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

Comparing Coro/Changes (file contents):
Revision 1.372 by root, Wed Sep 24 07:30:12 2008 UTC vs.
Revision 1.622 by root, Tue Nov 19 05:16:01 2013 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO. 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: maybe implement a default message channel, very much like Erlang's 4TODO: channel->maxsize(newsize)?
5 actor model (which is cool in a lot of important aspects (failures!), 5TODO: __GCC_HAVE_DWARF2_CFI_ASM
6 but very lacking in others (higher level ipc)). 6TODO: swap_sv, maybe add scope_swap_sv, andallow to unswap by repeat?
7TODO: swap_sv not undone in pool
7 8
84.748 9 - FATAL: $Coro::IDLE blocked itself will now use Carp::confess
10 to report a full stacktrace, which should help find out
11 where the actual call is.
12
136.33 Mon Nov 18 11:26:27 CET 2013
14 - do not crash when freeing padlists with holes (in 5.18).
15 - tentative SVt_BIND 5.19 port/fix.
16
176.32 Tue Nov 5 15:35:35 CET 2013
18 - use a new algorithm to derive padlists for perl 5.18. The old
19 one could lead to 0-pointer accesses inside perl (reported
20 by Darin McBride).
21
226.31 Thu May 9 07:39:48 CEST 2013
23 - Coro::AIO requests would crash if the thread was ready'd
24 while the request was ongoing.
25
266.29 Wed May 8 02:55:18 CEST 2013
27 - when an on_destroy handler destructs the coro currently being
28 destructed a perl scalar could be accessed after being freed,
29 likely causing a crash.
30
316.28 Wed Mar 6 06:58:02 CET 2013
32 - clean remnants of existing __DIE__ and __WARN__ handlers so
33 they lose their magic and will not cause segfaults later
34 (testcase by Andrey Sagulin).
35 - improved Coro::State documentation a bit.
36 - Coro::Debug::command now flushes the output.
37 - add hack detection code for x32 abi, because the braindead slugs
38 who designed that made it look exactly like x86_64 without
39 providing proper compile time symbols to test for it. as a result,
40 this detection cannot work reliably.
41 - valgrind stack registering was broken.
42 - do not rely on Time::HiRes anymore in Coro::Debug.
43
446.23 Fri Dec 7 23:36:37 CET 2012
45 - use experimental fiber implementation on native windows
46 perls.
47 - use sizeof (void *) as multiplication factor for stack sizes,
48 to accomodate the totally braindamaged microsoft 64 bit "os".
49 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
50 activeperl 5.16 crashes when PerlIO_define_layer is called due
51 to some bug in the perl dll, strawberry perl at least passes
52 the testsuite.
53 - implement Coro::Handle->peeraddr/host/port, for slightly
54 improved compatibility with LWP.
55 - implement 5.17 compatibility by almost blindly applying a
56 good-looking patch by Father Chrysostomos.
57 - move stack management functions into libcoro 3.
58 - libcoro version 3 "released".
59 - support magic values as timed_io_once args.
60 - recommend AnyEvent 7+ or EV 4+, also require EV
61 version 4 or newer for Coro::EV.
62
636.10 Tue Oct 9 01:14:27 CEST 2012
64 - updated ecb.h, it had a typo that caused it to not compile on many
65 big endian systems (reported by many people).
66 - disable memory fences in ecb.h to improve portability.
67
686.09 Sat Oct 6 23:25:02 CEST 2012
69 - Coro::EV I/O watchers were not interruptible by exceptions
70 (Coro::State::throw) (testcase by sten).
71 - ->throw now puts threads into the ready queue, as this seems to
72 be expected by existing code, and code that doesn't cope with spurious
73 wakeups needs fixing anyway.
74 - use fd -1 in mmap.
75 - cast I32 to int in error message printf.
76 - warn about broken so-called "hardened" kernels.
77
786.08 Fri Apr 13 12:05:47 CEST 2012
79 - be more aggressive about exiting like perl does - formerly,
80 exiting from the non-main thread would not execute END blocks.
81
826.07 Fri Nov 11 21:21:48 CET 2011
83 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
84 - update ecb.h.
85
866.06 Mon Aug 8 23:59:48 CEST 2011
87 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
88 backend, which is an order of magnitude slower. unfortunately, cygwins
89 pthread implementation isn't very complete either, so allocate the stack
90 twice just to be sure.
91 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
92 in its ithreaded perl - best recompile cygwin and use the 'w'indows
93 backend for much better performance. also disable ithreads for
94 even better performance...).
95
966.05 Thu Aug 4 21:36:36 CEST 2011
97 - blush, condvar values would not be propagated from send to recv anymore
98 (reported by Chip Salzenberg).
99 - use exponential increase for the readline buffer length in
100 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
101
1026.04 Wed Aug 3 17:15:45 CEST 2011
103 - use even more efficient and more compatible condvars for
104 compatibility to AnyEvent 6.x :)
105 - more inconsequential ecb.h updates.
106
1076.03 Wed Aug 3 11:41:30 CEST 2011
108 - change how Coro patches AnyEvent condvars for compatibility to
109 AnyEvent 6.x.
110 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
111 hasn't been included.
112
1136.02 Wed Jul 13 04:35:19 CEST 2011
114 - "improve portability to Gentoo" - gentoo manages to put perl variables
115 in memory areas that are farther than 2gb apart, which the jit couldn't
116 handle and barfed. now it's just a bit slower on gentoo and similar
117 systems.
118
1196.01 Sun Jul 3 12:31:14 CEST 2011
120 - workarounds are good, but the test for whether pthreads are used
121 was not good. thisone should be better.
122 - check differently whether gcc generates cfi instructions itself.
123
1246.0 Wed Jun 29 19:43:35 CEST 2011
125 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
126 destroyed and cleaned up automatically (e.g. async { schedule }).
127 - implement a JIT compiler for part of the thread switch code,
128 which gives a 50% speed improvement on threaded perls, and
129 about 4% on non-thraeded perls (so threaded perls now finally
130 reach about half the speed of non-threaded perls).
131 - slightly modernise Coro::Intro, add section about rouse functions.
132 - avoid DEFSV and ERRSV, giving another 10% improvement
133 in thread switching.
134 - Coro::State->is_destroyed is now called is_zombie.
135 - implement a Coro->safe_cancel method that might fail, but
136 cancels in a "safer" way if it succeeds.
137 - add preliminary support for DEBUGGING perls.
138 - get rid of two hash-accesses when initialising a new Coro - this
139 speeds up coro creation by almost a factor of two.
140 - croak when a coro that is being cancelled tries to block
141 (e.g. while executing a guard block), instead of crashing or
142 deadlocking.
143 - use a more robust and also faster method to identify Coro::State
144 objects - speeds up everything a bit.
145 - implement Coro->cancel in XS for a 20% speed improvement, and to
146 be able to implement mutual cancellation.
147 - speed up context switches by a percent or two by more efficiently
148 allocating context stack entries.
149 - implement Coro->join and Coro->on_destroy in XS for a speedup and
150 a reduction in memory use.
151 - cancelling a coro while it itself is cancelling another coro is
152 now supported and working, instead of triggering an assertion.
153 - be a bit more crash-resistant when calling (buggy) on_destroy
154 callbacks (best effort).
155 - move on_destroy into the slf_frame, to allow extension slf
156 functions to have destructors.
157 - get rid if coro refcounting - simply crash in other interpreter
158 threads by nulling the pointers on clone.
159 - simplify warn/die hook handling when loading Coro - the convoluted
160 logic seems to be no longer neccessary.
161 - use libecb instead of our own home-grown gcc hacks.
162 - document alternatives to Coro::LWP. Please use them :)
163 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
164 sigaltstack. Really. wine suffers from it, erlang suffers from it,
165 and it's known since at least 2006.
166
1675.372 Wed Feb 23 06:14:30 CET 2011
168 - apparently mingw doesn't provide a working gettimeofday, try to
169 work around that by relying on Time::HiRes (indirectly brought to
170 my attention by Max Maischein).
171 - fix some portability issues when Time::HiRes was used.
172
1735.371 Mon Feb 21 14:36:08 CET 2011
174 - backport to windows process emulation code again.
175
1765.37 Sat Feb 19 07:49:44 CET 2011
177 - add a big "Coro thread life cycle" section to "man Coro".
178 - try a tentative workaround against the breakage that 5.13 has
179 introduced without depreciation period. sigh.
180 - no longer use Time::HiRes if gettimeofday is available, which
181 saves quite a lot of memory.
182
1835.36 Sun Feb 13 05:33:41 CET 2011
184 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
185 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
186 method call.
187 - undocument Coro::Timer::sleep and obsolete whole module.
188 - optimise Coro::Timer::timeout memory and cpu usage.
189 - slightly updated Coro::Intro for recent changes.
190 - do not initialise PL_dirty anymore.
191
1925.25 Thu Nov 11 01:08:39 CET 2010
193 - try a different approach on netbsd - netbsd 5 finally has marginally
194 working pthreads, but still broken ucontext/sigaltstack.
195 - openbsd 4.8 finally got their act together, Coro works out of the box
196 with asm, setjmp and pthreads (no change, just informational).
197
1985.24 Sat Oct 23 11:27:12 CEST 2010
199 - port to the EV 4.0 API.
200 - work around bugs in mingw32, making strawberry perl work
201 out of the box.
202 - correctly modify Coro::AIO function prototypes
203 so that they reflect the "no optional parameters" rule.
204 - "ported" libcoro to C++.
205
2065.23 Mon May 17 18:50:42 CEST 2010
207 - be more resistant to ordering changes when initialising
208 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
209 Waldorf).
210 - document that perl 5.12 deliberately removed support for cloning.
211
2125.22 Wed Apr 14 03:55:35 CEST 2010
213 - correctly return udnef on errors in Coro::Handle::read/write
214 (testcase by Marc Mims).
215 - convert Coro::Util into a "perl compatibility wrapper" - the functions
216 are less useful now, but are drop-in replacements for existing
217 functions, listing better alternatives in the documentation. This also
218 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
219 with Coro::Util::inet_aton.
220 - do not override $Coro::idle unconditionally in Coro.pm, as other
221 modules could have provided their own idle coro already
222 (for exmaple, Coro::AnyEvent).
223 - fix Coro::Util::gethost* functions.
224 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
225
2265.21 Wed Dec 16 07:19:51 CET 2009
227 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
228 together.
229 - add some examples on how to combine other event loops with Coro in
230 Coro::AnyEvent, and how to run it (and not to block). Seems to be
231 the most common source of confusion.
232 - try to catch people naively blocking in an event callback.
233 - work around the perl filehandle bug issue in conjunction with
234 older common::sense (as indirectly pointed out by ZSystem).
235 - clarify the "not from signal handlers" section.
236
2375.2 Sun Oct 4 14:54:24 CEST 2009
238 - Coro::Storable destroyed the prototypes of the functions it wrapped.
239 - export rouse_cb and rouse_wait by default now.
240 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
241 - new method $state->swap_sv.
242 - added section on "windows process emulation" to the manpage,
243 after a not-so-fruitful (nor-friendly) "discussion" with chip
244 salzenberg (discussion implies arguments, but his only arguments
245 were ad-hominems, one wonders why he started it in the first
246 place). I hope this explains it well enough for him to understand,
247 and maybe well enough for others to understand.
248 - use common::sense everywhere now.
249 - idle callbacks are no longer supported, use idle coros instead.
250 - print a thread listing when a deadlock is detected.
251
2525.17 Sat Aug 22 23:09:31 CEST 2009
253 - work around a bug in the perl debugger causing crashes
254 when running under the debugger by marking _pool_handler as nodebug.
255 - speed up Coro::async considerably.
256 - try some hacks to get netbsd to work "more often" - their broken
257 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
258 - convert Coro to AE 5.0 API.
259
2605.162 Tue Jul 28 04:04:03 CEST 2009
261 - perl 5.8.2 is now minimum requirement.
262 - skip t/19_handle.t on broken windows perls.
263
2645.161 Wed Jul 22 04:47:38 CEST 2009
265 - Coro::AnyEvent::poll could have a different prototype when EV was
266 used as backend (analyzed by Tatsuhiko Miyagawa).
267 - Coro::AnyEvent errornously initialised the event loop when loaded,
268 not on demand.
269 - try to workaround rare */t/01_unblock.t failures.
270
2715.16 Tue Jul 21 01:44:37 CEST 2009
272 - Coro::AnyEvent failed to hook into the event loop
273 when no threads had been readied between detecting
274 the event loop and actually running it.
275 - considerably speed up Coro::Select by taking avdantage
276 of AnyEvent > 4.8 and some other optimisations.
277 - implement paragraph readline mode in Coro::Handle
278 (based on patches by Zsbán Ambrus).
279 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
280 and analyzed by Yasuhiro MATSUMOTO).
281 - clarified libcoro license and copyright.
282 - someone stole my EXTRA_META!!!
283 - implement Coro::Select::patch_pp_sselect and it's brother,
284 for hardcode select overriding.
285
2865.151 Mon Jul 6 05:41:57 CEST 2009
287 - backport to windows process emulation code again (patch by
288 Yasuhiro MATSUMOTO).
289 - slightly update Coro::MakeMaker.
290
2915.15 Tue Jun 30 10:28:06 CEST 2009
292 - deprecate Coro::Socket, document how to get ipv6 support via
293 AnyEvent::Socket instead.
294 - implement signal->wait ($cb) interface, similar to semaphores.
295 - work around SvOK not supporting getmagic, so we have to getmagic
296 to test for undef :/ (reported by Matthias Waldorf).
297 - load Coro::AnyEvent in all modules using AnyEvent.
298 - work around perl corrupting our internal data structures,
299 reported by Tokuhiro Matsuno.
300 - enable per-coroutine real and cpu time gathering
301 (Coro::State::enable_times).
302
3035.14 Wed Jun 24 01:37:48 CEST 2009
304 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
305 as this seems to have been a difficult concept (poll, sleep, idle,
306 idle_upto).
307 - add Coro::AnyEvent::readable/writable functions.
308 - clarify Coro::EV/Event/AnyEvent manpages.
309 - free per-thread global scalars in the thread calling ->cancel, to
310 avoid crashes when $_, $@ etc., are magical but some of those
311 had already been freed.
312 - "unexperimentalise" the callback interface for Coro::Semaphore.
313 - speed up ready queue management/context switching by using a linked
314 list instead of an array (~5-10%).
315 - implement "watch" command in Coro::Debug shells.
316 - for fun, implement time-slicing as an example in the manpage.
317 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
318 Coro::Event, use the normal AnyEvent handling instead of dieing
319 (the same is true for Coro::Handle).
320 - properly document Coro::EV::timed_io_once.
321 - avoid unneccessary ->cancel calls in Coro::Handle.
322 - maybe make it work on mingw32 with win32 backend
323 (based on patch by Yasuhiro Matsumoto).
324
3255.132 Fri May 29 09:00:39 CEST 2009
326 - do not keep a reference to the argument itself in
327 Coro::Semaphore::guard, as it could change later.
328 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
329 should support a prepare callback.
330
3315.131 Mon Mar 16 23:20:37 CET 2009
332 - implement and document Coro->suspend, Coro->resume.
333 - fix Coro::Select implementation to not (often) close
334 the passed file descriptors (testcase provided by pippijn).
335
3365.13 Mon Dec 15 21:51:42 CET 2008
337 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
338 - don't set diehook to C<undef> but instead to NULL, to avoid
339 spurious warnings.
340 - fix a lot of bugs in Coro::SemaphoreSet.
341 - Coro::SemaphoreSet will less often create a semaphore needlessly.
342 - add Coro::SemaphoreSet::count and wait methods.
343 - take advantage of the new Guard module.
344 - deprecate Coro::guard.
345 - try to fix the dreaded 01_unblock tests once more. I hate it when
346 testsuites need more fixing than the code they are supposed to test.
347 - croak in more cases when a required callback isn't resolvable.
348 - fix some minor issues in Coro::State->call/eval.
349 - use current coroutine context instead of a temporary one
350 when temporarily switching to another coroutine.
351 - do not call C-level on_destroy handlers during global destruction,
352 to avoid needless segfaults.
353
3545.12 Sun Dec 7 13:30:38 CET 2008
355 - add default config for MirOS, which seems to be a bug-to-bug
356 compatible fork of openbsd ("world domination by releasing
357 openbsd cvs before the openbsd folks do it" or so :).
358 - free_padlist did destroy the names pad, not good, but didn't
359 seem to bother perl - this could fix issues such as eval ""
360 inside a function called from multiple coroutines.
361 - use a different method to detect destruction time.
362 - be more careful when freeing padlists just before global
363 destruction.
364 - fixed and expanded the call/cc example.
365 - renamed _terminate to _coro_run.
366 - new method Coro::Channel->shutdown.
367 - try pthreads on openbsd <4.4 (broken sigaltstack, will
368 pthreads fare better?).
369 - be less picky about destroying "a" running coroutine.
370
3715.11 Tue Nov 25 21:49:05 CET 2008
372 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
373 working code with failed assertions, introducing more bugs than
374 it fixes, requiring elaborate workarounds :(
375
3765.1 Mon Nov 24 08:54:59 CET 2008
377 - wrote a small introductory tutorial - Coro::Intro.
378 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
379 - Coro::Select did errornously dup the file descriptors
380 and didn't work with all AnyEvent backends.
381 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
382 LWP data transfers.
383 - disassociate c contexts from coro objects - this is agruably more
384 correct, but mostly allows sharing of cctxs between coro and state
385 objects, for added memory savings and speed increases.
386 - bumped $Coro::POOL_RSS up to 32kb by default.
387 - no longer set the optype to OP_CUSTOM, as B::* understandably
388 doesn't like this very much (and we *are* a type of entersub).
389 - implement state cloning, just to prove that call/cc can be done.
390 - automatically load Coro::AnyEvent in Coro::Handle.
391 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
392 do not have this method (and don't need it either).
393 - speed up generic anyevent methods in Coro::Handle by using rouse
394 callbacks.
395 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
396 others. It also makes the debugger happier, as you can trace
397 through the idle threads now.
398 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
399 - eg/event was pretty broken.
400 - better 5.8.6 compatibility.
401
4025.0 Thu Nov 20 10:35:05 CET 2008
403 - NEW ARCHITECTURE: use the latest 4.x version if you experience
404 stability issues.
405 - bump API version to 7 - all dependents must be recompiled.
406 - removed timed_* functions - they were not being used anyways
407 and should be replaced by a more generic mechanism -
408 and were annoying to support anyways :)
409 - removed SemaphoreSet's waiter method - use sem method instead.
410 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
411 - async_pool did free a scalar value twice
412 ("Attempt to unreference...").
413 - fix a longstanding bug where calling terminate on a coro that
414 was waiting for a semaphore that was just becoming available
415 would cause a deadlock (semaphore would get into a state where
416 it was available but waiters were still blocked).
417 - calling throw on a coroutine that is waiting for a semaphore will
418 no longer make it acquire the semaphore (and thus leak a count).
419 - perl's process emulation is now not even theoretically supported
420 anymore.
421 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
422 conversion of callback-style to blocking-style.
423 - new methods $coro->schedule_to and ->cede_to, to specifically
424 schedule or cede to a specific coroutine.
425 - new function Coro::Semaphore::wait.
426 - use named constants in Coro::Channel (Richard Hundt).
427 - directly patch the entersub opcode calling SLF functions (cede,
428 transfer and so on). this does speed up context switching, but
429 more importanly, it frees us from the hardcoded behaviour of
430 entersub, so we might actually be able to return something from
431 those functions and atcually create new ones.
432 - take advantage of __builtin_frame_address on gcc.
433 - expose THX in coroapi (not sure whether this was a wise decision,
434 as "threaded" perls are running at half speed anyways).
435 - implement execute_slf (schedule-like-function) interface that makes
436 it possible to implement schedule-like-functions in XS.
437 - use new SLF interface to massively speed up Coro::EV by roughly a
438 factor of two.
439 - used new SLF interface to massively speed up Coro::Semaphore by a
440 factor of three.
441 - used new SLF interface to speed up Coro::AIO by roughly a factor of
442 four and reduce its memory usage considerably.
443 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
444 for a nice speedup and vastly more correct behaviour. Also implement
445 a new method "sem" to get at the underlying semaphore object.
446 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
447 (in comparison) 20-40% speedup.
448 - used new SLF interface to reimplement Coro::Signal gaining
449 some unknown (because I was too lazy), but certain, speedup, and also
450 making signals reliable for the first time.
451 - used new SLF interface and other optimisations to speed up async_pool
452 by a factor of two. It also doesn't rely on perl's exception mechanism
453 to exit anymore. The overhead for terminating an async_pool, coro over
454 a normal async is now very small.
455 - sped up coroutine creation/destruction by 40%.
456 - forgot to include Coro/libcoro/README in the dist for all these years.
457 - work around a freebsd pthreads bug (manual testcancel is required as
458 pthread_cond_wait isn't a cancellation point on freebsd).
459 - use new rouse functions to speed up and simplify Coro::BDB.
460 - make "prefer perl native functions" work with threaded perls.
461 - condense Coro::Debug ps output, hint at v and w flags.
462 - (libcoro) lots of minor cleanups and portability improvements.
463
4644.914 Wed Nov 19 12:54:18 CET 2008
465 - fix a disastrous bug in the readline optimisation
466 introduced in 4.801.
467
4684.913 Sat Nov 15 07:58:28 CET 2008
469 - async_pool did free a scalar value twice
470 ("Attempt to unreference...").
471
4724.912 Thu Nov 13 18:31:23 CET 2008
473 - minor cleanups.
474 - use much larger stacks on linux and perl < 5.8.8.
475 - Coro::Debug::new_unix_server did not unlink the socket
476 when destroyed.
477
4784.911 Tue Nov 11 04:26:17 CET 2008
479 - "port" to threaded perls.
480
4814.91 Mon Nov 10 05:36:38 CET 2008
482 - the ->throw exception object no longer leaks.
483 - creating a new cctx leaked a scopestack entry (memleak).
484 - new coroutines didn't get created with a zero flags field
485 (unknown impact).
486 - calling ->throw on a not-yet-started coroutine should now work
487 instead of being ignored.
488 - ->throw is now supported on Coro::State objects.
489 - clean up cctx creation code a bit.
490 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
491
4924.9 Sat Nov 8 17:45:27 CET 2008
493 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
494 - (libcoro) no longer leak threads in the experimental pthread backend,
495 also speed it up considerably.
496 - (libcoro) do not rely on makecontext passing void *'s unscathed.
497 - fix compiletime dependencies on libcoro in the Makefile.
498 - cctx_count wasn't always updated properly.
499 - Coro::State::cctx_stacksize wasn't applied correctly.
500 - new function Coro::State::cctx_max_idle.
501 - the default max number of idle C contexts is now 4.
502 - (libcoro) try harder to get _setjmp/_longjmp.
503 - (libcoro) cleanup and extend the libcoro API to officially
504 allow the creation of empty source contexts.
505 - very experimental workaround for the totally broken netbsd platform.
506 - tried to hack around openbsd bugs.
507
5084.804 Wed Nov 5 16:36:00 CET 2008
509 - Coro::Debug::new_unix_server would not create a non-blocking listening
510 socket, sometimes causing freezes.
511 - (libcoro) fix misaligned stack points for setjmp and assembly
512 methods, which can cause crashes on x86/x86_64 with a sufficiently
513 aggressive compiler.
514 - new function: Coro::Debug::new_tcp_server.
515 - move ->throw into the Coro class because it only works on coro objects.
516
5174.803 Mon Nov 3 17:16:12 CET 2008
518 - (libcoro) use a global asm statement to become independent of gcc
519 otpimisations for CORO_ASM (thanks to pippijn for the idea).
520 - try to workaround yet another broken bsd, this time dragonfly.
521
5224.802 Thu Oct 30 10:56:12 CET 2008
523 - support -fno-omit-frame-pointer on x86 with the assembly method.
524 - tune 01_unblock.t tests a bit.
525
5264.801 Wed Oct 22 18:33:37 CEST 2008
527 - improve readline speed for very long "lines".
528 - backport to 5.8.8.
529
5304.8 Thu Oct 2 13:34:40 CEST 2008
531 - new function Coro::AIO::aio_wait.
532 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
533 - greatly speed up and reduce memory usage of Coro::AIO requests.
534 - implement some other µ-optimisations.
535
5364.749 Mon Sep 29 14:40:12 CEST 2008
537 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
538
5394.748 Sat Sep 27 14:03:19 CEST 2008
540 - implement, but do not document, PerlIO::cede(granularity).
9 - Coro::Storable forgot to wrap Storable::pstore. 541 - Coro::Storable forgot to wrap Storable::pstore.
542 - work around the multitude of leaks and memory corruption
543 bugs in PerlIO::via by using our own C-level perliol. As a side
544 effect, Coro::Storable is now much, much, much faster.
10 545
114.747 Tue Sep 23 01:59:41 CEST 2008 5464.747 Tue Sep 23 01:59:41 CEST 2008
12 - fix a per-cv memleak (one empty array was leaked per 547 - fix a per-cv memleak (one empty array was leaked per
13 code reference). 548 code reference).
14 - avoid a crash in coro->call|rss when the coroutine was already 549 - avoid a crash in coro->call|rss when the coroutine was already
… …
25 - remove debugging code related to MgPV_nolen_const, also try to 560 - remove debugging code related to MgPV_nolen_const, also try to
26 make it compile with perl 5.8.6 (yes, apple apparently loves 561 make it compile with perl 5.8.6 (yes, apple apparently loves
27 outdated software). Reported by John S. 562 outdated software). Reported by John S.
28 563
294.744 Tue Jul 8 22:06:35 CEST 2008 5644.744 Tue Jul 8 22:06:35 CEST 2008
30 - correctly provide default DIE/WARN handlers as documented. 565 - correctly provide default DIE/WARN handlers as documented.
31 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 566 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
32 implementations inside perl work fine for us. 567 implementations inside perl work fine for us.
33 568
344.743 Mon Jun 16 00:21:57 CEST 2008 5694.743 Mon Jun 16 00:21:57 CEST 2008
35 - when using Coro::EV without running EV::loop it could 570 - when using Coro::EV without running EV::loop it could
… …
57 - sprinkle "no warnings" freely over everything, also suppress 592 - sprinkle "no warnings" freely over everything, also suppress
58 warnings for some other modules. 593 warnings for some other modules.
59 - fix typo in WSAEWOULDBLOCK. 594 - fix typo in WSAEWOULDBLOCK.
60 595
614.72 Sun May 25 05:14:36 CEST 2008 5964.72 Sun May 25 05:14:36 CEST 2008
62 - tweak META.yaml a bit, unfortunately, there is no documented way 597 - tweak META.yaml a bit, unfortunately, there is no documented way
63 to have optional dependencies with CPAN. doh :( 598 to have optional dependencies with CPAN. doh :(
64 - avoid running some tests on windows because they would fail due to 599 - avoid running some tests on windows because they would fail due to
65 perl bug (broken fork, broken pipes...). 600 perl bug (broken fork, broken pipes...).
66 - work around perl on windows bugs where perl returns undocumented 601 - work around perl on windows bugs where perl returns undocumented
67 error codes for sysread, syswrite etc. by taking advantage 602 error codes for sysread, syswrite etc. by taking advantage
… …
71 - use unix domain sockets in testsuite to work around 606 - use unix domain sockets in testsuite to work around
72 common perl implementation bugs on widows (they are emulated by 607 common perl implementation bugs on widows (they are emulated by
73 tcp sockets on windows. ugh.) 608 tcp sockets on windows. ugh.)
74 609
754.71 Sat May 24 20:01:27 CEST 2008 6104.71 Sat May 24 20:01:27 CEST 2008
76 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 611 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
77 - take advantage of async name resolution of AnyEvent::Util. 612 - take advantage of async name resolution of AnyEvent::Util.
78 - work around brutal inet_aton override in Coro::LWP. 613 - work around brutal inet_aton override in Coro::LWP.
79 - take advantage of the readyhook in Coro::EV, for smoother 614 - take advantage of the readyhook in Coro::EV, for smoother
80 scheduling. 615 scheduling.
81 616
824.7 Sun May 11 00:32:19 CEST 2008 6174.7 Sun May 11 00:32:19 CEST 2008
83 - completely reworked the Coro manpage. 618 - completely reworked the Coro manpage.
84 - added Coro::AnyEvent, generic event loop integration. 619 - added Coro::AnyEvent, generic event loop integration.
85 - implement cancel, ready and kill commands in Coro::Debug. 620 - implement cancel, ready and kill commands in Coro::Debug.
86 - document find_coro in Coro::Debug. 621 - document find_coro in Coro::Debug.
87 - incompatible change: rename has_stack to has_cctx. 622 - incompatible change: rename has_stack to has_cctx.
88 - Coro::AIO and Coro::BDB no longer force event model detection, 623 - Coro::AIO and Coro::BDB no longer force event model detection,
89 use AnyEvent::AIO and AnyEvent::BDB. 624 use AnyEvent::AIO and AnyEvent::BDB.
… …
213 748
2144.11 Thu Oct 11 02:40:24 CEST 2007 7494.11 Thu Oct 11 02:40:24 CEST 2007
215 - port to threaded perls. 750 - port to threaded perls.
216 751
2174.1 Thu Oct 11 02:38:16 CEST 2007 7524.1 Thu Oct 11 02:38:16 CEST 2007
218 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 753 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
219 be local to each coro (see Coro::State). 754 be local to each coro (see Coro::State).
220 - incompatible change: for very deep reasons, cede and cede_notself 755 - incompatible change: for very deep reasons, cede and cede_notself
221 cannot return anything, so nothing will be returned. 756 cannot return anything, so nothing will be returned.
222 - possibly bring back 5.10 compatibility (untested). 757 - possibly bring back 5.10 compatibility (untested).
223 - work around stupid (and wrong) warning on 5.10 :(. 758 - work around stupid (and wrong) warning on 5.10 :(.
… …
295 but stateful semantics. 830 but stateful semantics.
296 - fixed a lot of typos in Coro.pm (patch submitted by David 831 - fixed a lot of typos in Coro.pm (patch submitted by David
297 Steinbrunner, which applied flawlessly). 832 Steinbrunner, which applied flawlessly).
298 833
2993.6 Sat Apr 14 17:13:31 CEST 2007 8343.6 Sat Apr 14 17:13:31 CEST 2007
300 - added some bugfixes to get eg/myhttpd working again. 835 - added some bugfixes to get eg/myhttpd working again.
301 - added Coro::Storable for often-cede'ing freeze/thaw. 836 - added Coro::Storable for often-cede'ing freeze/thaw.
302 - try to do a clean exit when a coroutine calls exit 837 - try to do a clean exit when a coroutine calls exit
303 (EXPERIMENTAL). 838 (EXPERIMENTAL).
304 - got rid of indirect call through _coro_init. 839 - got rid of indirect call through _coro_init.
305 - updated the partly antique examples in eg/ to 840 - updated the partly antique examples in eg/ to

Diff Legend

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