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

Comparing Coro/Changes (file contents):
Revision 1.291 by root, Fri Nov 23 16:35:20 2007 UTC vs.
Revision 1.672 by root, Thu Aug 31 16:28:49 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 4TODO: channel->maxsize(newsize)?
5 (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
96.514 Thu Aug 31 18:25:31 CEST 2017
10 - safe_cancel no longer croaks when called on already-destroyed
11 thread, but instead "just works".
12
136.513 Sat Jul 15 01:18:51 CEST 2017
14 - experimental 5.22 comaptibility changes (PadlistNAMES).
15 - removed attributes example and made sure the other examples
16 still work.
17 - fix some statevar types for newer perls, should be of no consequence.
18
196.512 Fri Jul 14 05:03:17 CEST 2017
20 - seems to pass test suite down till 5.8.9, so make this the
21 minimum requirement in Makefile.PL.
22 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
23 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
24 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
25 fortified longjmp variants in some libcs.
26 - re-try JIT allocation without PROT_EXEC, to hopefully improve
27 portability (still doesn't seem to work around broken selinux
28 mmap everywhere).
29 - speling fixes (Xavier Guimard).
30
316.511 Sun Jun 26 23:44:50 CEST 2016
32 - make it compile with threaded perls again.
33 - simplify and speed up __DIE__ and __WARN__ handling.
34
356.51 Sat Jun 25 20:57:20 CEST 2016
36 - 6.5 release didn't compile on most older perls due to
37 a logic error enabling experimental 5.24 code for all versions.
38 - since various broken or deficient patches float around
39 to work around the vtbl API breakage in perl, let's provide
40 our own workaround, which seems to be much less invasive
41 and more compatible than the approaches seen so far. YMMV.
42 - work around assertion failure in perl_destruct on 5.24
43 (analyzed by Dave Mitchell).
44
456.5 Wed Jun 22 22:23:50 CEST 2016
46 - swap_sv swapping order was not symmetrical, causing
47 wrong swaps when swap_sv was used multiple times on the
48 same sv in the same thread.
49 - swap_sv calls can now be undone by calling it again
50 with the same variables.
51 - swap_sv calls will now be undone in async_pool threads.
52 - split Coro::Semaphore::up/adjust into separate xs functions
53 for better error reporting, at slight codesize increase.
54 - (libcoro) arm assembly support, please test and report.
55 - adjust to PL_savestack changes in perl 5.24 (adapted from
56 the debian patch, which unfortunately gets it wrong).
57
586.49 Sat Oct 17 01:40:12 CEST 2015
59 - throwing an exception to a thread waiting in
60 Coro::Handle using Coro::EV did not stop the watchers,
61 causing the next call to fail (testcase by Martin Pritchard).
62 - bump minimum perl version to 5.10.
63
646.48 Sun Oct 4 19:03:51 CEST 2015
65 - fix memory leak when Coro::Handle uses Coro::EV internally
66 (testcase by Sten Sten).
67 - update code to libev 4 API (internally, Coro still used the
68 version 3 API calls).
69 - remove unused hv_sig.
70
716.47 Sat Jul 11 03:58:20 CEST 2015
72 - swap_svs - NVs can be in the SV head beginning with 5.20.
73
746.46 Tue Jun 30 14:36:30 CEST 2015
75 - restore portability to perls with windows fork emulation
76 (patch by Petr Písař).
77
786.45 Tue Jun 30 01:40:08 CEST 2015
79 - restore portability to perl 5.16 and below.
80
816.44 Tue Jun 30 00:41:54 CEST 2015
82 - avoid segfaulting (or worse) when tracing a canceled thread.
83 - the Event module silently broke it's public hook API, causing
84 failures on perls compiled with -Duselongdouble. Requiring
85 the latest version of Event because I don't know which release
86 changed this (it's not mentioned in the ChangeLog).
87 - allow xs level enterleave hooks via CoroAPI.
88
896.43 Thu Jun 4 15:38:14 CEST 2015
90 - use stability canary.
91 - port to stableperl-5.22.0-1.001.
92 - update libecb.
93
946.42 Wed Feb 11 20:29:52 CET 2015
95 - Coro::SemaphoreSet->try did not actually work (analyzed by
96 SATO Kentaro).
97 - upgrade libecb to be C11 compliant.
98
996.41 Sat Sep 6 22:08:46 CEST 2014
100 - restore portability to perl 5.8.x.
101 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
102 up some memory, but fixes "use feature" and similar modules
103 when used inside a Coro.
104 - allow zero as argument to Coro::Channel to mean the same thing
105 as no argument. this works with older versions as well,
106 but wasn't legal until now.
107 - slightly better c header file detection.
108
1096.39 Mon Jun 2 00:00:08 CEST 2014
110 - work around more incompatible changes in 5.20.
111
1126.38 Sun Jun 1 21:54:23 CEST 2014
113 - check that perl slots actually have enough space to hold
114 interpreter variables.
115 - untested port to perl 5.19 (64 bit tmps indices) (reported
116 by Andreas König).
117 - croak when cancel is called without a thread context.
118
1196.37 Tue Mar 4 13:27:33 CET 2014
120 - *sigh*, removed leftover debugging code from debugging a
121 perl bug, of all things.
122
1236.36 Tue Mar 4 07:11:59 CET 2014
124 - semaphores would not clear the destroy hook when interrupted
125 by ->throw, causing segfaults or worse.
126 - ->throw on a thread waiting for a semaphore did not acquire
127 the semaphore, but also didn't wake up other waiters,
128 possibly causing a deadlock.
129 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
130 to report a full stacktrace, which should help find out
131 where the actual call is.
132 - "atomically" destroy data in slf_destroy, because it is
133 the right thing to do, just in case.
134 - disable ecb.h libm dependency, to compile on ancient systems
135 or under adverse conditions.
136
1376.33 Mon Nov 18 11:26:27 CET 2013
138 - do not crash when freeing padlists with holes (in 5.18).
139 - tentative SVt_BIND 5.19 port/fix.
140
1416.32 Tue Nov 5 15:35:35 CET 2013
142 - use a new algorithm to derive padlists for perl 5.18. The old
143 one could lead to 0-pointer accesses inside perl (reported
144 by Darin McBride).
145
1466.31 Thu May 9 07:39:48 CEST 2013
147 - Coro::AIO requests would crash if the thread was ready'd
148 while the request was ongoing.
149
1506.29 Wed May 8 02:55:18 CEST 2013
151 - when an on_destroy handler destructs the coro currently being
152 destructed a perl scalar could be accessed after being freed,
153 likely causing a crash.
154
1556.28 Wed Mar 6 06:58:02 CET 2013
156 - clean remnants of existing __DIE__ and __WARN__ handlers so
157 they lose their magic and will not cause segfaults later
158 (testcase by Andrey Sagulin).
159 - improved Coro::State documentation a bit.
160 - Coro::Debug::command now flushes the output.
161 - add hack detection code for x32 abi, because the braindead slugs
162 who designed that made it look exactly like x86_64 without
163 providing proper compile time symbols to test for it. as a result,
164 this detection cannot work reliably.
165 - valgrind stack registering was broken.
166 - do not rely on Time::HiRes anymore in Coro::Debug.
167
1686.23 Fri Dec 7 23:36:37 CET 2012
169 - use experimental fiber implementation on native windows
170 perls.
171 - use sizeof (void *) as multiplication factor for stack sizes,
172 to accomodate the totally braindamaged microsoft 64 bit "os".
173 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
174 activeperl 5.16 crashes when PerlIO_define_layer is called due
175 to some bug in the perl dll, strawberry perl at least passes
176 the testsuite.
177 - implement Coro::Handle->peeraddr/host/port, for slightly
178 improved compatibility with LWP.
179 - implement 5.17 compatibility by almost blindly applying a
180 good-looking patch by Father Chrysostomos.
181 - move stack management functions into libcoro 3.
182 - libcoro version 3 "released".
183 - support magic values as timed_io_once args.
184 - recommend AnyEvent 7+ or EV 4+, also require EV
185 version 4 or newer for Coro::EV.
186
1876.10 Tue Oct 9 01:14:27 CEST 2012
188 - updated ecb.h, it had a typo that caused it to not compile on many
189 big endian systems (reported by many people).
190 - disable memory fences in ecb.h to improve portability.
191
1926.09 Sat Oct 6 23:25:02 CEST 2012
193 - Coro::EV I/O watchers were not interruptible by exceptions
194 (Coro::State::throw) (testcase by sten).
195 - ->throw now puts threads into the ready queue, as this seems to
196 be expected by existing code, and code that doesn't cope with spurious
197 wakeups needs fixing anyway.
198 - use fd -1 in mmap.
199 - cast I32 to int in error message printf.
200 - warn about broken so-called "hardened" kernels.
201
2026.08 Fri Apr 13 12:05:47 CEST 2012
203 - be more aggressive about exiting like perl does - formerly,
204 exiting from the non-main thread would not execute END blocks.
205
2066.07 Fri Nov 11 21:21:48 CET 2011
207 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
208 - update ecb.h.
209
2106.06 Mon Aug 8 23:59:48 CEST 2011
211 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
212 backend, which is an order of magnitude slower. unfortunately, cygwins
213 pthread implementation isn't very complete either, so allocate the stack
214 twice just to be sure.
215 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
216 in its ithreaded perl - best recompile cygwin and use the 'w'indows
217 backend for much better performance. also disable ithreads for
218 even better performance...).
219
2206.05 Thu Aug 4 21:36:36 CEST 2011
221 - blush, condvar values would not be propagated from send to recv anymore
222 (reported by Chip Salzenberg).
223 - use exponential increase for the readline buffer length in
224 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
225
2266.04 Wed Aug 3 17:15:45 CEST 2011
227 - use even more efficient and more compatible condvars for
228 compatibility to AnyEvent 6.x :)
229 - more inconsequential ecb.h updates.
230
2316.03 Wed Aug 3 11:41:30 CEST 2011
232 - change how Coro patches AnyEvent condvars for compatibility to
233 AnyEvent 6.x.
234 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
235 hasn't been included.
236
2376.02 Wed Jul 13 04:35:19 CEST 2011
238 - "improve portability to Gentoo" - gentoo manages to put perl variables
239 in memory areas that are farther than 2gb apart, which the jit couldn't
240 handle and barfed. now it's just a bit slower on gentoo and similar
241 systems.
242
2436.01 Sun Jul 3 12:31:14 CEST 2011
244 - workarounds are good, but the test for whether pthreads are used
245 was not good. this one should be better.
246 - check differently whether gcc generates cfi instructions itself.
247
2486.0 Wed Jun 29 19:43:35 CEST 2011
249 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
250 destroyed and cleaned up automatically (e.g. async { schedule }).
251 - implement a JIT compiler for part of the thread switch code,
252 which gives a 50% speed improvement on threaded perls, and
253 about 4% on non-threaded perls (so threaded perls now finally
254 reach about half the speed of non-threaded perls).
255 - slightly modernise Coro::Intro, add section about rouse functions.
256 - avoid DEFSV and ERRSV, giving another 10% improvement
257 in thread switching.
258 - Coro::State->is_destroyed is now called is_zombie.
259 - implement a Coro->safe_cancel method that might fail, but
260 cancels in a "safer" way if it succeeds.
261 - add preliminary support for DEBUGGING perls.
262 - get rid of two hash-accesses when initialising a new Coro - this
263 speeds up coro creation by almost a factor of two.
264 - croak when a coro that is being cancelled tries to block
265 (e.g. while executing a guard block), instead of crashing or
266 deadlocking.
267 - use a more robust and also faster method to identify Coro::State
268 objects - speeds up everything a bit.
269 - implement Coro->cancel in XS for a 20% speed improvement, and to
270 be able to implement mutual cancellation.
271 - speed up context switches by a percent or two by more efficiently
272 allocating context stack entries.
273 - implement Coro->join and Coro->on_destroy in XS for a speedup and
274 a reduction in memory use.
275 - cancelling a coro while it itself is cancelling another coro is
276 now supported and working, instead of triggering an assertion.
277 - be a bit more crash-resistant when calling (buggy) on_destroy
278 callbacks (best effort).
279 - move on_destroy into the slf_frame, to allow extension slf
280 functions to have destructors.
281 - get rid if coro refcounting - simply crash in other interpreter
282 threads by nulling the pointers on clone.
283 - simplify warn/die hook handling when loading Coro - the convoluted
284 logic seems to be no longer neccessary.
285 - use libecb instead of our own home-grown gcc hacks.
286 - document alternatives to Coro::LWP. Please use them :)
287 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
288 sigaltstack. Really. wine suffers from it, erlang suffers from it,
289 and it's known since at least 2006.
290
2915.372 Wed Feb 23 06:14:30 CET 2011
292 - apparently mingw doesn't provide a working gettimeofday, try to
293 work around that by relying on Time::HiRes (indirectly brought to
294 my attention by Max Maischein).
295 - fix some portability issues when Time::HiRes was used.
296
2975.371 Mon Feb 21 14:36:08 CET 2011
298 - backport to windows process emulation code again.
299
3005.37 Sat Feb 19 07:49:44 CET 2011
301 - add a big "Coro thread life cycle" section to "man Coro".
302 - try a tentative workaround against the breakage that 5.13 has
303 introduced without depreciation period. sigh.
304 - no longer use Time::HiRes if gettimeofday is available, which
305 saves quite a lot of memory.
306
3075.36 Sun Feb 13 05:33:41 CET 2011
308 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
309 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
310 method call.
311 - undocument Coro::Timer::sleep and obsolete whole module.
312 - optimise Coro::Timer::timeout memory and cpu usage.
313 - slightly updated Coro::Intro for recent changes.
314 - do not initialise PL_dirty anymore.
315
3165.25 Thu Nov 11 01:08:39 CET 2010
317 - try a different approach on netbsd - netbsd 5 finally has marginally
318 working pthreads, but still broken ucontext/sigaltstack.
319 - openbsd 4.8 finally got their act together, Coro works out of the box
320 with asm, setjmp and pthreads (no change, just informational).
321
3225.24 Sat Oct 23 11:27:12 CEST 2010
323 - port to the EV 4.0 API.
324 - work around bugs in mingw32, making strawberry perl work
325 out of the box.
326 - correctly modify Coro::AIO function prototypes
327 so that they reflect the "no optional parameters" rule.
328 - "ported" libcoro to C++.
329
3305.23 Mon May 17 18:50:42 CEST 2010
331 - be more resistant to ordering changes when initialising
332 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
333 Waldorf).
334 - document that perl 5.12 deliberately removed support for cloning.
335
3365.22 Wed Apr 14 03:55:35 CEST 2010
337 - correctly return udnef on errors in Coro::Handle::read/write
338 (testcase by Marc Mims).
339 - convert Coro::Util into a "perl compatibility wrapper" - the functions
340 are less useful now, but are drop-in replacements for existing
341 functions, listing better alternatives in the documentation. This also
342 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
343 with Coro::Util::inet_aton.
344 - do not override $Coro::idle unconditionally in Coro.pm, as other
345 modules could have provided their own idle coro already
346 (for exmaple, Coro::AnyEvent).
347 - fix Coro::Util::gethost* functions.
348 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
349
3505.21 Wed Dec 16 07:19:51 CET 2009
351 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
352 together.
353 - add some examples on how to combine other event loops with Coro in
354 Coro::AnyEvent, and how to run it (and not to block). Seems to be
355 the most common source of confusion.
356 - try to catch people naively blocking in an event callback.
357 - work around the perl filehandle bug issue in conjunction with
358 older common::sense (as indirectly pointed out by ZSystem).
359 - clarify the "not from signal handlers" section.
360
3615.2 Sun Oct 4 14:54:24 CEST 2009
362 - Coro::Storable destroyed the prototypes of the functions it wrapped.
363 - export rouse_cb and rouse_wait by default now.
364 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
365 - new method $state->swap_sv.
366 - added section on "windows process emulation" to the manpage,
367 after a not-so-fruitful (nor-friendly) "discussion" with chip
368 salzenberg (discussion implies arguments, but his only arguments
369 were ad-hominems, one wonders why he started it in the first
370 place). I hope this explains it well enough for him to understand,
371 and maybe well enough for others to understand.
372 - use common::sense everywhere now.
373 - idle callbacks are no longer supported, use idle coros instead.
374 - print a thread listing when a deadlock is detected.
375
3765.17 Sat Aug 22 23:09:31 CEST 2009
377 - work around a bug in the perl debugger causing crashes
378 when running under the debugger by marking _pool_handler as nodebug.
379 - speed up Coro::async considerably.
380 - try some hacks to get netbsd to work "more often" - their broken
381 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
382 - convert Coro to AE 5.0 API.
383
3845.162 Tue Jul 28 04:04:03 CEST 2009
385 - perl 5.8.2 is now minimum requirement.
386 - skip t/19_handle.t on broken windows perls.
387
3885.161 Wed Jul 22 04:47:38 CEST 2009
389 - Coro::AnyEvent::poll could have a different prototype when EV was
390 used as backend (analyzed by Tatsuhiko Miyagawa).
391 - Coro::AnyEvent errornously initialised the event loop when loaded,
392 not on demand.
393 - try to workaround rare */t/01_unblock.t failures.
394
3955.16 Tue Jul 21 01:44:37 CEST 2009
396 - Coro::AnyEvent failed to hook into the event loop
397 when no threads had been readied between detecting
398 the event loop and actually running it.
399 - considerably speed up Coro::Select by taking avdantage
400 of AnyEvent > 4.8 and some other optimisations.
401 - implement paragraph readline mode in Coro::Handle
402 (based on patches by Zsbán Ambrus).
403 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
404 and analyzed by Yasuhiro MATSUMOTO).
405 - clarified libcoro license and copyright.
406 - someone stole my EXTRA_META!!!
407 - implement Coro::Select::patch_pp_sselect and it's brother,
408 for hardcode select overriding.
409
4105.151 Mon Jul 6 05:41:57 CEST 2009
411 - backport to windows process emulation code again (patch by
412 Yasuhiro MATSUMOTO).
413 - slightly update Coro::MakeMaker.
414
4155.15 Tue Jun 30 10:28:06 CEST 2009
416 - deprecate Coro::Socket, document how to get ipv6 support via
417 AnyEvent::Socket instead.
418 - implement signal->wait ($cb) interface, similar to semaphores.
419 - work around SvOK not supporting getmagic, so we have to getmagic
420 to test for undef :/ (reported by Matthias Waldorf).
421 - load Coro::AnyEvent in all modules using AnyEvent.
422 - work around perl corrupting our internal data structures,
423 reported by Tokuhiro Matsuno.
424 - enable per-coroutine real and cpu time gathering
425 (Coro::State::enable_times).
426
4275.14 Wed Jun 24 01:37:48 CEST 2009
428 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
429 as this seems to have been a difficult concept (poll, sleep, idle,
430 idle_upto).
431 - add Coro::AnyEvent::readable/writable functions.
432 - clarify Coro::EV/Event/AnyEvent manpages.
433 - free per-thread global scalars in the thread calling ->cancel, to
434 avoid crashes when $_, $@ etc., are magical but some of those
435 had already been freed.
436 - "unexperimentalise" the callback interface for Coro::Semaphore.
437 - speed up ready queue management/context switching by using a linked
438 list instead of an array (~5-10%).
439 - implement "watch" command in Coro::Debug shells.
440 - for fun, implement time-slicing as an example in the manpage.
441 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
442 Coro::Event, use the normal AnyEvent handling instead of dieing
443 (the same is true for Coro::Handle).
444 - properly document Coro::EV::timed_io_once.
445 - avoid unneccessary ->cancel calls in Coro::Handle.
446 - maybe make it work on mingw32 with win32 backend
447 (based on patch by Yasuhiro Matsumoto).
448
4495.132 Fri May 29 09:00:39 CEST 2009
450 - do not keep a reference to the argument itself in
451 Coro::Semaphore::guard, as it could change later.
452 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
453 should support a prepare callback.
454
4555.131 Mon Mar 16 23:20:37 CET 2009
456 - implement and document Coro->suspend, Coro->resume.
457 - fix Coro::Select implementation to not (often) close
458 the passed file descriptors (testcase provided by pippijn).
459
4605.13 Mon Dec 15 21:51:42 CET 2008
461 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
462 - don't set diehook to C<undef> but instead to NULL, to avoid
463 spurious warnings.
464 - fix a lot of bugs in Coro::SemaphoreSet.
465 - Coro::SemaphoreSet will less often create a semaphore needlessly.
466 - add Coro::SemaphoreSet::count and wait methods.
467 - take advantage of the new Guard module.
468 - deprecate Coro::guard.
469 - try to fix the dreaded 01_unblock tests once more. I hate it when
470 testsuites need more fixing than the code they are supposed to test.
471 - croak in more cases when a required callback isn't resolvable.
472 - fix some minor issues in Coro::State->call/eval.
473 - use current coroutine context instead of a temporary one
474 when temporarily switching to another coroutine.
475 - do not call C-level on_destroy handlers during global destruction,
476 to avoid needless segfaults.
477
4785.12 Sun Dec 7 13:30:38 CET 2008
479 - add default config for MirOS, which seems to be a bug-to-bug
480 compatible fork of openbsd ("world domination by releasing
481 openbsd cvs before the openbsd folks do it" or so :).
482 - free_padlist did destroy the names pad, not good, but didn't
483 seem to bother perl - this could fix issues such as eval ""
484 inside a function called from multiple coroutines.
485 - use a different method to detect destruction time.
486 - be more careful when freeing padlists just before global
487 destruction.
488 - fixed and expanded the call/cc example.
489 - renamed _terminate to _coro_run.
490 - new method Coro::Channel->shutdown.
491 - try pthreads on openbsd <4.4 (broken sigaltstack, will
492 pthreads fare better?).
493 - be less picky about destroying "a" running coroutine.
494
4955.11 Tue Nov 25 21:49:05 CET 2008
496 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
497 working code with failed assertions, introducing more bugs than
498 it fixes, requiring elaborate workarounds :(
499
5005.1 Mon Nov 24 08:54:59 CET 2008
501 - wrote a small introductory tutorial - Coro::Intro.
502 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
503 - Coro::Select did errornously dup the file descriptors
504 and didn't work with all AnyEvent backends.
505 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
506 LWP data transfers.
507 - disassociate c contexts from coro objects - this is agruably more
508 correct, but mostly allows sharing of cctxs between coro and state
509 objects, for added memory savings and speed increases.
510 - bumped $Coro::POOL_RSS up to 32kb by default.
511 - no longer set the optype to OP_CUSTOM, as B::* understandably
512 doesn't like this very much (and we *are* a type of entersub).
513 - implement state cloning, just to prove that call/cc can be done.
514 - automatically load Coro::AnyEvent in Coro::Handle.
515 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
516 do not have this method (and don't need it either).
517 - speed up generic anyevent methods in Coro::Handle by using rouse
518 callbacks.
519 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
520 others. It also makes the debugger happier, as you can trace
521 through the idle threads now.
522 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
523 - eg/event was pretty broken.
524 - better 5.8.6 compatibility.
525
5265.0 Thu Nov 20 10:35:05 CET 2008
527 - NEW ARCHITECTURE: use the latest 4.x version if you experience
528 stability issues.
529 - bump API version to 7 - all dependents must be recompiled.
530 - removed timed_* functions - they were not being used anyways
531 and should be replaced by a more generic mechanism -
532 and were annoying to support anyways :)
533 - removed SemaphoreSet's waiter method - use sem method instead.
534 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
535 - async_pool did free a scalar value twice
536 ("Attempt to unreference...").
537 - fix a longstanding bug where calling terminate on a coro that
538 was waiting for a semaphore that was just becoming available
539 would cause a deadlock (semaphore would get into a state where
540 it was available but waiters were still blocked).
541 - calling throw on a coroutine that is waiting for a semaphore will
542 no longer make it acquire the semaphore (and thus leak a count).
543 - perl's process emulation is now not even theoretically supported
544 anymore.
545 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
546 conversion of callback-style to blocking-style.
547 - new methods $coro->schedule_to and ->cede_to, to specifically
548 schedule or cede to a specific coroutine.
549 - new function Coro::Semaphore::wait.
550 - use named constants in Coro::Channel (Richard Hundt).
551 - directly patch the entersub opcode calling SLF functions (cede,
552 transfer and so on). this does speed up context switching, but
553 more importanly, it frees us from the hardcoded behaviour of
554 entersub, so we might actually be able to return something from
555 those functions and atcually create new ones.
556 - take advantage of __builtin_frame_address on gcc.
557 - expose THX in coroapi (not sure whether this was a wise decision,
558 as "threaded" perls are running at half speed anyways).
559 - implement execute_slf (schedule-like-function) interface that makes
560 it possible to implement schedule-like-functions in XS.
561 - use new SLF interface to massively speed up Coro::EV by roughly a
562 factor of two.
563 - used new SLF interface to massively speed up Coro::Semaphore by a
564 factor of three.
565 - used new SLF interface to speed up Coro::AIO by roughly a factor of
566 four and reduce its memory usage considerably.
567 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
568 for a nice speedup and vastly more correct behaviour. Also implement
569 a new method "sem" to get at the underlying semaphore object.
570 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
571 (in comparison) 20-40% speedup.
572 - used new SLF interface to reimplement Coro::Signal gaining
573 some unknown (because I was too lazy), but certain, speedup, and also
574 making signals reliable for the first time.
575 - used new SLF interface and other optimisations to speed up async_pool
576 by a factor of two. It also doesn't rely on perl's exception mechanism
577 to exit anymore. The overhead for terminating an async_pool, coro over
578 a normal async is now very small.
579 - sped up coroutine creation/destruction by 40%.
580 - forgot to include Coro/libcoro/README in the dist for all these years.
581 - work around a freebsd pthreads bug (manual testcancel is required as
582 pthread_cond_wait isn't a cancellation point on freebsd).
583 - use new rouse functions to speed up and simplify Coro::BDB.
584 - make "prefer perl native functions" work with threaded perls.
585 - condense Coro::Debug ps output, hint at v and w flags.
586 - (libcoro) lots of minor cleanups and portability improvements.
587
5884.914 Wed Nov 19 12:54:18 CET 2008
589 - fix a disastrous bug in the readline optimisation
590 introduced in 4.801.
591
5924.913 Sat Nov 15 07:58:28 CET 2008
593 - async_pool did free a scalar value twice
594 ("Attempt to unreference...").
595
5964.912 Thu Nov 13 18:31:23 CET 2008
597 - minor cleanups.
598 - use much larger stacks on linux and perl < 5.8.8.
599 - Coro::Debug::new_unix_server did not unlink the socket
600 when destroyed.
601
6024.911 Tue Nov 11 04:26:17 CET 2008
603 - "port" to threaded perls.
604
6054.91 Mon Nov 10 05:36:38 CET 2008
606 - the ->throw exception object no longer leaks.
607 - creating a new cctx leaked a scopestack entry (memleak).
608 - new coroutines didn't get created with a zero flags field
609 (unknown impact).
610 - calling ->throw on a not-yet-started coroutine should now work
611 instead of being ignored.
612 - ->throw is now supported on Coro::State objects.
613 - clean up cctx creation code a bit.
614 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
615
6164.9 Sat Nov 8 17:45:27 CET 2008
617 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
618 - (libcoro) no longer leak threads in the experimental pthread backend,
619 also speed it up considerably.
620 - (libcoro) do not rely on makecontext passing void *'s unscathed.
621 - fix compiletime dependencies on libcoro in the Makefile.
622 - cctx_count wasn't always updated properly.
623 - Coro::State::cctx_stacksize wasn't applied correctly.
624 - new function Coro::State::cctx_max_idle.
625 - the default max number of idle C contexts is now 4.
626 - (libcoro) try harder to get _setjmp/_longjmp.
627 - (libcoro) cleanup and extend the libcoro API to officially
628 allow the creation of empty source contexts.
629 - very experimental workaround for the totally broken netbsd platform.
630 - tried to hack around openbsd bugs.
631
6324.804 Wed Nov 5 16:36:00 CET 2008
633 - Coro::Debug::new_unix_server would not create a non-blocking listening
634 socket, sometimes causing freezes.
635 - (libcoro) fix misaligned stack points for setjmp and assembly
636 methods, which can cause crashes on x86/x86_64 with a sufficiently
637 aggressive compiler.
638 - new function: Coro::Debug::new_tcp_server.
639 - move ->throw into the Coro class because it only works on coro objects.
640
6414.803 Mon Nov 3 17:16:12 CET 2008
642 - (libcoro) use a global asm statement to become independent of gcc
643 otpimisations for CORO_ASM (thanks to pippijn for the idea).
644 - try to workaround yet another broken bsd, this time dragonfly.
645
6464.802 Thu Oct 30 10:56:12 CET 2008
647 - support -fno-omit-frame-pointer on x86 with the assembly method.
648 - tune 01_unblock.t tests a bit.
649
6504.801 Wed Oct 22 18:33:37 CEST 2008
651 - improve readline speed for very long "lines".
652 - backport to 5.8.8.
653
6544.8 Thu Oct 2 13:34:40 CEST 2008
655 - new function Coro::AIO::aio_wait.
656 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
657 - greatly speed up and reduce memory usage of Coro::AIO requests.
658 - implement some other µ-optimisations.
659
6604.749 Mon Sep 29 14:40:12 CEST 2008
661 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
662
6634.748 Sat Sep 27 14:03:19 CEST 2008
664 - implement, but do not document, PerlIO::cede(granularity).
665 - Coro::Storable forgot to wrap Storable::pstore.
666 - work around the multitude of leaks and memory corruption
667 bugs in PerlIO::via by using our own C-level perliol. As a side
668 effect, Coro::Storable is now much, much, much faster.
669
6704.747 Tue Sep 23 01:59:41 CEST 2008
671 - fix a per-cv memleak (one empty array was leaked per
672 code reference).
673 - avoid a crash in coro->call|rss when the coroutine was already
674 destroyed (most noticably when using Coro::Debug::ps :)
675 - also protect *Storable::FILE.
676 - push up default storable granularity to 20ms.
677
6784.746 Sun Sep 21 03:22:20 CEST 2008
679 - be more insistent on locking Storable against reentrancy
680 in Coro::Storable.
681 - move swap_def?v and throw to Coro::State, as documented.
682
6834.745 Thu Jul 24 00:14:38 CEST 2008
684 - remove debugging code related to MgPV_nolen_const, also try to
685 make it compile with perl 5.8.6 (yes, apple apparently loves
686 outdated software). Reported by John S.
687
6884.744 Tue Jul 8 22:06:35 CEST 2008
689 - correctly provide default DIE/WARN handlers as documented.
690 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
691 implementations inside perl work fine for us.
692
6934.743 Mon Jun 16 00:21:57 CEST 2008
694 - when using Coro::EV without running EV::loop it could
695 result in busy-waiting for events, this has been fixed.
696 - reduce codesize and improve performance by using EV_DEFAULT_UC.
697
6984.742 Sat May 31 14:10:21 CEST 2008
699 - implement a workaround for (some) perl <5.8.8 versions.
700 - require EV 3.3+.
701
7024.741 Fri May 30 23:33:09 CEST 2008
703 - tell netbsd how utterly broken their imitation of an OS is
704 and refuse to build by default if pthreads are in use.
705 - switch to "s" method on all bsd's by default, as their ucontext
706 stuff seems just too broken.
707 - fix a bug in Coro::Select.
708
7094.74 Thu May 29 20:05:31 CEST 2008
710 - do not test Coro::LWP for lack of dependencies.
711 - do not test Coro::Select for lack of working perls.
712
7134.73 Thu May 29 2008
714 - fix a bug in Coro::EV which would cause it to block despite
715 there being runnable coroutines.
716 - sprinkle "no warnings" freely over everything, also suppress
717 warnings for some other modules.
718 - fix typo in WSAEWOULDBLOCK.
719
7204.72 Sun May 25 05:14:36 CEST 2008
721 - tweak META.yaml a bit, unfortunately, there is no documented way
722 to have optional dependencies with CPAN. doh :(
723 - avoid running some tests on windows because they would fail due to
724 perl bug (broken fork, broken pipes...).
725 - work around perl on windows bugs where perl returns undocumented
726 error codes for sysread, syswrite etc. by taking advantage
727 of AnyEvent's workaround for that problem.
728 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
729 a common perl implementation bug on windows.
730 - use unix domain sockets in testsuite to work around
731 common perl implementation bugs on widows (they are emulated by
732 tcp sockets on windows. ugh.)
733
7344.71 Sat May 24 20:01:27 CEST 2008
735 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
736 - take advantage of async name resolution of AnyEvent::Util.
737 - work around brutal inet_aton override in Coro::LWP.
738 - take advantage of the readyhook in Coro::EV, for smoother
739 scheduling.
740
7414.7 Sun May 11 00:32:19 CEST 2008
742 - completely reworked the Coro manpage.
743 - added Coro::AnyEvent, generic event loop integration.
744 - implement cancel, ready and kill commands in Coro::Debug.
745 - document find_coro in Coro::Debug.
746 - incompatible change: rename has_stack to has_cctx.
747 - Coro::AIO and Coro::BDB no longer force event model detection,
748 use AnyEvent::AIO and AnyEvent::BDB.
749
7504.6 Sat Apr 26 10:05:14 CEST 2008
751 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
752 by the Coro module. It was a mistake to have it in the first place.
753 - (experimental) support for activestate perl 5.10 (method "w").
754 - (experimental) support for strawberry perl (method "a").
755 - coro_sigelem_set did not return a value although it had to,
756 actual impact unknown.
757
7584.51 Mon Apr 14 13:28:27 CEST 2008
759 - make it compile again on 5.8.
760
7614.50 Thu Apr 10 09:43:17 CEST 2008
762 - I did it twice! (see 4.49).
763
7644.49 Sun Apr 6 21:23:31 CEST 2008
765 - grr, instead of compiling the recent changes on 5.10 only they
766 were compiled on 5.8 only.
767
7684.48 Sun Apr 6 20:36:46 CEST 2008
769 - allow coroutine switches during eval's under 5.10.x, as it seems
770 the parser is a per-interpreter option now, so this is safe
771 (this might fix the odd crashing bug).
772 - drop support for 5.9.x versions: they are dead, jim.
773
7744.47 Sun Apr 6 00:37:52 CEST 2008
775 - force cctx allocation on API calls: we know we need to force one
776 and gcc actually manages to confuse our heuristic nowadays,
777 leading to crashes and worse.
778 - document force_cctx.
779
7804.46 Fri Apr 4 22:05:43 CEST 2008
781 - upgrade libcoro, resulting in pthread-backend (which was only created
782 to fulfill the rules of the programming languages shootout).
783
7844.45 Thu Mar 13 11:55:36 CET 2008
785 - fix a file leaking bug in eg/mhyttpd that would allow
786 downloading of any file (reported by oesi).
787 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
788 (also add testcase).
789 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
790 apparently having loads of fun with that).
791 - implement $state->swap_defsv, ->swap_defav and document ->throw.
792
7934.4 Wed Feb 13 16:44:29 CET 2008
794 - only assume .cfi pseudo isns exist on GNU/Linux.
795 - add get_timed method to Coro::Channels.
796 - fixed Coro::Debug manpage.
797 - perl 5.11 compatbility improvement based on patch by
798 Andreas König.
799
8004.37 Sun Jan 20 11:25:23 CET 2008
801 - soften the check for an active parser for perl 5.10 (avoiding
802 "Coro::State::transfer called while parsing" in many cases).
803
8044.36 Sun Jan 13 10:53:56 CET 2008
805 - reset diehook when terminating from an async_pool as to not
806 trigger any __DIE__ handlers.
807
8084.35 Sun Jan 13 04:14:13 CET 2008
809 - "bt" debug command now displays any exceptions
810 from longmess and also skips the innermost
811 stackframes, giving more useufl output.
812 - allow backtraces in coroutines blocked in Coro::EV,
813 at a <1% speed hit.
814 - handle localising of $SIG{__WARN__} and __DIE__
815 properly (with a proper amount of dirty hacking).
816
8174.34 Sat Dec 22 17:49:53 CET 2007
818 - upgrade to EV version 2.0 API.
819
8204.33 Mon Dec 17 08:36:12 CET 2007
821 - make Coro::AIO etc. loadable in the absence of EV.
822
8234.32 Mon Dec 17 07:46:02 CET 2007
824 - majorly improved Coro::Handle's performance with Coro::EV.
825 - implemented the until now mythical Coro::BDB module.
826 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
827
8284.31 Wed Dec 5 12:32:39 CET 2007
829 - remove warn statement form Coro::Util.
830
8314.3 Tue Dec 4 20:33:14 CET 2007
832 - calls to the idle function could cause stack corruption
833 when the stack changed.
834 - do no longer rely on the presence of EV::DNS if EV is used
835 (because it is gone), but instead take avdantage of EV::ADNS
836 if available.
837 - add ($) prototypes to all functions in Coro::Storable.
838 - use a conventional (safer) idle callback in Coro::EV.
839 - do accept quickly in Coro::Debug to avoid endless loops.
840
8414.22 Fri Nov 30 16:04:04 CET 2007
842 - really use optimised versions for Event and EV in Coro::Util
843 and Coro::Handle.
844
8454.21 Sun Nov 25 10:48:59 CET 2007
8 - fix a spurious memory read. 846 - fix a spurious memory read.
9 - Coro::EV no longer keeps the eventloop "alive". 847 - Coro::EV no longer keeps the eventloop "alive".
10 848
114.2 Fri Nov 9 20:47:05 CET 2007 8494.2 Fri Nov 9 20:47:05 CET 2007
12 - enable/disable tracing from a new coroutine, not a pooled one. 850 - enable/disable tracing from a new coroutine, not a pooled one.
… …
34 872
354.11 Thu Oct 11 02:40:24 CEST 2007 8734.11 Thu Oct 11 02:40:24 CEST 2007
36 - port to threaded perls. 874 - port to threaded perls.
37 875
384.1 Thu Oct 11 02:38:16 CEST 2007 8764.1 Thu Oct 11 02:38:16 CEST 2007
39 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 877 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
40 be local to each coro (see Coro::State). 878 be local to each coro (see Coro::State).
41 - incompatible change: for very deep reasons, cede and cede_notself 879 - incompatible change: for very deep reasons, cede and cede_notself
42 cannot return anything, so nothing will be returned. 880 cannot return anything, so nothing will be returned.
43 - possibly bring back 5.10 compatibility (untested). 881 - possibly bring back 5.10 compatibility (untested).
44 - work around stupid (and wrong) warning on 5.10 :(. 882 - work around stupid (and wrong) warning on 5.10 :(.
… …
116 but stateful semantics. 954 but stateful semantics.
117 - fixed a lot of typos in Coro.pm (patch submitted by David 955 - fixed a lot of typos in Coro.pm (patch submitted by David
118 Steinbrunner, which applied flawlessly). 956 Steinbrunner, which applied flawlessly).
119 957
1203.6 Sat Apr 14 17:13:31 CEST 2007 9583.6 Sat Apr 14 17:13:31 CEST 2007
121 - added some bugfixes to get eg/myhttpd working again. 959 - added some bugfixes to get eg/myhttpd working again.
122 - added Coro::Storable for often-cede'ing freeze/thaw. 960 - added Coro::Storable for often-cede'ing freeze/thaw.
123 - try to do a clean exit when a coroutine calls exit 961 - try to do a clean exit when a coroutine calls exit
124 (EXPERIMENTAL). 962 (EXPERIMENTAL).
125 - got rid of indirect call through _coro_init. 963 - got rid of indirect call through _coro_init.
126 - updated the partly antique examples in eg/ to 964 - updated the partly antique examples in eg/ to

Diff Legend

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