIO requests would crash if the thread was ready'd
136 while the request was ongoing.
137
1386.29 Wed May 8 02:55:18 CEST 2013
139 - when an on_destroy handler destructs the coro currently being
140 destructed a perl scalar could be accessed after being freed,
141 likely causing a crash.
142
1436.28 Wed Mar 6 06:58:02 CET 2013
144 - clean remnants of existing __DIE__ and __WARN__ handlers so
145 they lose their magic and will not cause segfaults later
146 (testcase by Andrey Sagulin).
147 - improved Coro::State documentation a bit.
148 - Coro::Debug::command now flushes the output.
149 - add hack detection code for x32 abi, because the braindead slugs
150 who designed that made it look exactly like x86_64 without
151 providing proper compile time symbols to test for it. as a result,
152 this detection cannot work reliably.
153 - valgrind stack registering was broken.
154 - do not rely on Time::HiRes anymore in Coro::Debug.
155
1566.23 Fri Dec 7 23:36:37 CET 2012
157 - use experimental fiber implementation on native windows
158 perls.
159 - use sizeof (void *) as multiplication factor for stack sizes,
160 to accomodate the totally braindamaged microsoft 64 bit "os".
161 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
162 activeperl 5.16 crashes when PerlIO_define_layer is called due
163 to some bug in the perl dll, strawberry perl at least passes
164 the testsuite.
165 - implement Coro::Handle->peeraddr/host/port, for slightly
166 improved compatibility with LWP.
167 - implement 5.17 compatibility by almost blindly applying a
168 good-looking patch by Father Chrysostomos.
169 - move stack management functions into libcoro 3.
170 - libcoro version 3 "released".
171 - support magic values as timed_io_once args.
172 - recommend AnyEvent 7+ or EV 4+, also require EV
173 version 4 or newer for Coro::EV.
174
1756.10 Tue Oct 9 01:14:27 CEST 2012
176 - updated ecb.h, it had a typo that caused it to not compile on many
177 big endian systems (reported by many people).
178 - disable memory fences in ecb.h to improve portability.
179
1806.09 Sat Oct 6 23:25:02 CEST 2012
181 - Coro::EV I/O watchers were not interruptible by exceptions
182 (Coro::State::throw) (testcase by sten).
183 - ->throw now puts threads into the ready queue, as this seems to
184 be expected by existing code, and code that doesn't cope with spurious
185 wakeups needs fixing anyway.
186 - use fd -1 in mmap.
187 - cast I32 to int in error message printf.
188 - warn about broken so-called "hardened" kernels.
189
1906.08 Fri Apr 13 12:05:47 CEST 2012
191 - be more aggressive about exiting like perl does - formerly,
192 exiting from the non-main thread would not execute END blocks.
193
1946.07 Fri Nov 11 21:21:48 CET 2011
195 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
196 - update ecb.h.
197
1986.06 Mon Aug 8 23:59:48 CEST 2011
199 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
200 backend, which is an order of magnitude slower. unfortunately, cygwins
201 pthread implementation isn't very complete either, so allocate the stack
202 twice just to be sure.
203 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
204 in its ithreaded perl - best recompile cygwin and use the 'w'indows
205 backend for much better performance. also disable ithreads for
206 even better performance...).
207
2086.05 Thu Aug 4 21:36:36 CEST 2011
209 - blush, condvar values would not be propagated from send to recv anymore
210 (reported by Chip Salzenberg).
211 - use exponential increase for the readline buffer length in
212 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
213
2146.04 Wed Aug 3 17:15:45 CEST 2011
215 - use even more efficient and more compatible condvars for
216 compatibility to AnyEvent 6.x :)
217 - more inconsequential ecb.h updates.
218
2196.03 Wed Aug 3 11:41:30 CEST 2011
220 - change how Coro patches AnyEvent condvars for compatibility to
221 AnyEvent 6.x.
222 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
223 hasn't been included.
224
2256.02 Wed Jul 13 04:35:19 CEST 2011
226 - "improve portability to Gentoo" - gentoo manages to put perl variables
227 in memory areas that are farther than 2gb apart, which the jit couldn't
228 handle and barfed. now it's just a bit slower on gentoo and similar
229 systems.
230
2316.01 Sun Jul 3 12:31:14 CEST 2011
232 - workarounds are good, but the test for whether pthreads are used
233 was not good. this one should be better.
234 - check differently whether gcc generates cfi instructions itself.
235
2366.0 Wed Jun 29 19:43:35 CEST 2011
237 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
238 destroyed and cleaned up automatically (e.g. async { schedule }).
239 - implement a JIT compiler for part of the thread switch code,
240 which gives a 50% speed improvement on threaded perls, and
241 about 4% on non-threaded perls (so threaded perls now finally
242 reach about half the speed of non-threaded perls).
243 - slightly modernise Coro::Intro, add section about rouse functions.
244 - avoid DEFSV and ERRSV, giving another 10% improvement
245 in thread switching.
246 - Coro::State->is_destroyed is now called is_zombie.
247 - implement a Coro->safe_cancel method that might fail, but
248 cancels in a "safer" way if it succeeds.
249 - add preliminary support for DEBUGGING perls.
250 - get rid of two hash-accesses when initialising a new Coro - this
251 speeds up coro creation by almost a factor of two.
252 - croak when a coro that is being cancelled tries to block
253 (e.g. while executing a guard block), instead of crashing or
254 deadlocking.
255 - use a more robust and also faster method to identify Coro::State
256 objects - speeds up everything a bit.
257 - implement Coro->cancel in XS for a 20% speed improvement, and to
258 be able to implement mutual cancellation.
259 - speed up context switches by a percent or two by more efficiently
260 allocating context stack entries.
261 - implement Coro->join and Coro->on_destroy in XS for a speedup and
262 a reduction in memory use.
263 - cancelling a coro while it itself is cancelling another coro is
264 now supported and working, instead of triggering an assertion.
265 - be a bit more crash-resistant when calling (buggy) on_destroy
266 callbacks (best effort).
267 - move on_destroy into the slf_frame, to allow extension slf
268 functions to have destructors.
269 - get rid if coro refcounting - simply crash in other interpreter
270 threads by nulling the pointers on clone.
271 - simplify warn/die hook handling when loading Coro - the convoluted
272 logic seems to be no longer neccessary.
273 - use libecb instead of our own home-grown gcc hacks.
274 - document alternatives to Coro::LWP. Please use them :)
275 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
276 sigaltstack. Really. wine suffers from it, erlang suffers from it,
277 and it's known since at least 2006.
278
2795.372 Wed Feb 23 06:14:30 CET 2011
280 - apparently mingw doesn't provide a working gettimeofday, try to
281 work around that by relying on Time::HiRes (indirectly brought to
282 my attention by Max Maischein).
283 - fix some portability issues when Time::HiRes was used.
284
2855.371 Mon Feb 21 14:36:08 CET 2011
286 - backport to windows process emulation code again.
287
2885.37 Sat Feb 19 07:49:44 CET 2011
289 - add a big "Coro thread life cycle" section to "man Coro".
290 - try a tentative workaround against the breakage that 5.13 has
291 introduced without depreciation period. sigh.
292 - no longer use Time::HiRes if gettimeofday is available, which
293 saves quite a lot of memory.
294
2955.36 Sun Feb 13 05:33:41 CET 2011
296 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
297 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
298 method call.
299 - undocument Coro::Timer::sleep and obsolete whole module.
300 - optimise Coro::Timer::timeout memory and cpu usage.
301 - slightly updated Coro::Intro for recent changes.
302 - do not initialise PL_dirty anymore.
303
3045.25 Thu Nov 11 01:08:39 CET 2010
305 - try a different approach on netbsd - netbsd 5 finally has marginally
306 working pthreads, but still broken ucontext/sigaltstack.
307 - openbsd 4.8 finally got their act together, Coro works out of the box
308 with asm, setjmp and pthreads (no change, just informational).
309
3105.24 Sat Oct 23 11:27:12 CEST 2010
311 - port to the EV 4.0 API.
312 - work around bugs in mingw32, making strawberry perl work
313 out of the box.
314 - correctly modify Coro::AIO function prototypes
315 so that they reflect the "no optional parameters" rule.
316 - "ported" libcoro to C++.
317
3185.23 Mon May 17 18:50:42 CEST 2010
319 - be more resistant to ordering changes when initialising
320 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
321 Waldorf).
322 - document that perl 5.12 deliberately removed support for cloning.
323
3245.22 Wed Apr 14 03:55:35 CEST 2010
325 - correctly return udnef on errors in Coro::Handle::read/write
326 (testcase by Marc Mims).
327 - convert Coro::Util into a "perl compatibility wrapper" - the functions
328 are less useful now, but are drop-in replacements for existing
329 functions, listing better alternatives in the documentation. This also
330 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
331 with Coro::Util::inet_aton.
332 - do not override $Coro::idle unconditionally in Coro.pm, as other
333 modules could have provided their own idle coro already
334 (for exmaple, Coro::AnyEvent).
335 - fix Coro::Util::gethost* functions.
336 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
337
3385.21 Wed Dec 16 07:19:51 CET 2009
339 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
340 together.
341 - add some examples on how to combine other event loops with Coro in
342 Coro::AnyEvent, and how to run it (and not to block). Seems to be
343 the most common source of confusion.
344 - try to catch people naively blocking in an event callback.
345 - work around the perl filehandle bug issue in conjunction with
346 older common::sense (as indirectly pointed out by ZSystem).
347 - clarify the "not from signal handlers" section.
348
3495.2 Sun Oct 4 14:54:24 CEST 2009
350 - Coro::Storable destroyed the prototypes of the functions it wrapped.
351 - export rouse_cb and rouse_wait by default now.
352 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
353 - new method $state->swap_sv.
354 - added section on "windows process emulation" to the manpage,
355 after a not-so-fruitful (nor-friendly) "discussion" with chip
356 salzenberg (discussion implies arguments, but his only arguments
357 were ad-hominems, one wonders why he started it in the first
358 place). I hope this explains it well enough for him to understand,
359 and maybe well enough for others to understand.
360 - use common::sense everywhere now.
361 - idle callbacks are no longer supported, use idle coros instead.
362 - print a thread listing when a deadlock is detected.
363
3645.17 Sat Aug 22 23:09:31 CEST 2009
365 - work around a bug in the perl debugger causing crashes
366 when running under the debugger by marking _pool_handler as nodebug.
367 - speed up Coro::async considerably.
368 - try some hacks to get netbsd to work "more often" - their broken
369 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
370 - convert Coro to AE 5.0 API.
371
3725.162 Tue Jul 28 04:04:03 CEST 2009
373 - perl 5.8.2 is now minimum requirement.
374 - skip t/19_handle.t on broken windows perls.
375
3765.161 Wed Jul 22 04:47:38 CEST 2009
377 - Coro::AnyEvent::poll could have a different prototype when EV was
378 used as backend (analyzed by Tatsuhiko Miyagawa).
379 - Coro::AnyEvent errornously initialised the event loop when loaded,
380 not on demand.
381 - try to workaround rare */t/01_unblock.t failures.
382
3835.16 Tue Jul 21 01:44:37 CEST 2009
384 - Coro::AnyEvent failed to hook into the event loop
385 when no threads had been readied between detecting
386 the event loop and actually running it.
387 - considerably speed up Coro::Select by taking avdantage
388 of AnyEvent > 4.8 and some other optimisations.
389 - implement paragraph readline mode in Coro::Handle
390 (based on patches by Zsbán Ambrus).
391 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
392 and analyzed by Yasuhiro MATSUMOTO).
393 - clarified libcoro license and copyright.
394 - someone stole my EXTRA_META!!!
395 - implement Coro::Select::patch_pp_sselect and it's brother,
396 for hardcode select overriding.
397
3985.151 Mon Jul 6 05:41:57 CEST 2009
399 - backport to windows process emulation code again (patch by
400 Yasuhiro MATSUMOTO).
401 - slightly update Coro::MakeMaker.
402
4035.15 Tue Jun 30 10:28:06 CEST 2009
404 - deprecate Coro::Socket, document how to get ipv6 support via
405 AnyEvent::Socket instead.
406 - implement signal->wait ($cb) interface, similar to semaphores.
407 - work around SvOK not supporting getmagic, so we have to getmagic
408 to test for undef :/ (reported by Matthias Waldorf).
409 - load Coro::AnyEvent in all modules using AnyEvent.
410 - work around perl corrupting our internal data structures,
411 reported by Tokuhiro Matsuno.
412 - enable per-coroutine real and cpu time gathering
413 (Coro::State::enable_times).
414
4155.14 Wed Jun 24 01:37:48 CEST 2009
416 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
417 as this seems to have been a difficult concept (poll, sleep, idle,
418 idle_upto).
419 - add Coro::AnyEvent::readable/writable functions.
420 - clarify Coro::EV/Event/AnyEvent manpages.
421 - free per-thread global scalars in the thread calling ->cancel, to
422 avoid crashes when $_, $@ etc., are magical but some of those
423 had already been freed.
424 - "unexperimentalise" the callback interface for Coro::Semaphore.
425 - speed up ready queue management/context switching by using a linked
426 list instead of an array (~5-10%).
427 - implement "watch" command in Coro::Debug shells.
428 - for fun, implement time-slicing as an example in the manpage.
429 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
430 Coro::Event, use the normal AnyEvent handling instead of dieing
431 (the same is true for Coro::Handle).
432 - properly document Coro::EV::timed_io_once.
433 - avoid unneccessary ->cancel calls in Coro::Handle.
434 - maybe make it work on mingw32 with win32 backend
435 (based on patch by Yasuhiro Matsumoto).
436
4375.132 Fri May 29 09:00:39 CEST 2009
438 - do not keep a reference to the argument itself in
439 Coro::Semaphore::guard, as it could change later.
440 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
441 should support a prepare callback.
442
4435.131 Mon Mar 16 23:20:37 CET 2009
444 - implement and document Coro->suspend, Coro->resume.
445 - fix Coro::Select implementation to not (often) close
446 the passed file descriptors (testcase provided by pippijn).
447
4485.13 Mon Dec 15 21:51:42 CET 2008
449 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
450 - don't set diehook to C<undef> but instead to NULL, to avoid
451 spurious warnings.
452 - fix a lot of bugs in Coro::SemaphoreSet.
453 - Coro::SemaphoreSet will less often create a semaphore needlessly.
454 - add Coro::SemaphoreSet::count and wait methods.
455 - take advantage of the new Guard module.
456 - deprecate Coro::guard.
457 - try to fix the dreaded 01_unblock tests once more. I hate it when
458 testsuites need more fixing than the code they are supposed to test.
459 - croak in more cases when a required callback isn't resolvable.
460 - fix some minor issues in Coro::State->call/eval.
461 - use current coroutine context instead of a temporary one
462 when temporarily switching to another coroutine.
463 - do not call C-level on_destroy handlers during global destruction,
464 to avoid needless segfaults.
465
4665.12 Sun Dec 7 13:30:38 CET 2008
467 - add default config for MirOS, which seems to be a bug-to-bug
468 compatible fork of openbsd ("world domination by releasing
469 openbsd cvs before the openbsd folks do it" or so :).
470 - free_padlist did destroy the names pad, not good, but didn't
471 seem to bother perl - this could fix issues such as eval ""
472 inside a function called from multiple coroutines.
473 - use a different method to detect destruction time.
474 - be more careful when freeing padlists just before global
475 destruction.
476 - fixed and expanded the call/cc example.
477 - renamed _terminate to _coro_run.
478 - new method Coro::Channel->shutdown.
479 - try pthreads on openbsd <4.4 (broken sigaltstack, will
480 pthreads fare better?).
481 - be less picky about destroying "a" running coroutine.
482
4835.11 Tue Nov 25 21:49:05 CET 2008
484 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
485 working code with failed assertions, introducing more bugs than
486 it fixes, requiring elaborate workarounds :(
487
4885.1 Mon Nov 24 08:54:59 CET 2008
489 - wrote a small introductory tutorial - Coro::Intro.
490 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
491 - Coro::Select did errornously dup the file descriptors
492 and didn't work with all AnyEvent backends.
493 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
494 LWP data transfers.
495 - disassociate c contexts from coro objects - this is agruably more
496 correct, but mostly allows sharing of cctxs between coro and state
497 objects, for added memory savings and speed increases.
498 - bumped $Coro::POOL_RSS up to 32kb by default.
499 - no longer set the optype to OP_CUSTOM, as B::* understandably
500 doesn't like this very much (and we *are* a type of entersub).
501 - implement state cloning, just to prove that call/cc can be done.
502 - automatically load Coro::AnyEvent in Coro::Handle.
503 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
504 do not have this method (and don't need it either).
505 - speed up generic anyevent methods in Coro::Handle by using rouse
506 callbacks.
507 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
508 others. It also makes the debugger happier, as you can trace
509 through the idle threads now.
510 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
511 - eg/event was pretty broken.
512 - better 5.8.6 compatibility.
513
5145.0 Thu Nov 20 10:35:05 CET 2008
515 - NEW ARCHITECTURE: use the latest 4.x version if you experience
516 stability issues.
517 - bump API version to 7 - all dependents must be recompiled.
518 - removed timed_* functions - they were not being used anyways
519 and should be replaced by a more generic mechanism -
520 and were annoying to support anyways :)
521 - removed SemaphoreSet's waiter method - use sem method instead.
522 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
523 - async_pool did free a scalar value twice
524 ("Attempt to unreference...").
525 - fix a longstanding bug where calling terminate on a coro that
526 was waiting for a semaphore that was just becoming available
527 would cause a deadlock (semaphore would get into a state where
528 it was available but waiters were still blocked).
529 - calling throw on a coroutine that is waiting for a semaphore will
530 no longer make it acquire the semaphore (and thus leak a count).
531 - perl's process emulation is now not even theoretically supported
532 anymore.
533 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
534 conversion of callback-style to blocking-style.
535 - new methods $coro->schedule_to and ->cede_to, to specifically
536 schedule or cede to a specific coroutine.
537 - new function Coro::Semaphore::wait.
538 - use named constants in Coro::Channel (Richard Hundt).
539 - directly patch the entersub opcode calling SLF functions (cede,
540 transfer and so on). this does speed up context switching, but
541 more importanly, it frees us from the hardcoded behaviour of
542 entersub, so we might actually be able to return something from
543 those functions and atcually create new ones.
544 - take advantage of __builtin_frame_address on gcc.
545 - expose THX in coroapi (not sure whether this was a wise decision,
546 as "threaded" perls are running at half speed anyways).
547 - implement execute_slf (schedule-like-function) interface that makes
548 it possible to implement schedule-like-functions in XS.
549 - use new SLF interface to massively speed up Coro::EV by roughly a
550 factor of two.
551 - used new SLF interface to massively speed up Coro::Semaphore by a
552 factor of three.
553 - used new SLF interface to speed up Coro::AIO by roughly a factor of
554 four and reduce its memory usage considerably.
555 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
556 for a nice speedup and vastly more correct behaviour. Also implement
557 a new method "sem" to get at the underlying semaphore object.
558 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
559 (in comparison) 20-40% speedup.
560 - used new SLF interface to reimplement Coro::Signal gaining
561 some unknown (because I was too lazy), but certain, speedup, and also
562 making signals reliable for the first time.
563 - used new SLF interface and other optimisations to speed up async_pool
564 by a factor of two. It also doesn't rely on perl's exception mechanism
565 to exit anymore. The overhead for terminating an async_pool, coro over
566 a normal async is now very small.
567 - sped up coroutine creation/destruction by 40%.
568 - forgot to include Coro/libcoro/README in the dist for all these years.
569 - work around a freebsd pthreads bug (manual testcancel is required as
570 pthread_cond_wait isn't a cancellation point on freebsd).
571 - use new rouse functions to speed up and simplify Coro::BDB.
572 - make "prefer perl native functions" work with threaded perls.
573 - condense Coro::Debug ps output, hint at v and w flags.
574 - (libcoro) lots of minor cleanups and portability improvements.
575
5764.914 Wed Nov 19 12:54:18 CET 2008
577 - fix a disastrous bug in the readline optimisation
578 introduced in 4.801.
579
5804.913 Sat Nov 15 07:58:28 CET 2008
581 - async_pool did free a scalar value twice
582 ("Attempt to unreference...").
583
5844.912 Thu Nov 13 18:31:23 CET 2008
585 - minor cleanups.
586 - use much larger stacks on linux and perl < 5.8.8.
587 - Coro::Debug::new_unix_server did not unlink the socket
588 when destroyed.
589
5904.911 Tue Nov 11 04:26:17 CET 2008
591 - "port" to threaded perls.
592
5934.91 Mon Nov 10 05:36:38 CET 2008
594 - the ->throw exception object no longer leaks.
595 - creating a new cctx leaked a scopestack entry (memleak).
596 - new coroutines didn't get created with a zero flags field
597 (unknown impact).
598 - calling ->throw on a not-yet-started coroutine should now work
599 instead of being ignored.
600 - ->throw is now supported on Coro::State objects.
601 - clean up cctx creation code a bit.
602 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
603
6044.9 Sat Nov 8 17:45:27 CET 2008
605 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
606 - (libcoro) no longer leak threads in the experimental pthread backend,
607 also speed it up considerably.
608 - (libcoro) do not rely on makecontext passing void *'s unscathed.
609 - fix compiletime dependencies on libcoro in the Makefile.
610 - cctx_count wasn't always updated properly.
611 - Coro::State::cctx_stacksize wasn't applied correctly.
612 - new function Coro::State::cctx_max_idle.
613 - the default max number of idle C contexts is now 4.
614 - (libcoro) try harder to get _setjmp/_longjmp.
615 - (libcoro) cleanup and extend the libcoro API to officially
616 allow the creation of empty source contexts.
617 - very experimental workaround for the totally broken netbsd platform.
618 - tried to hack around openbsd bugs.
619
6204.804 Wed Nov 5 16:36:00 CET 2008
621 - Coro::Debug::new_unix_server would not create a non-blocking listening
622 socket, sometimes causing freezes.
623 - (libcoro) fix misaligned stack points for setjmp and assembly
624 methods, which can cause crashes on x86/x86_64 with a sufficiently
625 aggressive compiler.
626 - new function: Coro::Debug::new_tcp_server.
627 - move ->throw into the Coro class because it only works on coro objects.
628
6294.803 Mon Nov 3 17:16:12 CET 2008
630 - (libcoro) use a global asm statement to become independent of gcc
631 otpimisations for CORO_ASM (thanks to pippijn for the idea).
632 - try to workaround yet another broken bsd, this time dragonfly.
633
6344.802 Thu Oct 30 10:56:12 CET 2008
635 - support -fno-omit-frame-pointer on x86 with the assembly method.
636 - tune 01_unblock.t tests a bit.
637
6384.801 Wed Oct 22 18:33:37 CEST 2008
639 - improve readline speed for very long "lines".
640 - backport to 5.8.8.
641
6424.8 Thu Oct 2 13:34:40 CEST 2008
643 - new function Coro::AIO::aio_wait.
644 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
645 - greatly speed up and reduce memory usage of Coro::AIO requests.
646 - implement some other µ-optimisations.
647
6484.749 Mon Sep 29 14:40:12 CEST 2008
649 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
650
6514.748 Sat Sep 27 14:03:19 CEST 2008
652 - implement, but do not document, PerlIO::cede(granularity).
653 - Coro::Storable forgot to wrap Storable::pstore.
654 - work around the multitude of leaks and memory corruption
655 bugs in PerlIO::via by using our own C-level perliol. As a side
656 effect, Coro::Storable is now much, much, much faster.
657
6584.747 Tue Sep 23 01:59:41 CEST 2008
659 - fix a per-cv memleak (one empty array was leaked per
660 code reference).
661 - avoid a crash in coro->call|rss when the coroutine was already
662 destroyed (most noticably when using Coro::Debug::ps :)
663 - also protect *Storable::FILE.
664 - push up default storable granularity to 20ms.
665
6664.746 Sun Sep 21 03:22:20 CEST 2008
667 - be more insistent on locking Storable against reentrancy
668 in Coro::Storable.
669 - move swap_def?v and throw to Coro::State, as documented.
670
6714.745 Thu Jul 24 00:14:38 CEST 2008
672 - remove debugging code related to MgPV_nolen_const, also try to
673 make it compile with perl 5.8.6 (yes, apple apparently loves
674 outdated software). Reported by John S.
675
6764.744 Tue Jul 8 22:06:35 CEST 2008
677 - correctly provide default DIE/WARN handlers as documented.
678 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
679 implementations inside perl work fine for us.
680
6814.743 Mon Jun 16 00:21:57 CEST 2008
682 - when using Coro::EV without running EV::loop it could
683 result in busy-waiting for events, this has been fixed.
684 - reduce codesize and improve performance by using EV_DEFAULT_UC.
685
6864.742 Sat May 31 14:10:21 CEST 2008
687 - implement a workaround for (some) perl <5.8.8 versions.
688 - require EV 3.3+.
689
6904.741 Fri May 30 23:33:09 CEST 2008
691 - tell netbsd how utterly broken their imitation of an OS is
692 and refuse to build by default if pthreads are in use.
693 - switch to "s" method on all bsd's by default, as their ucontext
694 stuff seems just too broken.
695 - fix a bug in Coro::Select.
696
6974.74 Thu May 29 20:05:31 CEST 2008
698 - do not test Coro::LWP for lack of dependencies.
699 - do not test Coro::Select for lack of working perls.
700
7014.73 Thu May 29 2008
702 - fix a bug in Coro::EV which would cause it to block despite
703 there being runnable coroutines.
704 - sprinkle "no warnings" freely over everything, also suppress
705 warnings for some other modules.
706 - fix typo in WSAEWOULDBLOCK.
707
7084.72 Sun May 25 05:14:36 CEST 2008
8 - tweak META.yaml a bit, unfortunately, there is no documented way 709 - tweak META.yaml a bit, unfortunately, there is no documented way
9 to have optional dependencies with CPAN. doh :( 710 to have optional dependencies with CPAN. doh :(
10 - avoid running some tests on windows because they would fail due to 711 - avoid running some tests on windows because they would fail due to
11 perl bug (broken fork, broken pipes...). 712 perl bug (broken fork, broken pipes...).
12 - work around perl on windows bugs where perl returns undocumented 713 - work around perl on windows bugs where perl returns undocumented
13 error codes for sysread, syswrite etc. by taking advantage 714 error codes for sysread, syswrite etc. by taking advantage
… …
17 - use unix domain sockets in testsuite to work around 718 - use unix domain sockets in testsuite to work around
18 common perl implementation bugs on widows (they are emulated by 719 common perl implementation bugs on widows (they are emulated by
19 tcp sockets on windows. ugh.) 720 tcp sockets on windows. ugh.)
20 721
214.71 Sat May 24 20:01:27 CEST 2008 7224.71 Sat May 24 20:01:27 CEST 2008
22 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 723 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
23 - take advantage of async name resolution of AnyEvent::Util. 724 - take advantage of async name resolution of AnyEvent::Util.
24 - work around brutal inet_aton override in Coro::LWP. 725 - work around brutal inet_aton override in Coro::LWP.
25 - take advantage of the readyhook in Coro::EV, for smoother 726 - take advantage of the readyhook in Coro::EV, for smoother
26 scheduling. 727 scheduling.
27 728
284.7 Sun May 11 00:32:19 CEST 2008 7294.7 Sun May 11 00:32:19 CEST 2008
29 - completely reworked the Coro manpage. 730 - completely reworked the Coro manpage.
30 - added Coro::AnyEvent, generic event loop integration. 731 - added Coro::AnyEvent, generic event loop integration.
31 - implement cancel, ready and kill commands in Coro::Debug. 732 - implement cancel, ready and kill commands in Coro::Debug.
32 - document find_coro in Coro::Debug. 733 - document find_coro in Coro::Debug.
33 - incompatible change: rename has_stack to has_cctx. 734 - incompatible change: rename has_stack to has_cctx.
34 - Coro::AIO and Coro::BDB no longer force event model detection, 735 - Coro::AIO and Coro::BDB no longer force event model detection,
35 use AnyEvent::AIO and AnyEvent::BDB. 736 use AnyEvent::AIO and AnyEvent::BDB.
… …
159 860
1604.11 Thu Oct 11 02:40:24 CEST 2007 8614.11 Thu Oct 11 02:40:24 CEST 2007
161 - port to threaded perls. 862 - port to threaded perls.
162 863
1634.1 Thu Oct 11 02:38:16 CEST 2007 8644.1 Thu Oct 11 02:38:16 CEST 2007
164 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 865 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
165 be local to each coro (see Coro::State). 866 be local to each coro (see Coro::State).
166 - incompatible change: for very deep reasons, cede and cede_notself 867 - incompatible change: for very deep reasons, cede and cede_notself
167 cannot return anything, so nothing will be returned. 868 cannot return anything, so nothing will be returned.
168 - possibly bring back 5.10 compatibility (untested). 869 - possibly bring back 5.10 compatibility (untested).
169 - work around stupid (and wrong) warning on 5.10 :(. 870 - work around stupid (and wrong) warning on 5.10 :(.
… …
241 but stateful semantics. 942 but stateful semantics.
242 - fixed a lot of typos in Coro.pm (patch submitted by David 943 - fixed a lot of typos in Coro.pm (patch submitted by David
243 Steinbrunner, which applied flawlessly). 944 Steinbrunner, which applied flawlessly).
244 945
2453.6 Sat Apr 14 17:13:31 CEST 2007 9463.6 Sat Apr 14 17:13:31 CEST 2007
246 - added some bugfixes to get eg/myhttpd working again. 947 - added some bugfixes to get eg/myhttpd working again.
247 - added Coro::Storable for often-cede'ing freeze/thaw. 948 - added Coro::Storable for often-cede'ing freeze/thaw.
248 - try to do a clean exit when a coroutine calls exit 949 - try to do a clean exit when a coroutine calls exit
249 (EXPERIMENTAL). 950 (EXPERIMENTAL).
250 - got rid of indirect call through _coro_init. 951 - got rid of indirect call through _coro_init.
251 - updated the partly antique examples in eg/ to 952 - updated the partly antique examples in eg/ to

Diff Legend

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