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

Comparing Coro/Changes (file contents):
Revision 1.161 by root, Mon Nov 27 02:08:55 2006 UTC vs.
Revision 1.553 by root, Wed Feb 23 07:14:20 2011 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
33.0 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: channel->maxsize(newsize)?
5TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
6
75.372 Wed Feb 23 06:14:30 CET 2011
8 - apparently mingw doesn't provide a working gettimeofday, try to
9 work around that by relying on Time::HiRes (indirectly brought to
10 my attention by Max Maischein).
11 - fix some portability issues when Time::HiRes was used.
12
135.371 Mon Feb 21 14:36:08 CET 2011
14 - backport to windows process emulation code again.
15
165.37 Sat Feb 19 07:49:44 CET 2011
17 - add a big "Coro thread life cycle" section to "man Coro".
18 - try a tentative workaround against the breakage that 5.13 has
19 introduced without depreciation period. sigh.
20 - no longer use Time::HiRes if gettimeofday is available, which
21 saves quite a lot of memory.
22
235.36 Sun Feb 13 05:33:41 CET 2011
24 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
25 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
26 method call.
27 - undocument Coro::Timer::sleep and obsolete whole module.
28 - optimise Coro::Timer::timeout memory and cpu usage.
29 - slightly updated Coro::Intro for recent changes.
30 - do not initialise PL_dirty anymore.
31
325.25 Thu Nov 11 01:08:39 CET 2010
33 - try a different approach on netbsd - netbsd 5 finally has marginally
34 working pthreads, but still broken ucontext/sigaltstack.
35 - openbsd 4.8 finally got their act together, Coro works out of the box
36 with asm, setjmp and pthreads (no change, just informational).
37
385.24 Sat Oct 23 11:27:12 CEST 2010
39 - port to the EV 4.0 API.
40 - work around bugs in mingw32, making strawberry perl work
41 out of the box.
42 - correctly modify Coro::AIO function prototypes
43 so that they reflect the "no optional parameters" rule.
44 - "ported" libcoro to C++.
45
465.23 Mon May 17 18:50:42 CEST 2010
47 - be more resistant to ordering changes when initialising
48 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
49 Waldorf).
50 - document that perl 5.12 deliberately removed support for cloning.
51
525.22 Wed Apr 14 03:55:35 CEST 2010
53 - correctly return udnef on errors in Coro::Handle::read/write
54 (testcase by Marc Mims).
55 - convert Coro::Util into a "perl compatibility wrapper" - the functions
56 are less useful now, but are drop-in replacements for existing
57 functions, listing better alternatives in the documentation. This also
58 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
59 with Coro::Util::inet_aton.
60 - do not override $Coro::idle unconditionally in Coro.pm, as other
61 modules could have provided their own idle coro already
62 (for exmaple, Coro::AnyEvent).
63 - fix Coro::Util::gethost* functions.
64 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
65
665.21 Wed Dec 16 07:19:51 CET 2009
67 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
68 together.
69 - add some examples on how to combine other event loops with Coro in
70 Coro::AnyEvent, and how to run it (and not to block). Seems to be
71 the most common source of confusion.
72 - try to catch people naively blocking in an event callback.
73 - work around the perl filehandle bug issue in conjunction with
74 older common::sense (as indirectly pointed out by ZSystem).
75 - clarify the "not from signal handlers" section.
76
775.2 Sun Oct 4 14:54:24 CEST 2009
78 - Coro::Storable destroyed the prototypes of the functions it wrapped.
79 - export rouse_cb and rouse_wait by default now.
80 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
81 - new method $state->swap_sv.
82 - added section on "windows process emulation" to the manpage,
83 after a not-so-fruitful (nor-friendly) "discussion" with chip
84 salzenberg (discussion implies arguments, but his only arguments
85 were ad-hominems, one wonders why he started it in the first
86 place). I hope this explains it well enough for him to understand,
87 and maybe well enough for others to understand.
88 - use common::sense everywhere now.
89 - idle callbacks are no longer supported, use idle coros instead.
90 - print a thread listing when a deadlock is detected.
91
925.17 Sat Aug 22 23:09:31 CEST 2009
93 - work around a bug in the perl debugger causing crashes
94 when running under the debugger by marking _pool_handler as nodebug.
95 - speed up Coro::async considerably.
96 - try some hacks to get netbsd to work "more often" - their broken
97 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
98 - convert Coro to AE 5.0 API.
99
1005.162 Tue Jul 28 04:04:03 CEST 2009
101 - perl 5.8.2 is now minimum requirement.
102 - skip t/19_handle.t on broken windows perls.
103
1045.161 Wed Jul 22 04:47:38 CEST 2009
105 - Coro::AnyEvent::poll could have a different prototype when EV was
106 used as backend (analyzed by Tatsuhiko Miyagawa).
107 - Coro::AnyEvent errornously initialised the event loop when loaded,
108 not on demand.
109 - try to workaround rare */t/01_unblock.t failures.
110
1115.16 Tue Jul 21 01:44:37 CEST 2009
112 - Coro::AnyEvent failed to hook into the event loop
113 when no threads had been readied between detecting
114 the event loop and actually running it.
115 - considerably speed up Coro::Select by taking avdantage
116 of AnyEvent > 4.8 and some other optimisations.
117 - implement paragraph readline mode in Coro::Handle
118 (based on patches by Zsbán Ambrus).
119 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
120 and analyzed by Yasuhiro MATSUMOTO).
121 - clarified libcoro license and copyright.
122 - someone stole my EXTRA_META!!!
123 - implement Coro::Select::patch_pp_sselect and it's brother,
124 for hardcode select overriding.
125
1265.151 Mon Jul 6 05:41:57 CEST 2009
127 - backport to windows process emulation code again (patch by
128 Yasuhiro MATSUMOTO).
129 - slightly update Coro::MakeMaker.
130
1315.15 Tue Jun 30 10:28:06 CEST 2009
132 - deprecate Coro::Socket, document how to get ipv6 support via
133 AnyEvent::Socket instead.
134 - implement signal->wait ($cb) interface, similar to semaphores.
135 - work around SvOK not supporting getmagic, so we have to getmagic
136 to test for undef :/ (reported by Matthias Waldorf).
137 - load Coro::AnyEvent in all modules using AnyEvent.
138 - work around perl corrupting our internal data structures,
139 reported by Tokuhiro Matsuno.
140 - enable per-coroutine real and cpu time gathering
141 (Coro::State::enable_times).
142
1435.14 Wed Jun 24 01:37:48 CEST 2009
144 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
145 as this seems to have been a difficult concept (poll, sleep, idle,
146 idle_upto).
147 - add Coro::AnyEvent::readable/writable functions.
148 - clarify Coro::EV/Event/AnyEvent manpages.
149 - free per-thread global scalars in the thread calling ->cancel, to
150 avoid crashes when $_, $@ etc., are magical but some of those
151 had already been freed.
152 - "unexperimentalise" the callback interface for Coro::Semaphore.
153 - speed up ready queue management/context switching by using a linked
154 list instead of an array (~5-10%).
155 - implement "watch" command in Coro::Debug shells.
156 - for fun, implement time-slicing as an example in the manpage.
157 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
158 Coro::Event, use the normal AnyEvent handling instead of dieing
159 (the same is true for Coro::Handle).
160 - properly document Coro::EV::timed_io_once.
161 - avoid unneccessary ->cancel calls in Coro::Handle.
162 - maybe make it work on mingw32 with win32 backend
163 (based on patch by Yasuhiro Matsumoto).
164
1655.132 Fri May 29 09:00:39 CEST 2009
166 - do not keep a reference to the argument itself in
167 Coro::Semaphore::guard, as it could change later.
168 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
169 should support a prepare callback.
170
1715.131 Mon Mar 16 23:20:37 CET 2009
172 - implement and document Coro->suspend, Coro->resume.
173 - fix Coro::Select implementation to not (often) close
174 the passed file descriptors (testcase provided by pippijn).
175
1765.13 Mon Dec 15 21:51:42 CET 2008
177 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
178 - don't set diehook to C<undef> but instead to NULL, to avoid
179 spurious warnings.
180 - fix a lot of bugs in Coro::SemaphoreSet.
181 - Coro::SemaphoreSet will less often create a semaphore needlessly.
182 - add Coro::SemaphoreSet::count and wait methods.
183 - take advantage of the new Guard module.
184 - deprecate Coro::guard.
185 - try to fix the dreaded 01_unblock tests once more. I hate it when
186 testsuites need more fixing than the code they are supposed to test.
187 - croak in more cases when a required callback isn't resolvable.
188 - fix some minor issues in Coro::State->call/eval.
189 - use current coroutine context instead of a temporary one
190 when temporarily switching to another coroutine.
191 - do not call C-level on_destroy handlers during global destruction,
192 to avoid needless segfaults.
193
1945.12 Sun Dec 7 13:30:38 CET 2008
195 - add default config for MirOS, which seems to be a bug-to-bug
196 compatible fork of openbsd ("world domination by releasing
197 openbsd cvs before the openbsd folks do it" or so :).
198 - free_padlist did destroy the names pad, not good, but didn't
199 seem to bother perl - this could fix issues such as eval ""
200 inside a function called from multiple coroutines.
201 - use a different method to detect destruction time.
202 - be more careful when freeing padlists just before global
203 destruction.
204 - fixed and expanded the call/cc example.
205 - renamed _terminate to _coro_run.
206 - new method Coro::Channel->shutdown.
207 - try pthreads on openbsd <4.4 (broken sigaltstack, will
208 pthreads fare better?).
209 - be less picky about destroying "a" running coroutine.
210
2115.11 Tue Nov 25 21:49:05 CET 2008
212 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
213 working code with failed assertions, introducing more bugs than
214 it fixes, requiring elaborate workarounds :(
215
2165.1 Mon Nov 24 08:54:59 CET 2008
217 - wrote a small introductory tutorial - Coro::Intro.
218 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
219 - Coro::Select did errornously dup the file descriptors
220 and didn't work with all AnyEvent backends.
221 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
222 LWP data transfers.
223 - disassociate c contexts from coro objects - this is agruably more
224 correct, but mostly allows sharing of cctxs between coro and state
225 objects, for added memory savings and speed increases.
226 - bumped $Coro::POOL_RSS up to 32kb by default.
227 - no longer set the optype to OP_CUSTOM, as B::* understandably
228 doesn't like this very much (and we *are* a type of entersub).
229 - implement state cloning, just to prove that call/cc can be done.
230 - automatically load Coro::AnyEvent in Coro::Handle.
231 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
232 do not have this method (and don't need it either).
233 - speed up generic anyevent methods in Coro::Handle by using rouse
234 callbacks.
235 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
236 others. It also makes the debugger happier, as you can trace
237 through the idle threads now.
238 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
239 - eg/event was pretty broken.
240 - better 5.8.6 compatibility.
241
2425.0 Thu Nov 20 10:35:05 CET 2008
243 - NEW ARCHITECTURE: use the latest 4.x version if you experience
244 stability issues.
245 - bump API version to 7 - all dependents must be recompiled.
246 - removed timed_* functions - they were not being used anyways
247 and should be replaced by a more generic mechanism -
248 and were annoying to support anyways :)
249 - removed SemaphoreSet's waiter method - use sem method instead.
250 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
251 - async_pool did free a scalar value twice
252 ("Attempt to unreference...").
253 - fix a longstanding bug where calling terminate on a coro that
254 was waiting for a semaphore that was just becoming available
255 would cause a deadlock (semaphore would get into a state where
256 it was available but waiters were still blocked).
257 - calling throw on a coroutine that is waiting for a semaphore will
258 no longer make it acquire the semaphore (and thus leak a count).
259 - perl's process emulation is now not even theoretically supported
260 anymore.
261 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
262 conversion of callback-style to blocking-style.
263 - new methods $coro->schedule_to and ->cede_to, to specifically
264 schedule or cede to a specific coroutine.
265 - new function Coro::Semaphore::wait.
266 - use named constants in Coro::Channel (Richard Hundt).
267 - directly patch the entersub opcode calling SLF functions (cede,
268 transfer and so on). this does speed up context switching, but
269 more importanly, it frees us from the hardcoded behaviour of
270 entersub, so we might actually be able to return something from
271 those functions and atcually create new ones.
272 - take advantage of __builtin_frame_address on gcc.
273 - expose THX in coroapi (not sure whether this was a wise decision,
274 as "threaded" perls are running at half speed anyways).
275 - implement execute_slf (schedule-like-function) interface that makes
276 it possible to implement schedule-like-functions in XS.
277 - use new SLF interface to massively speed up Coro::EV by roughly a
278 factor of two.
279 - used new SLF interface to massively speed up Coro::Semaphore by a
280 factor of three.
281 - used new SLF interface to speed up Coro::AIO by roughly a factor of
282 four and reduce its memory usage considerably.
283 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
284 for a nice speedup and vastly more correct behaviour. Also implement
285 a new method "sem" to get at the underlying semaphore object.
286 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
287 (in comparison) 20-40% speedup.
288 - used new SLF interface to reimplement Coro::Signal gaining
289 some unknown (because I was too lazy), but certain, speedup, and also
290 making signals reliable for the first time.
291 - used new SLF interface and other optimisations to speed up async_pool
292 by a factor of two. It also doesn't rely on perl's exception mechanism
293 to exit anymore. The overhead for terminating an async_pool, coro over
294 a normal async is now very small.
295 - sped up coroutine creation/destruction by 40%.
296 - forgot to include Coro/libcoro/README in the dist for all these years.
297 - work around a freebsd pthreads bug (manual testcancel is required as
298 pthread_cond_wait isn't a cancellation point on freebsd).
299 - use new rouse functions to speed up and simplify Coro::BDB.
300 - make "prefer perl native functions" work with threaded perls.
301 - condense Coro::Debug ps output, hint at v and w flags.
302 - (libcoro) lots of minor cleanups and portability improvements.
303
3044.914 Wed Nov 19 12:54:18 CET 2008
305 - fix a disastrous bug in the readline optimisation
306 introduced in 4.801.
307
3084.913 Sat Nov 15 07:58:28 CET 2008
309 - async_pool did free a scalar value twice
310 ("Attempt to unreference...").
311
3124.912 Thu Nov 13 18:31:23 CET 2008
313 - minor cleanups.
314 - use much larger stacks on linux and perl < 5.8.8.
315 - Coro::Debug::new_unix_server did not unlink the socket
316 when destroyed.
317
3184.911 Tue Nov 11 04:26:17 CET 2008
319 - "port" to threaded perls.
320
3214.91 Mon Nov 10 05:36:38 CET 2008
322 - the ->throw exception object no longer leaks.
323 - creating a new cctx leaked a scopestack entry (memleak).
324 - new coroutines didn't get created with a zero flags field
325 (unknown impact).
326 - calling ->throw on a not-yet-started coroutine should now work
327 instead of being ignored.
328 - ->throw is now supported on Coro::State objects.
329 - clean up cctx creation code a bit.
330 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
331
3324.9 Sat Nov 8 17:45:27 CET 2008
333 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
334 - (libcoro) no longer leak threads in the experimental pthread backend,
335 also speed it up considerably.
336 - (libcoro) do not rely on makecontext passing void *'s unscathed.
337 - fix compiletime dependencies on libcoro in the Makefile.
338 - cctx_count wasn't always updated properly.
339 - Coro::State::cctx_stacksize wasn't applied correctly.
340 - new function Coro::State::cctx_max_idle.
341 - the default max number of idle C contexts is now 4.
342 - (libcoro) try harder to get _setjmp/_longjmp.
343 - (libcoro) cleanup and extend the libcoro API to officially
344 allow the creation of empty source contexts.
345 - very experimental workaround for the totally broken netbsd platform.
346 - tried to hack around openbsd bugs.
347
3484.804 Wed Nov 5 16:36:00 CET 2008
349 - Coro::Debug::new_unix_server would not create a non-blocking listening
350 socket, sometimes causing freezes.
351 - (libcoro) fix misaligned stack points for setjmp and assembly
352 methods, which can cause crashes on x86/x86_64 with a sufficiently
353 aggressive compiler.
354 - new function: Coro::Debug::new_tcp_server.
355 - move ->throw into the Coro class because it only works on coro objects.
356
3574.803 Mon Nov 3 17:16:12 CET 2008
358 - (libcoro) use a global asm statement to become independent of gcc
359 otpimisations for CORO_ASM (thanks to pippijn for the idea).
360 - try to workaround yet another broken bsd, this time dragonfly.
361
3624.802 Thu Oct 30 10:56:12 CET 2008
363 - support -fno-omit-frame-pointer on x86 with the assembly method.
364 - tune 01_unblock.t tests a bit.
365
3664.801 Wed Oct 22 18:33:37 CEST 2008
367 - improve readline speed for very long "lines".
368 - backport to 5.8.8.
369
3704.8 Thu Oct 2 13:34:40 CEST 2008
371 - new function Coro::AIO::aio_wait.
372 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
373 - greatly speed up and reduce memory usage of Coro::AIO requests.
374 - implement some other µ-optimisations.
375
3764.749 Mon Sep 29 14:40:12 CEST 2008
377 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
378
3794.748 Sat Sep 27 14:03:19 CEST 2008
380 - implement, but do not document, PerlIO::cede(granularity).
381 - Coro::Storable forgot to wrap Storable::pstore.
382 - work around the multitude of leaks and memory corruption
383 bugs in PerlIO::via by using our own C-level perliol. As a side
384 effect, Coro::Storable is now much, much, much faster.
385
3864.747 Tue Sep 23 01:59:41 CEST 2008
387 - fix a per-cv memleak (one empty array was leaked per
388 code reference).
389 - avoid a crash in coro->call|rss when the coroutine was already
390 destroyed (most noticably when using Coro::Debug::ps :)
391 - also protect *Storable::FILE.
392 - push up default storable granularity to 20ms.
393
3944.746 Sun Sep 21 03:22:20 CEST 2008
395 - be more insistent on locking Storable against reentrancy
396 in Coro::Storable.
397 - move swap_def?v and throw to Coro::State, as documented.
398
3994.745 Thu Jul 24 00:14:38 CEST 2008
400 - remove debugging code related to MgPV_nolen_const, also try to
401 make it compile with perl 5.8.6 (yes, apple apparently loves
402 outdated software). Reported by John S.
403
4044.744 Tue Jul 8 22:06:35 CEST 2008
405 - correctly provide default DIE/WARN handlers as documented.
406 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
407 implementations inside perl work fine for us.
408
4094.743 Mon Jun 16 00:21:57 CEST 2008
410 - when using Coro::EV without running EV::loop it could
411 result in busy-waiting for events, this has been fixed.
412 - reduce codesize and improve performance by using EV_DEFAULT_UC.
413
4144.742 Sat May 31 14:10:21 CEST 2008
415 - implement a workaround for (some) perl <5.8.8 versions.
416 - require EV 3.3+.
417
4184.741 Fri May 30 23:33:09 CEST 2008
419 - tell netbsd how utterly broken their imitation of an OS is
420 and refuse to build by default if pthreads are in use.
421 - switch to "s" method on all bsd's by default, as their ucontext
422 stuff seems just too broken.
423 - fix a bug in Coro::Select.
424
4254.74 Thu May 29 20:05:31 CEST 2008
426 - do not test Coro::LWP for lack of dependencies.
427 - do not test Coro::Select for lack of working perls.
428
4294.73 Thu May 29 2008
430 - fix a bug in Coro::EV which would cause it to block despite
431 there being runnable coroutines.
432 - sprinkle "no warnings" freely over everything, also suppress
433 warnings for some other modules.
434 - fix typo in WSAEWOULDBLOCK.
435
4364.72 Sun May 25 05:14:36 CEST 2008
437 - tweak META.yaml a bit, unfortunately, there is no documented way
438 to have optional dependencies with CPAN. doh :(
439 - avoid running some tests on windows because they would fail due to
440 perl bug (broken fork, broken pipes...).
441 - work around perl on windows bugs where perl returns undocumented
442 error codes for sysread, syswrite etc. by taking advantage
443 of AnyEvent's workaround for that problem.
444 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
445 a common perl implementation bug on windows.
446 - use unix domain sockets in testsuite to work around
447 common perl implementation bugs on widows (they are emulated by
448 tcp sockets on windows. ugh.)
449
4504.71 Sat May 24 20:01:27 CEST 2008
451 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
452 - take advantage of async name resolution of AnyEvent::Util.
453 - work around brutal inet_aton override in Coro::LWP.
454 - take advantage of the readyhook in Coro::EV, for smoother
455 scheduling.
456
4574.7 Sun May 11 00:32:19 CEST 2008
458 - completely reworked the Coro manpage.
459 - added Coro::AnyEvent, generic event loop integration.
460 - implement cancel, ready and kill commands in Coro::Debug.
461 - document find_coro in Coro::Debug.
462 - incompatible change: rename has_stack to has_cctx.
463 - Coro::AIO and Coro::BDB no longer force event model detection,
464 use AnyEvent::AIO and AnyEvent::BDB.
465
4664.6 Sat Apr 26 10:05:14 CEST 2008
467 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
468 by the Coro module. It was a mistake to have it in the first place.
469 - (experimental) support for activestate perl 5.10 (method "w").
470 - (experimental) support for strawberry perl (method "a").
471 - coro_sigelem_set did not return a value although it had to,
472 actual impact unknown.
473
4744.51 Mon Apr 14 13:28:27 CEST 2008
475 - make it compile again on 5.8.
476
4774.50 Thu Apr 10 09:43:17 CEST 2008
478 - I did it twice! (see 4.49).
479
4804.49 Sun Apr 6 21:23:31 CEST 2008
481 - grr, instead of compiling the recent changes on 5.10 only they
482 were compiled on 5.8 only.
483
4844.48 Sun Apr 6 20:36:46 CEST 2008
485 - allow coroutine switches during eval's under 5.10.x, as it seems
486 the parser is a per-interpreter option now, so this is safe
487 (this might fix the odd crashing bug).
488 - drop support for 5.9.x versions: they are dead, jim.
489
4904.47 Sun Apr 6 00:37:52 CEST 2008
491 - force cctx allocation on API calls: we know we need to force one
492 and gcc actually manages to confuse our heuristic nowadays,
493 leading to crashes and worse.
494 - document force_cctx.
495
4964.46 Fri Apr 4 22:05:43 CEST 2008
497 - upgrade libcoro, resulting in pthread-backend (which was only created
498 to fulfill the rules of the programming languages shootout).
499
5004.45 Thu Mar 13 11:55:36 CET 2008
501 - fix a file leaking bug in eg/mhyttpd that would allow
502 downloading of any file (reported by oesi).
503 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
504 (also add testcase).
505 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
506 apparently having loads of fun with that).
507 - implement $state->swap_defsv, ->swap_defav and document ->throw.
508
5094.4 Wed Feb 13 16:44:29 CET 2008
510 - only assume .cfi pseudo isns exist on GNU/Linux.
511 - add get_timed method to Coro::Channels.
512 - fixed Coro::Debug manpage.
513 - perl 5.11 compatbility improvement based on patch by
514 Andreas König.
515
5164.37 Sun Jan 20 11:25:23 CET 2008
517 - soften the check for an active parser for perl 5.10 (avoiding
518 "Coro::State::transfer called while parsing" in many cases).
519
5204.36 Sun Jan 13 10:53:56 CET 2008
521 - reset diehook when terminating from an async_pool as to not
522 trigger any __DIE__ handlers.
523
5244.35 Sun Jan 13 04:14:13 CET 2008
525 - "bt" debug command now displays any exceptions
526 from longmess and also skips the innermost
527 stackframes, giving more useufl output.
528 - allow backtraces in coroutines blocked in Coro::EV,
529 at a <1% speed hit.
530 - handle localising of $SIG{__WARN__} and __DIE__
531 properly (with a proper amount of dirty hacking).
532
5334.34 Sat Dec 22 17:49:53 CET 2007
534 - upgrade to EV version 2.0 API.
535
5364.33 Mon Dec 17 08:36:12 CET 2007
537 - make Coro::AIO etc. loadable in the absence of EV.
538
5394.32 Mon Dec 17 07:46:02 CET 2007
540 - majorly improved Coro::Handle's performance with Coro::EV.
541 - implemented the until now mythical Coro::BDB module.
542 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
543
5444.31 Wed Dec 5 12:32:39 CET 2007
545 - remove warn statement form Coro::Util.
546
5474.3 Tue Dec 4 20:33:14 CET 2007
548 - calls to the idle function could cause stack corruption
549 when the stack changed.
550 - do no longer rely on the presence of EV::DNS if EV is used
551 (because it is gone), but instead take avdantage of EV::ADNS
552 if available.
553 - add ($) prototypes to all functions in Coro::Storable.
554 - use a conventional (safer) idle callback in Coro::EV.
555 - do accept quickly in Coro::Debug to avoid endless loops.
556
5574.22 Fri Nov 30 16:04:04 CET 2007
558 - really use optimised versions for Event and EV in Coro::Util
559 and Coro::Handle.
560
5614.21 Sun Nov 25 10:48:59 CET 2007
562 - fix a spurious memory read.
563 - Coro::EV no longer keeps the eventloop "alive".
564
5654.2 Fri Nov 9 20:47:05 CET 2007
566 - enable/disable tracing from a new coroutine, not a pooled one.
567 - fix a memleak in Coro::Event.
568 - removed killall call from fork_eval.
569 - made sure store_fd is already loaded so that fork_eval
570 does not have to parse autoload in each subprocess.
571 - only use assembly method if -O switch is in $Config{optimize}.
572 - add (optional) Coro::EV module, so far the best event loop module
573 directly supported by Coro.
574 - if the event model is EV, use EV::DNS to resolve
575 stuff in Coro::Util.
576 - don't get confused by multiple event notifications in Coro::Handle.
577 - initial support for EV (libevent interface).
578 - require Event and EV using configure_requires, to force their existance.
579
5804.13 Wed Oct 24 07:26:45 CEST 2007
581 - add Coro::Storable::blocking_thaw.
582 - use a vastly more complicated technique to localise
583 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
584 - use a coroutine for the idle callback Coro::Event,
585 instead of running Event in the current coroutine context.
586 This also catches recursive invocations.
587 - actually report fork errors in gethostbyname and inet_aton.
588
5894.11 Thu Oct 11 02:40:24 CEST 2007
590 - port to threaded perls.
591
5924.1 Thu Oct 11 02:38:16 CEST 2007
593 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
594 be local to each coro (see Coro::State).
595 - incompatible change: for very deep reasons, cede and cede_notself
596 cannot return anything, so nothing will be returned.
597 - possibly bring back 5.10 compatibility (untested).
598 - work around stupid (and wrong) warning on 5.10 :(.
599 - overlay the saved state over the context stack. This saves
600 a few hundred bytes per coroutine on average and also
601 speeds up context switching a bit.
602 - further tune default stack sizes.
603 - (more) correctly calculate stack usage in coro_rss.
604 - Coro::Storable::blocking_* did not properly lock
605 resulting in races between coroutines.
606 - added Coro::Storable::guard.
607 - stopping to trace a coroutine could destroy the cctx of
608 an unrelated coroutine.
609 - explain the relationship between Perl and C coroutines in
610 more detail in Coro::State.
611 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
612 causing a fork per resolve. This also affected Coro::Socket.
613 - switch to a separate stack in $coro->call/eval to avoid
614 invalidating pointers.
615
6164.03 Sat Oct 6 21:24:00 CEST 2007
617 - added Coro::throw method.
618 - minor code cleanups.
619
6204.02 Sat Oct 6 02:36:47 CEST 2007
621 - fix a very minor per-coroutine memleak (a single codereference).
622 - fixed a bug where the currently in-use c context would be freed
623 prematurely (can happen only when programs change the stacksize
624 or use tracing).
625 - tracing can no longer keep a coro alive after it terminated.
626 - do static branch prediction in the common path for gcc. gives
627 about 2-5% speed improvement here.
628
6294.01 Fri Oct 5 22:10:49 CEST 2007
630 - instead of recreating *a* standard output handle we simply
631 use STDOUT, which is faster and hopefully more robust.
632
6334.0 Fri Oct 5 12:56:00 CEST 2007
634 - incompatibly changed Coro::Storable::freeze.
635 - major new feature: added Coro::Debug, for interactive coroutine
636 debugging, tracing and much more.
637 - major bug fix: unbelievable, but true: $_, $/ and many other
638 "saved" variables actually weren't being saved. This has been fixed,
639 of course, while increasing performance while losing all the save
640 flags.
641 - save flags are gone, and all the api functions dealing with them.
642 - added Coro::Semaphore::adjust.
643 - added Coro::Util::fork_eval.
644 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
645 - added Coro::killall.
646 - reduce initial stack sizes to allow for "micro-coroutines".
647 - better async_pool resource management, moved parts of async_pool
648 handling to XS (major speed improvement).
649 - actually croak before modifying important data structures.
650 - refuse to transfer while compiling.
651 - possibly support eval EXPR better now.
652 - enable assembly per default on linux+bsd x86+amd64.
653 - all internal members were renamed _something for easier subclassing.
654 - many minor tweaks.
655
6563.63 Wed May 16 14:10:06 CEST 2007
657 - implement handcoded assembly for x86/amd64 SVR ABI.
658
6593.62 Fri Apr 27 21:36:06 CEST 2007
660 - upgrade libcoro (which might set unwind info correctly).
661 - change default on linux to setjmp/longjmp.
662
6633.61 Thu Apr 19 12:36:18 CEST 2007
664 - Coro::Storable caused an endless loop when thawing invalid
665 pst's sometimes.
666 - use a Semaphore in Coro::Storable, as Storable doesn't
667 seem to be reentrant (although it is documented to
668 be threadsafe...).
669 - fix Coro::Signal to bring back the original unreliable
670 but stateful semantics.
671 - fixed a lot of typos in Coro.pm (patch submitted by David
672 Steinbrunner, which applied flawlessly).
673
6743.6 Sat Apr 14 17:13:31 CEST 2007
675 - added some bugfixes to get eg/myhttpd working again.
676 - added Coro::Storable for often-cede'ing freeze/thaw.
677 - try to do a clean exit when a coroutine calls exit
678 (EXPERIMENTAL).
679 - got rid of indirect call through _coro_init.
680 - updated the partly antique examples in eg/ to
681 work again and be a bit less magic, too.
682 - fixed Coro::Signal semantics to work as documented again.
683
6843.55 Sun Mar 25 01:20:47 CET 2007
685 - add SAVE_DEFFH to save the default filehandle and enable
686 it by default.
687 - finally move socket-operations from Coro::Socket to Coro::Handle
688 to be able to unblock foreign sockets.
689 - add Coro::State::save_also and guarded_save.
690 - add count accessor to Coro::Semaphore.
691 - add Coro::State::cctx_stacksize.
692 - just for the fun of it, do not rely on implicit context,
693 which can dramatically improve performance, but people
694 using windows-process-emulation perls do not care much about
695 performance.
696
6973.51 Sun Mar 4 14:18:04 CET 2007
698 - fixed a problem when you weakened references to Coro::State's
699 (patch by Yuval Kogman).
700
7013.501 Wed Feb 28 12:44:07 CET 2007
702 - rename some global symbols as macosx from hell redefines
703 them without asking.
704
7053.5 Tue Feb 13 20:22:53 CET 2007
706 - do AnyEvent model detection earlier, avoiding problems
707 caused by first using AnyEvent and later Coro::Event.
708 - implement and document Coro::Event event objects.
709 - fix a potential problem in Coro::Event causing crashes.
710 - initialise PL_comppad when creating a new coroutine,
711 avoids crashes on early coro destruction.
712
7133.41 Mon Jan 22 19:19:49 CET 2007
714 - readline on Coro::Handle did not support undefined $/,
715 nor did it deliver partial lines on EOF or error.
716 - implement malloc fallback for stack allocation because
717 stupid broken idiotic OSX has a stupid broken
718 idiotic fits-the-whole-os mmap "implementation" and
719 my dick feels longer if Coro is portable even to
720 obsolete platforms.
721
7223.4 Fri Jan 19 21:52:54 CET 2007
723 - remove t/09_timer.t, as it isn't really testing much
724 but was rather flaky in practise.
725 - async_pool coro would keep arguments and callback alive until
726 it was reused.
727 - cancellation of a coroutine could cause spurious idle calls
728 in cede_notself.
729
7303.3 Sat Jan 6 03:45:00 CET 2007
731 - implement $coro->on_destroy.
732 - Coro::Event blocking semantics have been changed,
733 documented and - hopefully - improved.
734 - fix nice adding, not subtracting, from priority.
735 - fix ->prio and api_is_ready (patch by Mark Hinds).
736 - fixed an assert ("... == prev__cctx->idle_te")
737 that could errronously trigger.
738 - fix various large and small memleaks.
739 - use a (hopefully) more stable cancel implementation
740 that immediately frees the coroutine data.
741 - cede/cede_notself return a status now.
742 - added Coro::guard function.
743 - added a global coroutine pool for jobs (on my machine,
744 I can create and execute 48k simple coros/s with async,
745 and 128k coros with async_pool).
746 - Coro::AIO now uses the coroutine priority as io priority.
747
7483.2 Fri Dec 22 05:07:09 CET 2006
749 - improve portability to slightly older perls.
750 - use cleaner coroutine destruction.
751 - simplify configuration for users.
752 - optionally (unrecommended) prefer perl functions over
753 their coro replacements.
754
7553.11 Tue Dec 5 13:11:24 CET 2006
756 - fixed some bogus assert's, but as perl.h disables assert even
757 without NDEBUG (thank you very much), not too many people should
758 notice (that did include myself). Andreas König noticed, though :)
759 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
760 it. Also noticed by Andreas König :)
761 - save/restore tainted status.
762 - verified to pass the testsuite on my 5.9.5.
763
7643.1 Mon Dec 4 23:03:40 CET 2006
765 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
766 - incompatible change: transfer flags are now per-state.
767 - give a better error message on deadlock.
768 - document Coro::nready.
769 - enhanced testsuite.
770
7713.01 Sun Dec 3 23:47:42 CET 2006
772 - forgot to include Coro::Timer.
773
7743.0 Sun Dec 3 22:57:25 CET 2006
775 - the "FINALLY COMPLETELY STABLE" release (coming soon:
776 the "FAMOUS LAST WORDS" release).
4 - reimplement a new stack sharing algorithm. 777 - implement a new stack sharing algorithm, which uses a stack
778 pool (size currently hardcoded).
5 - make stack sharing mandatory (it no longer uses a heuristic). 779 - make stack sharing mandatory (it no longer uses a heuristic).
6 - eval/die no longer cause weird problems under heavy use. 780 - eval/die no longer cause weird problems under heavy use.
7 - Coro::Event could cause livelocks if it was used but 781 - Coro::Event could cause livelocks if it was used but
8 no Coro::Event watchers were used. 782 no Coro::Event watchers were used.
783 - Coro::Event now uses asynccheck as prepare does not
784 check for changed watchers.
785 - Coro::Event allows multiple waiting coros and will wake up one
786 per event.
9 - Coro::Event should be cleaner and more efficient now. 787 - Coro::Event should be cleaner and more efficient now.
788 - new utility function Coro::unblock_sub.
789 - document the sad fact that Event is no longer reentrant.
790 - putting a coroutine into the ready queue twice could under
791 some circumstances lead to stack corruption.
10 - minor incompatibility: subclassing Coro::State is supported 792 - minor incompatibility: subclassing Coro::State is supported
11 directly now without going through a _coro_state member. 793 directly now without going through a _coro_state member.
12 - state/coro switching is much faster now. 794 - state/coro switching is much faster now.
13 - very minor optimisations and code/documentation cleanup. 795 - very minor optimisations and code/documentation cleanup.
14 - avoid problems due to compiler inlining. 796 - avoid problems due to compiler inlining.
15 - removed timers from Coro::Timer -> use AnyEvent instead. 797 - removed timers from Coro::Timer -> use AnyEvent instead.
16 - replaced Coro::idle coroutine by (cleaner) idle handler. 798 - replaced Coro::idle coroutine by (cleaner) idle handler.
17 - updated to newest libcoro. 799 - updated to newest libcoro.
18 - implement enhanced support for valgrind. 800 - implement enhanced support for valgrind.
801 - implement is_ready and return value for ready.
802 - removed Coro::Cont, it was a misnomer (it's generators),
803 and it was somewhat annoying to get it right. Will come back
804 if somebody asks for it :)
805 - many ->wait methods and Coro::Event could return
806 spuriously without the event having happened.
19 807
202.5 Tue Nov 7 12:22:33 CET 2006 8082.5 Tue Nov 7 12:22:33 CET 2006
21 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket 809 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
22 use AnyEvent, moved them to Coro/. 810 use AnyEvent, moved them to Coro/.
23 - added Coro::LWP which contains all the uglyness required to 811 - added Coro::LWP which contains all the uglyness required to

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines