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

Comparing Coro/Changes (file contents):
Revision 1.376 by root, Sat Sep 27 12:03:49 2008 UTC vs.
Revision 1.630 by root, Thu May 1 08:09:18 2014 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::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
8
9 - check that perl slots actually have enough space to hold
10 interpreter variables.
11 - untested port to perl 5.19 (64 bit tmps indices) (reported
12 by Andreas König).
13 - croak when cancel is called without a thread context.
14
156.37 Tue Mar 4 13:27:33 CET 2014
16 - *sigh*, removed leftover debugging code from debugging a
17 perl bug, of all things.
18
196.36 Tue Mar 4 07:11:59 CET 2014
20 - semaphores would not clear the destroy hook when interrupted
21 by ->throw, causing segfaults or worse.
22 - ->throw on a thread waiting for a semaphore did not acquire
23 the semaphore, but also didn't wake up other waiters,
24 possibly causing a deadlock.
25 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
26 to report a full stacktrace, which should help find out
27 where the actual call is.
28 - "atomically" destroy data in slf_destroy, because it is
29 the right thing to do, just in case.
30 - disable ecb.h libm dependency, to compile on ancient systems
31 or under adverse conditions.
32
336.33 Mon Nov 18 11:26:27 CET 2013
34 - do not crash when freeing padlists with holes (in 5.18).
35 - tentative SVt_BIND 5.19 port/fix.
36
376.32 Tue Nov 5 15:35:35 CET 2013
38 - use a new algorithm to derive padlists for perl 5.18. The old
39 one could lead to 0-pointer accesses inside perl (reported
40 by Darin McBride).
41
426.31 Thu May 9 07:39:48 CEST 2013
43 - Coro::AIO requests would crash if the thread was ready'd
44 while the request was ongoing.
45
466.29 Wed May 8 02:55:18 CEST 2013
47 - when an on_destroy handler destructs the coro currently being
48 destructed a perl scalar could be accessed after being freed,
49 likely causing a crash.
50
516.28 Wed Mar 6 06:58:02 CET 2013
52 - clean remnants of existing __DIE__ and __WARN__ handlers so
53 they lose their magic and will not cause segfaults later
54 (testcase by Andrey Sagulin).
55 - improved Coro::State documentation a bit.
56 - Coro::Debug::command now flushes the output.
57 - add hack detection code for x32 abi, because the braindead slugs
58 who designed that made it look exactly like x86_64 without
59 providing proper compile time symbols to test for it. as a result,
60 this detection cannot work reliably.
61 - valgrind stack registering was broken.
62 - do not rely on Time::HiRes anymore in Coro::Debug.
63
646.23 Fri Dec 7 23:36:37 CET 2012
65 - use experimental fiber implementation on native windows
66 perls.
67 - use sizeof (void *) as multiplication factor for stack sizes,
68 to accomodate the totally braindamaged microsoft 64 bit "os".
69 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
70 activeperl 5.16 crashes when PerlIO_define_layer is called due
71 to some bug in the perl dll, strawberry perl at least passes
72 the testsuite.
73 - implement Coro::Handle->peeraddr/host/port, for slightly
74 improved compatibility with LWP.
75 - implement 5.17 compatibility by almost blindly applying a
76 good-looking patch by Father Chrysostomos.
77 - move stack management functions into libcoro 3.
78 - libcoro version 3 "released".
79 - support magic values as timed_io_once args.
80 - recommend AnyEvent 7+ or EV 4+, also require EV
81 version 4 or newer for Coro::EV.
82
836.10 Tue Oct 9 01:14:27 CEST 2012
84 - updated ecb.h, it had a typo that caused it to not compile on many
85 big endian systems (reported by many people).
86 - disable memory fences in ecb.h to improve portability.
87
886.09 Sat Oct 6 23:25:02 CEST 2012
89 - Coro::EV I/O watchers were not interruptible by exceptions
90 (Coro::State::throw) (testcase by sten).
91 - ->throw now puts threads into the ready queue, as this seems to
92 be expected by existing code, and code that doesn't cope with spurious
93 wakeups needs fixing anyway.
94 - use fd -1 in mmap.
95 - cast I32 to int in error message printf.
96 - warn about broken so-called "hardened" kernels.
97
986.08 Fri Apr 13 12:05:47 CEST 2012
99 - be more aggressive about exiting like perl does - formerly,
100 exiting from the non-main thread would not execute END blocks.
101
1026.07 Fri Nov 11 21:21:48 CET 2011
103 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
104 - update ecb.h.
105
1066.06 Mon Aug 8 23:59:48 CEST 2011
107 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
108 backend, which is an order of magnitude slower. unfortunately, cygwins
109 pthread implementation isn't very complete either, so allocate the stack
110 twice just to be sure.
111 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
112 in its ithreaded perl - best recompile cygwin and use the 'w'indows
113 backend for much better performance. also disable ithreads for
114 even better performance...).
115
1166.05 Thu Aug 4 21:36:36 CEST 2011
117 - blush, condvar values would not be propagated from send to recv anymore
118 (reported by Chip Salzenberg).
119 - use exponential increase for the readline buffer length in
120 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
121
1226.04 Wed Aug 3 17:15:45 CEST 2011
123 - use even more efficient and more compatible condvars for
124 compatibility to AnyEvent 6.x :)
125 - more inconsequential ecb.h updates.
126
1276.03 Wed Aug 3 11:41:30 CEST 2011
128 - change how Coro patches AnyEvent condvars for compatibility to
129 AnyEvent 6.x.
130 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
131 hasn't been included.
132
1336.02 Wed Jul 13 04:35:19 CEST 2011
134 - "improve portability to Gentoo" - gentoo manages to put perl variables
135 in memory areas that are farther than 2gb apart, which the jit couldn't
136 handle and barfed. now it's just a bit slower on gentoo and similar
137 systems.
138
1396.01 Sun Jul 3 12:31:14 CEST 2011
140 - workarounds are good, but the test for whether pthreads are used
141 was not good. thisone should be better.
142 - check differently whether gcc generates cfi instructions itself.
143
1446.0 Wed Jun 29 19:43:35 CEST 2011
145 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
146 destroyed and cleaned up automatically (e.g. async { schedule }).
147 - implement a JIT compiler for part of the thread switch code,
148 which gives a 50% speed improvement on threaded perls, and
149 about 4% on non-thraeded perls (so threaded perls now finally
150 reach about half the speed of non-threaded perls).
151 - slightly modernise Coro::Intro, add section about rouse functions.
152 - avoid DEFSV and ERRSV, giving another 10% improvement
153 in thread switching.
154 - Coro::State->is_destroyed is now called is_zombie.
155 - implement a Coro->safe_cancel method that might fail, but
156 cancels in a "safer" way if it succeeds.
157 - add preliminary support for DEBUGGING perls.
158 - get rid of two hash-accesses when initialising a new Coro - this
159 speeds up coro creation by almost a factor of two.
160 - croak when a coro that is being cancelled tries to block
161 (e.g. while executing a guard block), instead of crashing or
162 deadlocking.
163 - use a more robust and also faster method to identify Coro::State
164 objects - speeds up everything a bit.
165 - implement Coro->cancel in XS for a 20% speed improvement, and to
166 be able to implement mutual cancellation.
167 - speed up context switches by a percent or two by more efficiently
168 allocating context stack entries.
169 - implement Coro->join and Coro->on_destroy in XS for a speedup and
170 a reduction in memory use.
171 - cancelling a coro while it itself is cancelling another coro is
172 now supported and working, instead of triggering an assertion.
173 - be a bit more crash-resistant when calling (buggy) on_destroy
174 callbacks (best effort).
175 - move on_destroy into the slf_frame, to allow extension slf
176 functions to have destructors.
177 - get rid if coro refcounting - simply crash in other interpreter
178 threads by nulling the pointers on clone.
179 - simplify warn/die hook handling when loading Coro - the convoluted
180 logic seems to be no longer neccessary.
181 - use libecb instead of our own home-grown gcc hacks.
182 - document alternatives to Coro::LWP. Please use them :)
183 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
184 sigaltstack. Really. wine suffers from it, erlang suffers from it,
185 and it's known since at least 2006.
186
1875.372 Wed Feb 23 06:14:30 CET 2011
188 - apparently mingw doesn't provide a working gettimeofday, try to
189 work around that by relying on Time::HiRes (indirectly brought to
190 my attention by Max Maischein).
191 - fix some portability issues when Time::HiRes was used.
192
1935.371 Mon Feb 21 14:36:08 CET 2011
194 - backport to windows process emulation code again.
195
1965.37 Sat Feb 19 07:49:44 CET 2011
197 - add a big "Coro thread life cycle" section to "man Coro".
198 - try a tentative workaround against the breakage that 5.13 has
199 introduced without depreciation period. sigh.
200 - no longer use Time::HiRes if gettimeofday is available, which
201 saves quite a lot of memory.
202
2035.36 Sun Feb 13 05:33:41 CET 2011
204 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
205 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
206 method call.
207 - undocument Coro::Timer::sleep and obsolete whole module.
208 - optimise Coro::Timer::timeout memory and cpu usage.
209 - slightly updated Coro::Intro for recent changes.
210 - do not initialise PL_dirty anymore.
211
2125.25 Thu Nov 11 01:08:39 CET 2010
213 - try a different approach on netbsd - netbsd 5 finally has marginally
214 working pthreads, but still broken ucontext/sigaltstack.
215 - openbsd 4.8 finally got their act together, Coro works out of the box
216 with asm, setjmp and pthreads (no change, just informational).
217
2185.24 Sat Oct 23 11:27:12 CEST 2010
219 - port to the EV 4.0 API.
220 - work around bugs in mingw32, making strawberry perl work
221 out of the box.
222 - correctly modify Coro::AIO function prototypes
223 so that they reflect the "no optional parameters" rule.
224 - "ported" libcoro to C++.
225
2265.23 Mon May 17 18:50:42 CEST 2010
227 - be more resistant to ordering changes when initialising
228 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
229 Waldorf).
230 - document that perl 5.12 deliberately removed support for cloning.
231
2325.22 Wed Apr 14 03:55:35 CEST 2010
233 - correctly return udnef on errors in Coro::Handle::read/write
234 (testcase by Marc Mims).
235 - convert Coro::Util into a "perl compatibility wrapper" - the functions
236 are less useful now, but are drop-in replacements for existing
237 functions, listing better alternatives in the documentation. This also
238 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
239 with Coro::Util::inet_aton.
240 - do not override $Coro::idle unconditionally in Coro.pm, as other
241 modules could have provided their own idle coro already
242 (for exmaple, Coro::AnyEvent).
243 - fix Coro::Util::gethost* functions.
244 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
245
2465.21 Wed Dec 16 07:19:51 CET 2009
247 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
248 together.
249 - add some examples on how to combine other event loops with Coro in
250 Coro::AnyEvent, and how to run it (and not to block). Seems to be
251 the most common source of confusion.
252 - try to catch people naively blocking in an event callback.
253 - work around the perl filehandle bug issue in conjunction with
254 older common::sense (as indirectly pointed out by ZSystem).
255 - clarify the "not from signal handlers" section.
256
2575.2 Sun Oct 4 14:54:24 CEST 2009
258 - Coro::Storable destroyed the prototypes of the functions it wrapped.
259 - export rouse_cb and rouse_wait by default now.
260 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
261 - new method $state->swap_sv.
262 - added section on "windows process emulation" to the manpage,
263 after a not-so-fruitful (nor-friendly) "discussion" with chip
264 salzenberg (discussion implies arguments, but his only arguments
265 were ad-hominems, one wonders why he started it in the first
266 place). I hope this explains it well enough for him to understand,
267 and maybe well enough for others to understand.
268 - use common::sense everywhere now.
269 - idle callbacks are no longer supported, use idle coros instead.
270 - print a thread listing when a deadlock is detected.
271
2725.17 Sat Aug 22 23:09:31 CEST 2009
273 - work around a bug in the perl debugger causing crashes
274 when running under the debugger by marking _pool_handler as nodebug.
275 - speed up Coro::async considerably.
276 - try some hacks to get netbsd to work "more often" - their broken
277 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
278 - convert Coro to AE 5.0 API.
279
2805.162 Tue Jul 28 04:04:03 CEST 2009
281 - perl 5.8.2 is now minimum requirement.
282 - skip t/19_handle.t on broken windows perls.
283
2845.161 Wed Jul 22 04:47:38 CEST 2009
285 - Coro::AnyEvent::poll could have a different prototype when EV was
286 used as backend (analyzed by Tatsuhiko Miyagawa).
287 - Coro::AnyEvent errornously initialised the event loop when loaded,
288 not on demand.
289 - try to workaround rare */t/01_unblock.t failures.
290
2915.16 Tue Jul 21 01:44:37 CEST 2009
292 - Coro::AnyEvent failed to hook into the event loop
293 when no threads had been readied between detecting
294 the event loop and actually running it.
295 - considerably speed up Coro::Select by taking avdantage
296 of AnyEvent > 4.8 and some other optimisations.
297 - implement paragraph readline mode in Coro::Handle
298 (based on patches by Zsbán Ambrus).
299 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
300 and analyzed by Yasuhiro MATSUMOTO).
301 - clarified libcoro license and copyright.
302 - someone stole my EXTRA_META!!!
303 - implement Coro::Select::patch_pp_sselect and it's brother,
304 for hardcode select overriding.
305
3065.151 Mon Jul 6 05:41:57 CEST 2009
307 - backport to windows process emulation code again (patch by
308 Yasuhiro MATSUMOTO).
309 - slightly update Coro::MakeMaker.
310
3115.15 Tue Jun 30 10:28:06 CEST 2009
312 - deprecate Coro::Socket, document how to get ipv6 support via
313 AnyEvent::Socket instead.
314 - implement signal->wait ($cb) interface, similar to semaphores.
315 - work around SvOK not supporting getmagic, so we have to getmagic
316 to test for undef :/ (reported by Matthias Waldorf).
317 - load Coro::AnyEvent in all modules using AnyEvent.
318 - work around perl corrupting our internal data structures,
319 reported by Tokuhiro Matsuno.
320 - enable per-coroutine real and cpu time gathering
321 (Coro::State::enable_times).
322
3235.14 Wed Jun 24 01:37:48 CEST 2009
324 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
325 as this seems to have been a difficult concept (poll, sleep, idle,
326 idle_upto).
327 - add Coro::AnyEvent::readable/writable functions.
328 - clarify Coro::EV/Event/AnyEvent manpages.
329 - free per-thread global scalars in the thread calling ->cancel, to
330 avoid crashes when $_, $@ etc., are magical but some of those
331 had already been freed.
332 - "unexperimentalise" the callback interface for Coro::Semaphore.
333 - speed up ready queue management/context switching by using a linked
334 list instead of an array (~5-10%).
335 - implement "watch" command in Coro::Debug shells.
336 - for fun, implement time-slicing as an example in the manpage.
337 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
338 Coro::Event, use the normal AnyEvent handling instead of dieing
339 (the same is true for Coro::Handle).
340 - properly document Coro::EV::timed_io_once.
341 - avoid unneccessary ->cancel calls in Coro::Handle.
342 - maybe make it work on mingw32 with win32 backend
343 (based on patch by Yasuhiro Matsumoto).
344
3455.132 Fri May 29 09:00:39 CEST 2009
346 - do not keep a reference to the argument itself in
347 Coro::Semaphore::guard, as it could change later.
348 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
349 should support a prepare callback.
350
3515.131 Mon Mar 16 23:20:37 CET 2009
352 - implement and document Coro->suspend, Coro->resume.
353 - fix Coro::Select implementation to not (often) close
354 the passed file descriptors (testcase provided by pippijn).
355
3565.13 Mon Dec 15 21:51:42 CET 2008
357 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
358 - don't set diehook to C<undef> but instead to NULL, to avoid
359 spurious warnings.
360 - fix a lot of bugs in Coro::SemaphoreSet.
361 - Coro::SemaphoreSet will less often create a semaphore needlessly.
362 - add Coro::SemaphoreSet::count and wait methods.
363 - take advantage of the new Guard module.
364 - deprecate Coro::guard.
365 - try to fix the dreaded 01_unblock tests once more. I hate it when
366 testsuites need more fixing than the code they are supposed to test.
367 - croak in more cases when a required callback isn't resolvable.
368 - fix some minor issues in Coro::State->call/eval.
369 - use current coroutine context instead of a temporary one
370 when temporarily switching to another coroutine.
371 - do not call C-level on_destroy handlers during global destruction,
372 to avoid needless segfaults.
373
3745.12 Sun Dec 7 13:30:38 CET 2008
375 - add default config for MirOS, which seems to be a bug-to-bug
376 compatible fork of openbsd ("world domination by releasing
377 openbsd cvs before the openbsd folks do it" or so :).
378 - free_padlist did destroy the names pad, not good, but didn't
379 seem to bother perl - this could fix issues such as eval ""
380 inside a function called from multiple coroutines.
381 - use a different method to detect destruction time.
382 - be more careful when freeing padlists just before global
383 destruction.
384 - fixed and expanded the call/cc example.
385 - renamed _terminate to _coro_run.
386 - new method Coro::Channel->shutdown.
387 - try pthreads on openbsd <4.4 (broken sigaltstack, will
388 pthreads fare better?).
389 - be less picky about destroying "a" running coroutine.
390
3915.11 Tue Nov 25 21:49:05 CET 2008
392 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
393 working code with failed assertions, introducing more bugs than
394 it fixes, requiring elaborate workarounds :(
395
3965.1 Mon Nov 24 08:54:59 CET 2008
397 - wrote a small introductory tutorial - Coro::Intro.
398 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
399 - Coro::Select did errornously dup the file descriptors
400 and didn't work with all AnyEvent backends.
401 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
402 LWP data transfers.
403 - disassociate c contexts from coro objects - this is agruably more
404 correct, but mostly allows sharing of cctxs between coro and state
405 objects, for added memory savings and speed increases.
406 - bumped $Coro::POOL_RSS up to 32kb by default.
407 - no longer set the optype to OP_CUSTOM, as B::* understandably
408 doesn't like this very much (and we *are* a type of entersub).
409 - implement state cloning, just to prove that call/cc can be done.
410 - automatically load Coro::AnyEvent in Coro::Handle.
411 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
412 do not have this method (and don't need it either).
413 - speed up generic anyevent methods in Coro::Handle by using rouse
414 callbacks.
415 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
416 others. It also makes the debugger happier, as you can trace
417 through the idle threads now.
418 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
419 - eg/event was pretty broken.
420 - better 5.8.6 compatibility.
421
4225.0 Thu Nov 20 10:35:05 CET 2008
423 - NEW ARCHITECTURE: use the latest 4.x version if you experience
424 stability issues.
425 - bump API version to 7 - all dependents must be recompiled.
426 - removed timed_* functions - they were not being used anyways
427 and should be replaced by a more generic mechanism -
428 and were annoying to support anyways :)
429 - removed SemaphoreSet's waiter method - use sem method instead.
430 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
431 - async_pool did free a scalar value twice
432 ("Attempt to unreference...").
433 - fix a longstanding bug where calling terminate on a coro that
434 was waiting for a semaphore that was just becoming available
435 would cause a deadlock (semaphore would get into a state where
436 it was available but waiters were still blocked).
437 - calling throw on a coroutine that is waiting for a semaphore will
438 no longer make it acquire the semaphore (and thus leak a count).
439 - perl's process emulation is now not even theoretically supported
440 anymore.
441 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
442 conversion of callback-style to blocking-style.
443 - new methods $coro->schedule_to and ->cede_to, to specifically
444 schedule or cede to a specific coroutine.
445 - new function Coro::Semaphore::wait.
446 - use named constants in Coro::Channel (Richard Hundt).
447 - directly patch the entersub opcode calling SLF functions (cede,
448 transfer and so on). this does speed up context switching, but
449 more importanly, it frees us from the hardcoded behaviour of
450 entersub, so we might actually be able to return something from
451 those functions and atcually create new ones.
452 - take advantage of __builtin_frame_address on gcc.
453 - expose THX in coroapi (not sure whether this was a wise decision,
454 as "threaded" perls are running at half speed anyways).
455 - implement execute_slf (schedule-like-function) interface that makes
456 it possible to implement schedule-like-functions in XS.
457 - use new SLF interface to massively speed up Coro::EV by roughly a
458 factor of two.
459 - used new SLF interface to massively speed up Coro::Semaphore by a
460 factor of three.
461 - used new SLF interface to speed up Coro::AIO by roughly a factor of
462 four and reduce its memory usage considerably.
463 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
464 for a nice speedup and vastly more correct behaviour. Also implement
465 a new method "sem" to get at the underlying semaphore object.
466 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
467 (in comparison) 20-40% speedup.
468 - used new SLF interface to reimplement Coro::Signal gaining
469 some unknown (because I was too lazy), but certain, speedup, and also
470 making signals reliable for the first time.
471 - used new SLF interface and other optimisations to speed up async_pool
472 by a factor of two. It also doesn't rely on perl's exception mechanism
473 to exit anymore. The overhead for terminating an async_pool, coro over
474 a normal async is now very small.
475 - sped up coroutine creation/destruction by 40%.
476 - forgot to include Coro/libcoro/README in the dist for all these years.
477 - work around a freebsd pthreads bug (manual testcancel is required as
478 pthread_cond_wait isn't a cancellation point on freebsd).
479 - use new rouse functions to speed up and simplify Coro::BDB.
480 - make "prefer perl native functions" work with threaded perls.
481 - condense Coro::Debug ps output, hint at v and w flags.
482 - (libcoro) lots of minor cleanups and portability improvements.
483
4844.914 Wed Nov 19 12:54:18 CET 2008
485 - fix a disastrous bug in the readline optimisation
486 introduced in 4.801.
487
4884.913 Sat Nov 15 07:58:28 CET 2008
489 - async_pool did free a scalar value twice
490 ("Attempt to unreference...").
491
4924.912 Thu Nov 13 18:31:23 CET 2008
493 - minor cleanups.
494 - use much larger stacks on linux and perl < 5.8.8.
495 - Coro::Debug::new_unix_server did not unlink the socket
496 when destroyed.
497
4984.911 Tue Nov 11 04:26:17 CET 2008
499 - "port" to threaded perls.
500
5014.91 Mon Nov 10 05:36:38 CET 2008
502 - the ->throw exception object no longer leaks.
503 - creating a new cctx leaked a scopestack entry (memleak).
504 - new coroutines didn't get created with a zero flags field
505 (unknown impact).
506 - calling ->throw on a not-yet-started coroutine should now work
507 instead of being ignored.
508 - ->throw is now supported on Coro::State objects.
509 - clean up cctx creation code a bit.
510 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
511
5124.9 Sat Nov 8 17:45:27 CET 2008
513 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
514 - (libcoro) no longer leak threads in the experimental pthread backend,
515 also speed it up considerably.
516 - (libcoro) do not rely on makecontext passing void *'s unscathed.
517 - fix compiletime dependencies on libcoro in the Makefile.
518 - cctx_count wasn't always updated properly.
519 - Coro::State::cctx_stacksize wasn't applied correctly.
520 - new function Coro::State::cctx_max_idle.
521 - the default max number of idle C contexts is now 4.
522 - (libcoro) try harder to get _setjmp/_longjmp.
523 - (libcoro) cleanup and extend the libcoro API to officially
524 allow the creation of empty source contexts.
525 - very experimental workaround for the totally broken netbsd platform.
526 - tried to hack around openbsd bugs.
527
5284.804 Wed Nov 5 16:36:00 CET 2008
529 - Coro::Debug::new_unix_server would not create a non-blocking listening
530 socket, sometimes causing freezes.
531 - (libcoro) fix misaligned stack points for setjmp and assembly
532 methods, which can cause crashes on x86/x86_64 with a sufficiently
533 aggressive compiler.
534 - new function: Coro::Debug::new_tcp_server.
535 - move ->throw into the Coro class because it only works on coro objects.
536
5374.803 Mon Nov 3 17:16:12 CET 2008
538 - (libcoro) use a global asm statement to become independent of gcc
539 otpimisations for CORO_ASM (thanks to pippijn for the idea).
540 - try to workaround yet another broken bsd, this time dragonfly.
541
5424.802 Thu Oct 30 10:56:12 CET 2008
543 - support -fno-omit-frame-pointer on x86 with the assembly method.
544 - tune 01_unblock.t tests a bit.
545
5464.801 Wed Oct 22 18:33:37 CEST 2008
547 - improve readline speed for very long "lines".
548 - backport to 5.8.8.
549
5504.8 Thu Oct 2 13:34:40 CEST 2008
551 - new function Coro::AIO::aio_wait.
552 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
553 - greatly speed up and reduce memory usage of Coro::AIO requests.
554 - implement some other µ-optimisations.
555
5564.749 Mon Sep 29 14:40:12 CEST 2008
557 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
7 558
84.748 Sat Sep 27 14:03:19 CEST 2008 5594.748 Sat Sep 27 14:03:19 CEST 2008
9 - implement, but do not document, PerlIO::cede(granularity). 560 - implement, but do not document, PerlIO::cede(granularity).
10 - Coro::Storable forgot to wrap Storable::pstore. 561 - Coro::Storable forgot to wrap Storable::pstore.
11 - work around the multitude of leaks and memory corruption 562 - work around the multitude of leaks and memory corruption
… …
29 - remove debugging code related to MgPV_nolen_const, also try to 580 - remove debugging code related to MgPV_nolen_const, also try to
30 make it compile with perl 5.8.6 (yes, apple apparently loves 581 make it compile with perl 5.8.6 (yes, apple apparently loves
31 outdated software). Reported by John S. 582 outdated software). Reported by John S.
32 583
334.744 Tue Jul 8 22:06:35 CEST 2008 5844.744 Tue Jul 8 22:06:35 CEST 2008
34 - correctly provide default DIE/WARN handlers as documented. 585 - correctly provide default DIE/WARN handlers as documented.
35 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 586 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
36 implementations inside perl work fine for us. 587 implementations inside perl work fine for us.
37 588
384.743 Mon Jun 16 00:21:57 CEST 2008 5894.743 Mon Jun 16 00:21:57 CEST 2008
39 - when using Coro::EV without running EV::loop it could 590 - when using Coro::EV without running EV::loop it could
… …
61 - sprinkle "no warnings" freely over everything, also suppress 612 - sprinkle "no warnings" freely over everything, also suppress
62 warnings for some other modules. 613 warnings for some other modules.
63 - fix typo in WSAEWOULDBLOCK. 614 - fix typo in WSAEWOULDBLOCK.
64 615
654.72 Sun May 25 05:14:36 CEST 2008 6164.72 Sun May 25 05:14:36 CEST 2008
66 - tweak META.yaml a bit, unfortunately, there is no documented way 617 - tweak META.yaml a bit, unfortunately, there is no documented way
67 to have optional dependencies with CPAN. doh :( 618 to have optional dependencies with CPAN. doh :(
68 - avoid running some tests on windows because they would fail due to 619 - avoid running some tests on windows because they would fail due to
69 perl bug (broken fork, broken pipes...). 620 perl bug (broken fork, broken pipes...).
70 - work around perl on windows bugs where perl returns undocumented 621 - work around perl on windows bugs where perl returns undocumented
71 error codes for sysread, syswrite etc. by taking advantage 622 error codes for sysread, syswrite etc. by taking advantage
… …
75 - use unix domain sockets in testsuite to work around 626 - use unix domain sockets in testsuite to work around
76 common perl implementation bugs on widows (they are emulated by 627 common perl implementation bugs on widows (they are emulated by
77 tcp sockets on windows. ugh.) 628 tcp sockets on windows. ugh.)
78 629
794.71 Sat May 24 20:01:27 CEST 2008 6304.71 Sat May 24 20:01:27 CEST 2008
80 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 631 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
81 - take advantage of async name resolution of AnyEvent::Util. 632 - take advantage of async name resolution of AnyEvent::Util.
82 - work around brutal inet_aton override in Coro::LWP. 633 - work around brutal inet_aton override in Coro::LWP.
83 - take advantage of the readyhook in Coro::EV, for smoother 634 - take advantage of the readyhook in Coro::EV, for smoother
84 scheduling. 635 scheduling.
85 636
864.7 Sun May 11 00:32:19 CEST 2008 6374.7 Sun May 11 00:32:19 CEST 2008
87 - completely reworked the Coro manpage. 638 - completely reworked the Coro manpage.
88 - added Coro::AnyEvent, generic event loop integration. 639 - added Coro::AnyEvent, generic event loop integration.
89 - implement cancel, ready and kill commands in Coro::Debug. 640 - implement cancel, ready and kill commands in Coro::Debug.
90 - document find_coro in Coro::Debug. 641 - document find_coro in Coro::Debug.
91 - incompatible change: rename has_stack to has_cctx. 642 - incompatible change: rename has_stack to has_cctx.
92 - Coro::AIO and Coro::BDB no longer force event model detection, 643 - Coro::AIO and Coro::BDB no longer force event model detection,
93 use AnyEvent::AIO and AnyEvent::BDB. 644 use AnyEvent::AIO and AnyEvent::BDB.
… …
217 768
2184.11 Thu Oct 11 02:40:24 CEST 2007 7694.11 Thu Oct 11 02:40:24 CEST 2007
219 - port to threaded perls. 770 - port to threaded perls.
220 771
2214.1 Thu Oct 11 02:38:16 CEST 2007 7724.1 Thu Oct 11 02:38:16 CEST 2007
222 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 773 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
223 be local to each coro (see Coro::State). 774 be local to each coro (see Coro::State).
224 - incompatible change: for very deep reasons, cede and cede_notself 775 - incompatible change: for very deep reasons, cede and cede_notself
225 cannot return anything, so nothing will be returned. 776 cannot return anything, so nothing will be returned.
226 - possibly bring back 5.10 compatibility (untested). 777 - possibly bring back 5.10 compatibility (untested).
227 - work around stupid (and wrong) warning on 5.10 :(. 778 - work around stupid (and wrong) warning on 5.10 :(.
… …
299 but stateful semantics. 850 but stateful semantics.
300 - fixed a lot of typos in Coro.pm (patch submitted by David 851 - fixed a lot of typos in Coro.pm (patch submitted by David
301 Steinbrunner, which applied flawlessly). 852 Steinbrunner, which applied flawlessly).
302 853
3033.6 Sat Apr 14 17:13:31 CEST 2007 8543.6 Sat Apr 14 17:13:31 CEST 2007
304 - added some bugfixes to get eg/myhttpd working again. 855 - added some bugfixes to get eg/myhttpd working again.
305 - added Coro::Storable for often-cede'ing freeze/thaw. 856 - added Coro::Storable for often-cede'ing freeze/thaw.
306 - try to do a clean exit when a coroutine calls exit 857 - try to do a clean exit when a coroutine calls exit
307 (EXPERIMENTAL). 858 (EXPERIMENTAL).
308 - got rid of indirect call through _coro_init. 859 - got rid of indirect call through _coro_init.
309 - updated the partly antique examples in eg/ to 860 - updated the partly antique examples in eg/ to

Diff Legend

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