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

Comparing Coro/Changes (file contents):
Revision 1.266 by root, Sun Oct 7 13:53:37 2007 UTC vs.
Revision 1.538 by root, Fri May 7 20:28:30 2010 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: localise SIG{__DIE__} and/or __WARN__ 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO. 4TODO: unready_all
5TODO: maybe implement a default message channel, very much like Erlang 5TODO: myhttpd header parsing
6 (which is cool in a lot of important aspects (failures!), 6TODO: channel->maxsize(newsize)?
7 but very lacking in others (higher level ipc)). 7TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
8 8
94.1 9 - be more resistant to ordering changes when initialising
10 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
11 Waldorf).
12 - document that perl 5.12 deliberately removed support for cloning.
13
145.22 Wed Apr 14 03:55:35 CEST 2010
15 - correctly return udnef on errors in Coro::Handle::read/write
16 (testcase by Marc Mims).
17 - convert Coro::Util into a "perl compatibility wrapper" - the functions
18 are less useful now, but are drop-in replacements for existing
19 functions, listing better alternatives in the documentation. This also
20 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
21 with Coro::Util::inet_aton.
22 - do not override $Coro::idle unconditionally in Coro.pm, as other
23 modules could have provided their own idle coro already
24 (for exmaple, Coro::AnyEvent).
25 - fix Coro::Util::gethost* functions.
26 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
27
285.21 Wed Dec 16 07:19:51 CET 2009
29 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
30 together.
31 - add some examples on how to combine other event loops with Coro in
32 Coro::AnyEvent, and how to run it (and not to block). Seems to be
33 the most common source of confusion.
34 - try to catch people naively blocking in an event callback.
35 - work around the perl filehandle bug issue in conjunction with
36 older common::sense (as indirectly pointed out by ZSystem).
37 - clarify the "not from signal handlers" section.
38
395.2 Sun Oct 4 14:54:24 CEST 2009
40 - Coro::Storable destroyed the prototypes of the functions it wrapped.
41 - export rouse_cb and rouse_wait by default now.
42 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
43 - new method $state->swap_sv.
44 - added section on "windows process emulation" to the manpage,
45 after a not-so-fruitful (nor-friendly) "discussion" with chip
46 salzenberg (discussion implies arguments, but his only arguments
47 were ad-hominems, one wonders why he started it in the first
48 place). I hope this explains it well enough for him to understand,
49 and maybe well enough for others to understand.
50 - use common::sense everywhere now.
51 - idle callbacks are no longer supported, use idle coros instead.
52 - print a thread listing when a deadlock is detected.
53
545.17 Sat Aug 22 23:09:31 CEST 2009
55 - work around a bug in the perl debugger causing crashes
56 when running under the debugger by marking _pool_handler as nodebug.
57 - speed up Coro::async considerably.
58 - try some hacks to get netbsd to work "more often" - their broken
59 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
60 - convert Coro to AE 5.0 API.
61
625.162 Tue Jul 28 04:04:03 CEST 2009
63 - perl 5.8.2 is now minimum requirement.
64 - skip t/19_handle.t on broken windows perls.
65
665.161 Wed Jul 22 04:47:38 CEST 2009
67 - Coro::AnyEvent::poll could have a different prototype when EV was
68 used as backend (analyzed by Tatsuhiko Miyagawa).
69 - Coro::AnyEvent errornously initialised the event loop when loaded,
70 not on demand.
71 - try to workaround rare */t/01_unblock.t failures.
72
735.16 Tue Jul 21 01:44:37 CEST 2009
74 - Coro::AnyEvent failed to hook into the event loop
75 when no threads had been readied between detecting
76 the event loop and actually running it.
77 - considerably speed up Coro::Select by taking avdantage
78 of AnyEvent > 4.8 and some other optimisations.
79 - implement paragraph readline mode in Coro::Handle
80 (based on patches by Zsbán Ambrus).
81 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
82 and analyzed by Yasuhiro MATSUMOTO).
83 - clarified libcoro license and copyright.
84 - someone stole my EXTRA_META!!!
85 - implement Coro::Select::patch_pp_sselect and it's brother,
86 for hardcode select overriding.
87
885.151 Mon Jul 6 05:41:57 CEST 2009
89 - backport to windows process emulation code again (patch by
90 Yasuhiro MATSUMOTO).
91 - slightly update Coro::MakeMaker.
92
935.15 Tue Jun 30 10:28:06 CEST 2009
94 - deprecate Coro::Socket, document how to get ipv6 support via
95 AnyEvent::Socket instead.
96 - implement signal->wait ($cb) interface, similar to semaphores.
97 - work around SvOK not supporting getmagic, so we have to getmagic
98 to test for undef :/ (reported by Matthias Waldorf).
99 - load Coro::AnyEvent in all modules using AnyEvent.
100 - work around perl corrupting our internal data structures,
101 reported by Tokuhiro Matsuno.
102 - enable per-coroutine real and cpu time gathering
103 (Coro::State::enable_times).
104
1055.14 Wed Jun 24 01:37:48 CEST 2009
106 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
107 as this seems to have been a difficult concept (poll, sleep, idle,
108 idle_upto).
109 - add Coro::AnyEvent::readable/writable functions.
110 - clarify Coro::EV/Event/AnyEvent manpages.
111 - free per-thread global scalars in the thread calling ->cancel, to
112 avoid crashes when $_, $@ etc., are magical but some of those
113 had already been freed.
114 - "unexperimentalise" the callback interface for Coro::Semaphore.
115 - speed up ready queue management/context switching by using a linked
116 list instead of an array (~5-10%).
117 - implement "watch" command in Coro::Debug shells.
118 - for fun, implement time-slicing as an example in the manpage.
119 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
120 Coro::Event, use the normal AnyEvent handling instead of dieing
121 (the same is true for Coro::Handle).
122 - properly document Coro::EV::timed_io_once.
123 - avoid unneccessary ->cancel calls in Coro::Handle.
124 - maybe make it work on mingw32 with win32 backend
125 (based on patch by Yasuhiro Matsumoto).
126
1275.132 Fri May 29 09:00:39 CEST 2009
128 - do not keep a reference to the argument itself in
129 Coro::Semaphore::guard, as it could change later.
130 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
131 should support a prepare callback.
132
1335.131 Mon Mar 16 23:20:37 CET 2009
134 - implement and document Coro->suspend, Coro->resume.
135 - fix Coro::Select implementation to not (often) close
136 the passed file descriptors (testcase provided by pippijn).
137
1385.13 Mon Dec 15 21:51:42 CET 2008
139 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
140 - don't set diehook to C<undef> but instead to NULL, to avoid
141 spurious warnings.
142 - fix a lot of bugs in Coro::SemaphoreSet.
143 - Coro::SemaphoreSet will less often create a semaphore needlessly.
144 - add Coro::SemaphoreSet::count and wait methods.
145 - take advantage of the new Guard module.
146 - deprecate Coro::guard.
147 - try to fix the dreaded 01_unblock tests once more. I hate it when
148 testsuites need more fixing than the code they are supposed to test.
149 - croak in more cases when a required callback isn't resolvable.
150 - fix some minor issues in Coro::State->call/eval.
151 - use current coroutine context instead of a temporary one
152 when temporarily switching to another coroutine.
153 - do not call C-level on_destroy handlers during global destruction,
154 to avoid needless segfaults.
155
1565.12 Sun Dec 7 13:30:38 CET 2008
157 - add default config for MirOS, which seems to be a bug-to-bug
158 compatible fork of openbsd ("world domination by releasing
159 openbsd cvs before the openbsd folks do it" or so :).
160 - free_padlist did destroy the names pad, not good, but didn't
161 seem to bother perl - this could fix issues such as eval ""
162 inside a function called from multiple coroutines.
163 - use a different method to detect destruction time.
164 - be more careful when freeing padlists just before global
165 destruction.
166 - fixed and expanded the call/cc example.
167 - renamed _terminate to _coro_run.
168 - new method Coro::Channel->shutdown.
169 - try pthreads on openbsd <4.4 (broken sigaltstack, will
170 pthreads fare better?).
171 - be less picky about destroying "a" running coroutine.
172
1735.11 Tue Nov 25 21:49:05 CET 2008
174 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
175 working code with failed assertions, introducing more bugs than
176 it fixes, requiring elaborate workarounds :(
177
1785.1 Mon Nov 24 08:54:59 CET 2008
179 - wrote a small introductory tutorial - Coro::Intro.
180 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
181 - Coro::Select did errornously dup the file descriptors
182 and didn't work with all AnyEvent backends.
183 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
184 LWP data transfers.
185 - disassociate c contexts from coro objects - this is agruably more
186 correct, but mostly allows sharing of cctxs between coro and state
187 objects, for added memory savings and speed increases.
188 - bumped $Coro::POOL_RSS up to 32kb by default.
189 - no longer set the optype to OP_CUSTOM, as B::* understandably
190 doesn't like this very much (and we *are* a type of entersub).
191 - implement state cloning, just to prove that call/cc can be done.
192 - automatically load Coro::AnyEvent in Coro::Handle.
193 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
194 do not have this method (and don't need it either).
195 - speed up generic anyevent methods in Coro::Handle by using rouse
196 callbacks.
197 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
198 others. It also makes the debugger happier, as you can trace
199 through the idle threads now.
200 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
201 - eg/event was pretty broken.
202 - better 5.8.6 compatibility.
203
2045.0 Thu Nov 20 10:35:05 CET 2008
205 - NEW ARCHITECTURE: use the latest 4.x version if you experience
206 stability issues.
207 - bump API version to 7 - all dependents must be recompiled.
208 - removed timed_* functions - they were not being used anyways
209 and should be replaced by a more generic mechanism -
210 and were annoying to support anyways :)
211 - removed SemaphoreSet's waiter method - use sem method instead.
212 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
213 - async_pool did free a scalar value twice
214 ("Attempt to unreference...").
215 - fix a longstanding bug where calling terminate on a coro that
216 was waiting for a semaphore that was just becoming available
217 would cause a deadlock (semaphore would get into a state where
218 it was available but waiters were still blocked).
219 - calling throw on a coroutine that is waiting for a semaphore will
220 no longer make it acquire the semaphore (and thus leak a count).
221 - perl's process emulation is now not even theoretically supported
222 anymore.
223 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
224 conversion of callback-style to blocking-style.
225 - new methods $coro->schedule_to and ->cede_to, to specifically
226 schedule or cede to a specific coroutine.
227 - new function Coro::Semaphore::wait.
228 - use named constants in Coro::Channel (Richard Hundt).
229 - directly patch the entersub opcode calling SLF functions (cede,
230 transfer and so on). this does speed up context switching, but
231 more importanly, it frees us from the hardcoded behaviour of
232 entersub, so we might actually be able to return something from
233 those functions and atcually create new ones.
234 - take advantage of __builtin_frame_address on gcc.
235 - expose THX in coroapi (not sure whether this was a wise decision,
236 as "threaded" perls are running at half speed anyways).
237 - implement execute_slf (schedule-like-function) interface that makes
238 it possible to implement schedule-like-functions in XS.
239 - use new SLF interface to massively speed up Coro::EV by roughly a
240 factor of two.
241 - used new SLF interface to massively speed up Coro::Semaphore by a
242 factor of three.
243 - used new SLF interface to speed up Coro::AIO by roughly a factor of
244 four and reduce its memory usage considerably.
245 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
246 for a nice speedup and vastly more correct behaviour. Also implement
247 a new method "sem" to get at the underlying semaphore object.
248 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
249 (in comparison) 20-40% speedup.
250 - used new SLF interface to reimplement Coro::Signal gaining
251 some unknown (because I was too lazy), but certain, speedup, and also
252 making signals reliable for the first time.
253 - used new SLF interface and other optimisations to speed up async_pool
254 by a factor of two. It also doesn't rely on perl's exception mechanism
255 to exit anymore. The overhead for terminating an async_pool, coro over
256 a normal async is now very small.
257 - sped up coroutine creation/destruction by 40%.
258 - forgot to include Coro/libcoro/README in the dist for all these years.
259 - work around a freebsd pthreads bug (manual testcancel is required as
260 pthread_cond_wait isn't a cancellation point on freebsd).
261 - use new rouse functions to speed up and simplify Coro::BDB.
262 - make "prefer perl native functions" work with threaded perls.
263 - condense Coro::Debug ps output, hint at v and w flags.
264 - (libcoro) lots of minor cleanups and portability improvements.
265
2664.914 Wed Nov 19 12:54:18 CET 2008
267 - fix a disastrous bug in the readline optimisation
268 introduced in 4.801.
269
2704.913 Sat Nov 15 07:58:28 CET 2008
271 - async_pool did free a scalar value twice
272 ("Attempt to unreference...").
273
2744.912 Thu Nov 13 18:31:23 CET 2008
275 - minor cleanups.
276 - use much larger stacks on linux and perl < 5.8.8.
277 - Coro::Debug::new_unix_server did not unlink the socket
278 when destroyed.
279
2804.911 Tue Nov 11 04:26:17 CET 2008
281 - "port" to threaded perls.
282
2834.91 Mon Nov 10 05:36:38 CET 2008
284 - the ->throw exception object no longer leaks.
285 - creating a new cctx leaked a scopestack entry (memleak).
286 - new coroutines didn't get created with a zero flags field
287 (unknown impact).
288 - calling ->throw on a not-yet-started coroutine should now work
289 instead of being ignored.
290 - ->throw is now supported on Coro::State objects.
291 - clean up cctx creation code a bit.
292 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
293
2944.9 Sat Nov 8 17:45:27 CET 2008
295 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
296 - (libcoro) no longer leak threads in the experimental pthread backend,
297 also speed it up considerably.
298 - (libcoro) do not rely on makecontext passing void *'s unscathed.
299 - fix compiletime dependencies on libcoro in the Makefile.
300 - cctx_count wasn't always updated properly.
301 - Coro::State::cctx_stacksize wasn't applied correctly.
302 - new function Coro::State::cctx_max_idle.
303 - the default max number of idle C contexts is now 4.
304 - (libcoro) try harder to get _setjmp/_longjmp.
305 - (libcoro) cleanup and extend the libcoro API to officially
306 allow the creation of empty source contexts.
307 - very experimental workaround for the totally broken netbsd platform.
308 - tried to hack around openbsd bugs.
309
3104.804 Wed Nov 5 16:36:00 CET 2008
311 - Coro::Debug::new_unix_server would not create a non-blocking listening
312 socket, sometimes causing freezes.
313 - (libcoro) fix misaligned stack points for setjmp and assembly
314 methods, which can cause crashes on x86/x86_64 with a sufficiently
315 aggressive compiler.
316 - new function: Coro::Debug::new_tcp_server.
317 - move ->throw into the Coro class because it only works on coro objects.
318
3194.803 Mon Nov 3 17:16:12 CET 2008
320 - (libcoro) use a global asm statement to become independent of gcc
321 otpimisations for CORO_ASM (thanks to pippijn for the idea).
322 - try to workaround yet another broken bsd, this time dragonfly.
323
3244.802 Thu Oct 30 10:56:12 CET 2008
325 - support -fno-omit-frame-pointer on x86 with the assembly method.
326 - tune 01_unblock.t tests a bit.
327
3284.801 Wed Oct 22 18:33:37 CEST 2008
329 - improve readline speed for very long "lines".
330 - backport to 5.8.8.
331
3324.8 Thu Oct 2 13:34:40 CEST 2008
333 - new function Coro::AIO::aio_wait.
334 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
335 - greatly speed up and reduce memory usage of Coro::AIO requests.
336 - implement some other µ-optimisations.
337
3384.749 Mon Sep 29 14:40:12 CEST 2008
339 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
340
3414.748 Sat Sep 27 14:03:19 CEST 2008
342 - implement, but do not document, PerlIO::cede(granularity).
343 - Coro::Storable forgot to wrap Storable::pstore.
344 - work around the multitude of leaks and memory corruption
345 bugs in PerlIO::via by using our own C-level perliol. As a side
346 effect, Coro::Storable is now much, much, much faster.
347
3484.747 Tue Sep 23 01:59:41 CEST 2008
349 - fix a per-cv memleak (one empty array was leaked per
350 code reference).
351 - avoid a crash in coro->call|rss when the coroutine was already
352 destroyed (most noticably when using Coro::Debug::ps :)
353 - also protect *Storable::FILE.
354 - push up default storable granularity to 20ms.
355
3564.746 Sun Sep 21 03:22:20 CEST 2008
357 - be more insistent on locking Storable against reentrancy
358 in Coro::Storable.
359 - move swap_def?v and throw to Coro::State, as documented.
360
3614.745 Thu Jul 24 00:14:38 CEST 2008
362 - remove debugging code related to MgPV_nolen_const, also try to
363 make it compile with perl 5.8.6 (yes, apple apparently loves
364 outdated software). Reported by John S.
365
3664.744 Tue Jul 8 22:06:35 CEST 2008
367 - correctly provide default DIE/WARN handlers as documented.
368 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
369 implementations inside perl work fine for us.
370
3714.743 Mon Jun 16 00:21:57 CEST 2008
372 - when using Coro::EV without running EV::loop it could
373 result in busy-waiting for events, this has been fixed.
374 - reduce codesize and improve performance by using EV_DEFAULT_UC.
375
3764.742 Sat May 31 14:10:21 CEST 2008
377 - implement a workaround for (some) perl <5.8.8 versions.
378 - require EV 3.3+.
379
3804.741 Fri May 30 23:33:09 CEST 2008
381 - tell netbsd how utterly broken their imitation of an OS is
382 and refuse to build by default if pthreads are in use.
383 - switch to "s" method on all bsd's by default, as their ucontext
384 stuff seems just too broken.
385 - fix a bug in Coro::Select.
386
3874.74 Thu May 29 20:05:31 CEST 2008
388 - do not test Coro::LWP for lack of dependencies.
389 - do not test Coro::Select for lack of working perls.
390
3914.73 Thu May 29 2008
392 - fix a bug in Coro::EV which would cause it to block despite
393 there being runnable coroutines.
394 - sprinkle "no warnings" freely over everything, also suppress
395 warnings for some other modules.
396 - fix typo in WSAEWOULDBLOCK.
397
3984.72 Sun May 25 05:14:36 CEST 2008
399 - tweak META.yaml a bit, unfortunately, there is no documented way
400 to have optional dependencies with CPAN. doh :(
401 - avoid running some tests on windows because they would fail due to
402 perl bug (broken fork, broken pipes...).
403 - work around perl on windows bugs where perl returns undocumented
404 error codes for sysread, syswrite etc. by taking advantage
405 of AnyEvent's workaround for that problem.
406 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
407 a common perl implementation bug on windows.
408 - use unix domain sockets in testsuite to work around
409 common perl implementation bugs on widows (they are emulated by
410 tcp sockets on windows. ugh.)
411
4124.71 Sat May 24 20:01:27 CEST 2008
413 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
414 - take advantage of async name resolution of AnyEvent::Util.
415 - work around brutal inet_aton override in Coro::LWP.
416 - take advantage of the readyhook in Coro::EV, for smoother
417 scheduling.
418
4194.7 Sun May 11 00:32:19 CEST 2008
420 - completely reworked the Coro manpage.
421 - added Coro::AnyEvent, generic event loop integration.
422 - implement cancel, ready and kill commands in Coro::Debug.
423 - document find_coro in Coro::Debug.
424 - incompatible change: rename has_stack to has_cctx.
425 - Coro::AIO and Coro::BDB no longer force event model detection,
426 use AnyEvent::AIO and AnyEvent::BDB.
427
4284.6 Sat Apr 26 10:05:14 CEST 2008
429 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
430 by the Coro module. It was a mistake to have it in the first place.
431 - (experimental) support for activestate perl 5.10 (method "w").
432 - (experimental) support for strawberry perl (method "a").
433 - coro_sigelem_set did not return a value although it had to,
434 actual impact unknown.
435
4364.51 Mon Apr 14 13:28:27 CEST 2008
437 - make it compile again on 5.8.
438
4394.50 Thu Apr 10 09:43:17 CEST 2008
440 - I did it twice! (see 4.49).
441
4424.49 Sun Apr 6 21:23:31 CEST 2008
443 - grr, instead of compiling the recent changes on 5.10 only they
444 were compiled on 5.8 only.
445
4464.48 Sun Apr 6 20:36:46 CEST 2008
447 - allow coroutine switches during eval's under 5.10.x, as it seems
448 the parser is a per-interpreter option now, so this is safe
449 (this might fix the odd crashing bug).
450 - drop support for 5.9.x versions: they are dead, jim.
451
4524.47 Sun Apr 6 00:37:52 CEST 2008
453 - force cctx allocation on API calls: we know we need to force one
454 and gcc actually manages to confuse our heuristic nowadays,
455 leading to crashes and worse.
456 - document force_cctx.
457
4584.46 Fri Apr 4 22:05:43 CEST 2008
459 - upgrade libcoro, resulting in pthread-backend (which was only created
460 to fulfill the rules of the programming languages shootout).
461
4624.45 Thu Mar 13 11:55:36 CET 2008
463 - fix a file leaking bug in eg/mhyttpd that would allow
464 downloading of any file (reported by oesi).
465 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
466 (also add testcase).
467 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
468 apparently having loads of fun with that).
469 - implement $state->swap_defsv, ->swap_defav and document ->throw.
470
4714.4 Wed Feb 13 16:44:29 CET 2008
472 - only assume .cfi pseudo isns exist on GNU/Linux.
473 - add get_timed method to Coro::Channels.
474 - fixed Coro::Debug manpage.
475 - perl 5.11 compatbility improvement based on patch by
476 Andreas König.
477
4784.37 Sun Jan 20 11:25:23 CET 2008
479 - soften the check for an active parser for perl 5.10 (avoiding
480 "Coro::State::transfer called while parsing" in many cases).
481
4824.36 Sun Jan 13 10:53:56 CET 2008
483 - reset diehook when terminating from an async_pool as to not
484 trigger any __DIE__ handlers.
485
4864.35 Sun Jan 13 04:14:13 CET 2008
487 - "bt" debug command now displays any exceptions
488 from longmess and also skips the innermost
489 stackframes, giving more useufl output.
490 - allow backtraces in coroutines blocked in Coro::EV,
491 at a <1% speed hit.
492 - handle localising of $SIG{__WARN__} and __DIE__
493 properly (with a proper amount of dirty hacking).
494
4954.34 Sat Dec 22 17:49:53 CET 2007
496 - upgrade to EV version 2.0 API.
497
4984.33 Mon Dec 17 08:36:12 CET 2007
499 - make Coro::AIO etc. loadable in the absence of EV.
500
5014.32 Mon Dec 17 07:46:02 CET 2007
502 - majorly improved Coro::Handle's performance with Coro::EV.
503 - implemented the until now mythical Coro::BDB module.
504 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
505
5064.31 Wed Dec 5 12:32:39 CET 2007
507 - remove warn statement form Coro::Util.
508
5094.3 Tue Dec 4 20:33:14 CET 2007
510 - calls to the idle function could cause stack corruption
511 when the stack changed.
512 - do no longer rely on the presence of EV::DNS if EV is used
513 (because it is gone), but instead take avdantage of EV::ADNS
514 if available.
515 - add ($) prototypes to all functions in Coro::Storable.
516 - use a conventional (safer) idle callback in Coro::EV.
517 - do accept quickly in Coro::Debug to avoid endless loops.
518
5194.22 Fri Nov 30 16:04:04 CET 2007
520 - really use optimised versions for Event and EV in Coro::Util
521 and Coro::Handle.
522
5234.21 Sun Nov 25 10:48:59 CET 2007
524 - fix a spurious memory read.
525 - Coro::EV no longer keeps the eventloop "alive".
526
5274.2 Fri Nov 9 20:47:05 CET 2007
528 - enable/disable tracing from a new coroutine, not a pooled one.
529 - fix a memleak in Coro::Event.
530 - removed killall call from fork_eval.
531 - made sure store_fd is already loaded so that fork_eval
532 does not have to parse autoload in each subprocess.
533 - only use assembly method if -O switch is in $Config{optimize}.
534 - add (optional) Coro::EV module, so far the best event loop module
535 directly supported by Coro.
536 - if the event model is EV, use EV::DNS to resolve
537 stuff in Coro::Util.
538 - don't get confused by multiple event notifications in Coro::Handle.
539 - initial support for EV (libevent interface).
540 - require Event and EV using configure_requires, to force their existance.
541
5424.13 Wed Oct 24 07:26:45 CEST 2007
543 - add Coro::Storable::blocking_thaw.
544 - use a vastly more complicated technique to localise
545 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
546 - use a coroutine for the idle callback Coro::Event,
547 instead of running Event in the current coroutine context.
548 This also catches recursive invocations.
549 - actually report fork errors in gethostbyname and inet_aton.
550
5514.11 Thu Oct 11 02:40:24 CEST 2007
552 - port to threaded perls.
553
5544.1 Thu Oct 11 02:38:16 CEST 2007
10 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 555 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
11 be local to each coro (see Coro::State). 556 be local to each coro (see Coro::State).
557 - incompatible change: for very deep reasons, cede and cede_notself
558 cannot return anything, so nothing will be returned.
12 - possibly bring back 5.10 compatibility (untested). 559 - possibly bring back 5.10 compatibility (untested).
13 - work around stupid (and wrong) warning on 5.10 :(. 560 - work around stupid (and wrong) warning on 5.10 :(.
14 - overlay the saved state over the context stack. This saves 561 - overlay the saved state over the context stack. This saves
15 a few hundred bytes per coroutine on average and also 562 a few hundred bytes per coroutine on average and also
16 speeds up context switching a bit. 563 speeds up context switching a bit.
17 - further tune default stack sizes. 564 - further tune default stack sizes.
18 - (more) correctly calculate stack usage in coro_rss. 565 - (more) correctly calculate stack usage in coro_rss.
566 - Coro::Storable::blocking_* did not properly lock
567 resulting in races between coroutines.
568 - added Coro::Storable::guard.
569 - stopping to trace a coroutine could destroy the cctx of
570 an unrelated coroutine.
571 - explain the relationship between Perl and C coroutines in
572 more detail in Coro::State.
573 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
574 causing a fork per resolve. This also affected Coro::Socket.
575 - switch to a separate stack in $coro->call/eval to avoid
576 invalidating pointers.
19 577
204.03 Sat Oct 6 21:24:00 CEST 2007 5784.03 Sat Oct 6 21:24:00 CEST 2007
21 - added Coro::throw method. 579 - added Coro::throw method.
22 - minor code cleanups. 580 - minor code cleanups.
23 581

Diff Legend

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