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

Comparing Coro/Changes (file contents):
Revision 1.181 by root, Tue Dec 12 04:21:09 2006 UTC vs.
Revision 1.597 by root, Mon Oct 8 23:13:36 2012 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: better (earlier) error messages on broken transfer's 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: "internals mode" 4TODO: channel->maxsize(newsize)?
5TODO: "use defaults for everything" 5TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
6TODO: __GCC_HAVE_DWARF2_CFI_ASM
6 7
86.10
9 - updated
10
116.09 Sat Oct 6 23:25:02 CEST 2012
12 - Coro::EV I/O watchers were not interruptible by exceptions
13 (Coro::State::throw) (testcase by sten).
14 - ->throw now puts threads into the ready queue, as this seems to
15 be expected by existing code, and code that doesn't cope with spurious
16 wakeups needs fixing anyway.
17 - use fd -1 in mmap.
18 - cast I32 to int in error message printf.
19 - warn about broken so-called "hardened" kernels.
20
216.08 Fri Apr 13 12:05:47 CEST 2012
22 - be more aggressive about exiting like perl does - formerly,
23 exiting from the non-main thread would not execute END blocks.
24
256.07 Fri Nov 11 21:21:48 CET 2011
26 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
27 - update ecb.h.
28
296.06 Mon Aug 8 23:59:48 CEST 2011
30 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
31 backend, which is an order of magnitude slower. unfortunately, cygwins
32 pthread implementation isn't very complete either, so allocate the stack
33 twice just to be sure.
34 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
35 in its ithreaded perl - best recompile cygwin and use the 'w'indows
36 backend for much better performance. also disable ithreads for
37 even better performance...).
38
396.05 Thu Aug 4 21:36:36 CEST 2011
40 - blush, condvar values would not be propagated from send to recv anymore
41 (reported by Chip Salzenberg).
42 - use exponential increase for the readline buffer length in
43 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
44
456.04 Wed Aug 3 17:15:45 CEST 2011
46 - use even more efficient and more compatible condvars for
47 compatibility to AnyEvent 6.x :)
48 - more inconsequential ecb.h updates.
49
506.03 Wed Aug 3 11:41:30 CEST 2011
51 - change how Coro patches AnyEvent condvars for compatibility to
52 AnyEvent 6.x.
53 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
54 hasn't been included.
55
566.02 Wed Jul 13 04:35:19 CEST 2011
57 - "improve portability to Gentoo" - gentoo manages to put perl variables
58 in memory areas that are farther than 2gb apart, which the jit couldn't
59 handle and barfed. now it's just a bit slower on gentoo and similar
60 systems.
61
626.01 Sun Jul 3 12:31:14 CEST 2011
63 - workarounds are good, but the test for whether pthreads are used
64 was not good. thisone should be better.
65 - check differently whether gcc generates cfi instructions itself.
66
676.0 Wed Jun 29 19:43:35 CEST 2011
68 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
69 destroyed and cleaned up automatically (e.g. async { schedule }).
70 - implement a JIT compiler for part of the thread switch code,
71 which gives a 50% speed improvement on threaded perls, and
72 about 4% on non-thraeded perls (so threaded perls now finally
73 reach about half the speed of non-threaded perls).
74 - slightly modernise Coro::Intro, add section about rouse functions.
75 - avoid DEFSV and ERRSV, giving another 10% improvement
76 in thread switching.
77 - Coro::State->is_destroyed is now called is_zombie.
78 - implement a Coro->safe_cancel method that might fail, but
79 cancels in a "safer" way if it succeeds.
80 - add preliminary support for DEBUGGING perls.
81 - get rid of two hash-accesses when initialising a new Coro - this
82 speeds up coro creation by almost a factor of two.
83 - croak when a coro that is being cancelled tries to block
84 (e.g. while executing a guard block), instead of crashing or
85 deadlocking.
86 - use a more robust and also faster method to identify Coro::State
87 objects - speeds up everything a bit.
88 - implement Coro->cancel in XS for a 20% speed improvement, and to
89 be able to implement mutual cancellation.
90 - speed up context switches by a percent or two by more efficiently
91 allocating context stack entries.
92 - implement Coro->join and Coro->on_destroy in XS for a speedup and
93 a reduction in memory use.
94 - cancelling a coro while it itself is cancelling another coro is
95 now supported and working, instead of triggering an assertion.
96 - be a bit more crash-resistant when calling (buggy) on_destroy
97 callbacks (best effort).
98 - move on_destroy into the slf_frame, to allow extension slf
99 functions to have destructors.
100 - get rid if coro refcounting - simply crash in other interpreter
101 threads by nulling the pointers on clone.
102 - simplify warn/die hook handling when loading Coro - the convoluted
103 logic seems to be no longer neccessary.
104 - use libecb instead of our own home-grown gcc hacks.
105 - document alternatives to Coro::LWP. Please use them :)
106 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
107 sigaltstack. Really. wine suffers from it, erlang suffers from it,
108 and it's known since at least 2006.
109
1105.372 Wed Feb 23 06:14:30 CET 2011
111 - apparently mingw doesn't provide a working gettimeofday, try to
112 work around that by relying on Time::HiRes (indirectly brought to
113 my attention by Max Maischein).
114 - fix some portability issues when Time::HiRes was used.
115
1165.371 Mon Feb 21 14:36:08 CET 2011
117 - backport to windows process emulation code again.
118
1195.37 Sat Feb 19 07:49:44 CET 2011
120 - add a big "Coro thread life cycle" section to "man Coro".
121 - try a tentative workaround against the breakage that 5.13 has
122 introduced without depreciation period. sigh.
123 - no longer use Time::HiRes if gettimeofday is available, which
124 saves quite a lot of memory.
125
1265.36 Sun Feb 13 05:33:41 CET 2011
127 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
128 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
129 method call.
130 - undocument Coro::Timer::sleep and obsolete whole module.
131 - optimise Coro::Timer::timeout memory and cpu usage.
132 - slightly updated Coro::Intro for recent changes.
133 - do not initialise PL_dirty anymore.
134
1355.25 Thu Nov 11 01:08:39 CET 2010
136 - try a different approach on netbsd - netbsd 5 finally has marginally
137 working pthreads, but still broken ucontext/sigaltstack.
138 - openbsd 4.8 finally got their act together, Coro works out of the box
139 with asm, setjmp and pthreads (no change, just informational).
140
1415.24 Sat Oct 23 11:27:12 CEST 2010
142 - port to the EV 4.0 API.
143 - work around bugs in mingw32, making strawberry perl work
144 out of the box.
145 - correctly modify Coro::AIO function prototypes
146 so that they reflect the "no optional parameters" rule.
147 - "ported" libcoro to C++.
148
1495.23 Mon May 17 18:50:42 CEST 2010
150 - be more resistant to ordering changes when initialising
151 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
152 Waldorf).
153 - document that perl 5.12 deliberately removed support for cloning.
154
1555.22 Wed Apr 14 03:55:35 CEST 2010
156 - correctly return udnef on errors in Coro::Handle::read/write
157 (testcase by Marc Mims).
158 - convert Coro::Util into a "perl compatibility wrapper" - the functions
159 are less useful now, but are drop-in replacements for existing
160 functions, listing better alternatives in the documentation. This also
161 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
162 with Coro::Util::inet_aton.
163 - do not override $Coro::idle unconditionally in Coro.pm, as other
164 modules could have provided their own idle coro already
165 (for exmaple, Coro::AnyEvent).
166 - fix Coro::Util::gethost* functions.
167 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
168
1695.21 Wed Dec 16 07:19:51 CET 2009
170 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
171 together.
172 - add some examples on how to combine other event loops with Coro in
173 Coro::AnyEvent, and how to run it (and not to block). Seems to be
174 the most common source of confusion.
175 - try to catch people naively blocking in an event callback.
176 - work around the perl filehandle bug issue in conjunction with
177 older common::sense (as indirectly pointed out by ZSystem).
178 - clarify the "not from signal handlers" section.
179
1805.2 Sun Oct 4 14:54:24 CEST 2009
181 - Coro::Storable destroyed the prototypes of the functions it wrapped.
182 - export rouse_cb and rouse_wait by default now.
183 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
184 - new method $state->swap_sv.
185 - added section on "windows process emulation" to the manpage,
186 after a not-so-fruitful (nor-friendly) "discussion" with chip
187 salzenberg (discussion implies arguments, but his only arguments
188 were ad-hominems, one wonders why he started it in the first
189 place). I hope this explains it well enough for him to understand,
190 and maybe well enough for others to understand.
191 - use common::sense everywhere now.
192 - idle callbacks are no longer supported, use idle coros instead.
193 - print a thread listing when a deadlock is detected.
194
1955.17 Sat Aug 22 23:09:31 CEST 2009
196 - work around a bug in the perl debugger causing crashes
197 when running under the debugger by marking _pool_handler as nodebug.
198 - speed up Coro::async considerably.
199 - try some hacks to get netbsd to work "more often" - their broken
200 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
201 - convert Coro to AE 5.0 API.
202
2035.162 Tue Jul 28 04:04:03 CEST 2009
204 - perl 5.8.2 is now minimum requirement.
205 - skip t/19_handle.t on broken windows perls.
206
2075.161 Wed Jul 22 04:47:38 CEST 2009
208 - Coro::AnyEvent::poll could have a different prototype when EV was
209 used as backend (analyzed by Tatsuhiko Miyagawa).
210 - Coro::AnyEvent errornously initialised the event loop when loaded,
211 not on demand.
212 - try to workaround rare */t/01_unblock.t failures.
213
2145.16 Tue Jul 21 01:44:37 CEST 2009
215 - Coro::AnyEvent failed to hook into the event loop
216 when no threads had been readied between detecting
217 the event loop and actually running it.
218 - considerably speed up Coro::Select by taking avdantage
219 of AnyEvent > 4.8 and some other optimisations.
220 - implement paragraph readline mode in Coro::Handle
221 (based on patches by Zsbán Ambrus).
222 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
223 and analyzed by Yasuhiro MATSUMOTO).
224 - clarified libcoro license and copyright.
225 - someone stole my EXTRA_META!!!
226 - implement Coro::Select::patch_pp_sselect and it's brother,
227 for hardcode select overriding.
228
2295.151 Mon Jul 6 05:41:57 CEST 2009
230 - backport to windows process emulation code again (patch by
231 Yasuhiro MATSUMOTO).
232 - slightly update Coro::MakeMaker.
233
2345.15 Tue Jun 30 10:28:06 CEST 2009
235 - deprecate Coro::Socket, document how to get ipv6 support via
236 AnyEvent::Socket instead.
237 - implement signal->wait ($cb) interface, similar to semaphores.
238 - work around SvOK not supporting getmagic, so we have to getmagic
239 to test for undef :/ (reported by Matthias Waldorf).
240 - load Coro::AnyEvent in all modules using AnyEvent.
241 - work around perl corrupting our internal data structures,
242 reported by Tokuhiro Matsuno.
243 - enable per-coroutine real and cpu time gathering
244 (Coro::State::enable_times).
245
2465.14 Wed Jun 24 01:37:48 CEST 2009
247 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
248 as this seems to have been a difficult concept (poll, sleep, idle,
249 idle_upto).
250 - add Coro::AnyEvent::readable/writable functions.
251 - clarify Coro::EV/Event/AnyEvent manpages.
252 - free per-thread global scalars in the thread calling ->cancel, to
253 avoid crashes when $_, $@ etc., are magical but some of those
254 had already been freed.
255 - "unexperimentalise" the callback interface for Coro::Semaphore.
256 - speed up ready queue management/context switching by using a linked
257 list instead of an array (~5-10%).
258 - implement "watch" command in Coro::Debug shells.
259 - for fun, implement time-slicing as an example in the manpage.
260 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
261 Coro::Event, use the normal AnyEvent handling instead of dieing
262 (the same is true for Coro::Handle).
263 - properly document Coro::EV::timed_io_once.
264 - avoid unneccessary ->cancel calls in Coro::Handle.
265 - maybe make it work on mingw32 with win32 backend
266 (based on patch by Yasuhiro Matsumoto).
267
2685.132 Fri May 29 09:00:39 CEST 2009
269 - do not keep a reference to the argument itself in
270 Coro::Semaphore::guard, as it could change later.
271 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
272 should support a prepare callback.
273
2745.131 Mon Mar 16 23:20:37 CET 2009
275 - implement and document Coro->suspend, Coro->resume.
276 - fix Coro::Select implementation to not (often) close
277 the passed file descriptors (testcase provided by pippijn).
278
2795.13 Mon Dec 15 21:51:42 CET 2008
280 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
281 - don't set diehook to C<undef> but instead to NULL, to avoid
282 spurious warnings.
283 - fix a lot of bugs in Coro::SemaphoreSet.
284 - Coro::SemaphoreSet will less often create a semaphore needlessly.
285 - add Coro::SemaphoreSet::count and wait methods.
286 - take advantage of the new Guard module.
287 - deprecate Coro::guard.
288 - try to fix the dreaded 01_unblock tests once more. I hate it when
289 testsuites need more fixing than the code they are supposed to test.
290 - croak in more cases when a required callback isn't resolvable.
291 - fix some minor issues in Coro::State->call/eval.
292 - use current coroutine context instead of a temporary one
293 when temporarily switching to another coroutine.
294 - do not call C-level on_destroy handlers during global destruction,
295 to avoid needless segfaults.
296
2975.12 Sun Dec 7 13:30:38 CET 2008
298 - add default config for MirOS, which seems to be a bug-to-bug
299 compatible fork of openbsd ("world domination by releasing
300 openbsd cvs before the openbsd folks do it" or so :).
301 - free_padlist did destroy the names pad, not good, but didn't
302 seem to bother perl - this could fix issues such as eval ""
303 inside a function called from multiple coroutines.
304 - use a different method to detect destruction time.
305 - be more careful when freeing padlists just before global
306 destruction.
307 - fixed and expanded the call/cc example.
308 - renamed _terminate to _coro_run.
309 - new method Coro::Channel->shutdown.
310 - try pthreads on openbsd <4.4 (broken sigaltstack, will
311 pthreads fare better?).
312 - be less picky about destroying "a" running coroutine.
313
3145.11 Tue Nov 25 21:49:05 CET 2008
315 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
316 working code with failed assertions, introducing more bugs than
317 it fixes, requiring elaborate workarounds :(
318
3195.1 Mon Nov 24 08:54:59 CET 2008
320 - wrote a small introductory tutorial - Coro::Intro.
321 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
322 - Coro::Select did errornously dup the file descriptors
323 and didn't work with all AnyEvent backends.
324 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
325 LWP data transfers.
326 - disassociate c contexts from coro objects - this is agruably more
327 correct, but mostly allows sharing of cctxs between coro and state
328 objects, for added memory savings and speed increases.
329 - bumped $Coro::POOL_RSS up to 32kb by default.
330 - no longer set the optype to OP_CUSTOM, as B::* understandably
331 doesn't like this very much (and we *are* a type of entersub).
332 - implement state cloning, just to prove that call/cc can be done.
333 - automatically load Coro::AnyEvent in Coro::Handle.
334 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
335 do not have this method (and don't need it either).
336 - speed up generic anyevent methods in Coro::Handle by using rouse
337 callbacks.
338 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
339 others. It also makes the debugger happier, as you can trace
340 through the idle threads now.
341 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
342 - eg/event was pretty broken.
343 - better 5.8.6 compatibility.
344
3455.0 Thu Nov 20 10:35:05 CET 2008
346 - NEW ARCHITECTURE: use the latest 4.x version if you experience
347 stability issues.
348 - bump API version to 7 - all dependents must be recompiled.
349 - removed timed_* functions - they were not being used anyways
350 and should be replaced by a more generic mechanism -
351 and were annoying to support anyways :)
352 - removed SemaphoreSet's waiter method - use sem method instead.
353 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
354 - async_pool did free a scalar value twice
355 ("Attempt to unreference...").
356 - fix a longstanding bug where calling terminate on a coro that
357 was waiting for a semaphore that was just becoming available
358 would cause a deadlock (semaphore would get into a state where
359 it was available but waiters were still blocked).
360 - calling throw on a coroutine that is waiting for a semaphore will
361 no longer make it acquire the semaphore (and thus leak a count).
362 - perl's process emulation is now not even theoretically supported
363 anymore.
364 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
365 conversion of callback-style to blocking-style.
366 - new methods $coro->schedule_to and ->cede_to, to specifically
367 schedule or cede to a specific coroutine.
368 - new function Coro::Semaphore::wait.
369 - use named constants in Coro::Channel (Richard Hundt).
370 - directly patch the entersub opcode calling SLF functions (cede,
371 transfer and so on). this does speed up context switching, but
372 more importanly, it frees us from the hardcoded behaviour of
373 entersub, so we might actually be able to return something from
374 those functions and atcually create new ones.
375 - take advantage of __builtin_frame_address on gcc.
376 - expose THX in coroapi (not sure whether this was a wise decision,
377 as "threaded" perls are running at half speed anyways).
378 - implement execute_slf (schedule-like-function) interface that makes
379 it possible to implement schedule-like-functions in XS.
380 - use new SLF interface to massively speed up Coro::EV by roughly a
381 factor of two.
382 - used new SLF interface to massively speed up Coro::Semaphore by a
383 factor of three.
384 - used new SLF interface to speed up Coro::AIO by roughly a factor of
385 four and reduce its memory usage considerably.
386 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
387 for a nice speedup and vastly more correct behaviour. Also implement
388 a new method "sem" to get at the underlying semaphore object.
389 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
390 (in comparison) 20-40% speedup.
391 - used new SLF interface to reimplement Coro::Signal gaining
392 some unknown (because I was too lazy), but certain, speedup, and also
393 making signals reliable for the first time.
394 - used new SLF interface and other optimisations to speed up async_pool
395 by a factor of two. It also doesn't rely on perl's exception mechanism
396 to exit anymore. The overhead for terminating an async_pool, coro over
397 a normal async is now very small.
398 - sped up coroutine creation/destruction by 40%.
399 - forgot to include Coro/libcoro/README in the dist for all these years.
400 - work around a freebsd pthreads bug (manual testcancel is required as
401 pthread_cond_wait isn't a cancellation point on freebsd).
402 - use new rouse functions to speed up and simplify Coro::BDB.
403 - make "prefer perl native functions" work with threaded perls.
404 - condense Coro::Debug ps output, hint at v and w flags.
405 - (libcoro) lots of minor cleanups and portability improvements.
406
4074.914 Wed Nov 19 12:54:18 CET 2008
408 - fix a disastrous bug in the readline optimisation
409 introduced in 4.801.
410
4114.913 Sat Nov 15 07:58:28 CET 2008
412 - async_pool did free a scalar value twice
413 ("Attempt to unreference...").
414
4154.912 Thu Nov 13 18:31:23 CET 2008
416 - minor cleanups.
417 - use much larger stacks on linux and perl < 5.8.8.
418 - Coro::Debug::new_unix_server did not unlink the socket
419 when destroyed.
420
4214.911 Tue Nov 11 04:26:17 CET 2008
422 - "port" to threaded perls.
423
4244.91 Mon Nov 10 05:36:38 CET 2008
425 - the ->throw exception object no longer leaks.
426 - creating a new cctx leaked a scopestack entry (memleak).
427 - new coroutines didn't get created with a zero flags field
428 (unknown impact).
429 - calling ->throw on a not-yet-started coroutine should now work
430 instead of being ignored.
431 - ->throw is now supported on Coro::State objects.
432 - clean up cctx creation code a bit.
433 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
434
4354.9 Sat Nov 8 17:45:27 CET 2008
436 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
437 - (libcoro) no longer leak threads in the experimental pthread backend,
438 also speed it up considerably.
439 - (libcoro) do not rely on makecontext passing void *'s unscathed.
440 - fix compiletime dependencies on libcoro in the Makefile.
441 - cctx_count wasn't always updated properly.
442 - Coro::State::cctx_stacksize wasn't applied correctly.
443 - new function Coro::State::cctx_max_idle.
444 - the default max number of idle C contexts is now 4.
445 - (libcoro) try harder to get _setjmp/_longjmp.
446 - (libcoro) cleanup and extend the libcoro API to officially
447 allow the creation of empty source contexts.
448 - very experimental workaround for the totally broken netbsd platform.
449 - tried to hack around openbsd bugs.
450
4514.804 Wed Nov 5 16:36:00 CET 2008
452 - Coro::Debug::new_unix_server would not create a non-blocking listening
453 socket, sometimes causing freezes.
454 - (libcoro) fix misaligned stack points for setjmp and assembly
455 methods, which can cause crashes on x86/x86_64 with a sufficiently
456 aggressive compiler.
457 - new function: Coro::Debug::new_tcp_server.
458 - move ->throw into the Coro class because it only works on coro objects.
459
4604.803 Mon Nov 3 17:16:12 CET 2008
461 - (libcoro) use a global asm statement to become independent of gcc
462 otpimisations for CORO_ASM (thanks to pippijn for the idea).
463 - try to workaround yet another broken bsd, this time dragonfly.
464
4654.802 Thu Oct 30 10:56:12 CET 2008
466 - support -fno-omit-frame-pointer on x86 with the assembly method.
467 - tune 01_unblock.t tests a bit.
468
4694.801 Wed Oct 22 18:33:37 CEST 2008
470 - improve readline speed for very long "lines".
471 - backport to 5.8.8.
472
4734.8 Thu Oct 2 13:34:40 CEST 2008
474 - new function Coro::AIO::aio_wait.
475 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
476 - greatly speed up and reduce memory usage of Coro::AIO requests.
477 - implement some other µ-optimisations.
478
4794.749 Mon Sep 29 14:40:12 CEST 2008
480 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
481
4824.748 Sat Sep 27 14:03:19 CEST 2008
483 - implement, but do not document, PerlIO::cede(granularity).
484 - Coro::Storable forgot to wrap Storable::pstore.
485 - work around the multitude of leaks and memory corruption
486 bugs in PerlIO::via by using our own C-level perliol. As a side
487 effect, Coro::Storable is now much, much, much faster.
488
4894.747 Tue Sep 23 01:59:41 CEST 2008
490 - fix a per-cv memleak (one empty array was leaked per
491 code reference).
492 - avoid a crash in coro->call|rss when the coroutine was already
493 destroyed (most noticably when using Coro::Debug::ps :)
494 - also protect *Storable::FILE.
495 - push up default storable granularity to 20ms.
496
4974.746 Sun Sep 21 03:22:20 CEST 2008
498 - be more insistent on locking Storable against reentrancy
499 in Coro::Storable.
500 - move swap_def?v and throw to Coro::State, as documented.
501
5024.745 Thu Jul 24 00:14:38 CEST 2008
503 - remove debugging code related to MgPV_nolen_const, also try to
504 make it compile with perl 5.8.6 (yes, apple apparently loves
505 outdated software). Reported by John S.
506
5074.744 Tue Jul 8 22:06:35 CEST 2008
508 - correctly provide default DIE/WARN handlers as documented.
509 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
510 implementations inside perl work fine for us.
511
5124.743 Mon Jun 16 00:21:57 CEST 2008
513 - when using Coro::EV without running EV::loop it could
514 result in busy-waiting for events, this has been fixed.
515 - reduce codesize and improve performance by using EV_DEFAULT_UC.
516
5174.742 Sat May 31 14:10:21 CEST 2008
518 - implement a workaround for (some) perl <5.8.8 versions.
519 - require EV 3.3+.
520
5214.741 Fri May 30 23:33:09 CEST 2008
522 - tell netbsd how utterly broken their imitation of an OS is
523 and refuse to build by default if pthreads are in use.
524 - switch to "s" method on all bsd's by default, as their ucontext
525 stuff seems just too broken.
526 - fix a bug in Coro::Select.
527
5284.74 Thu May 29 20:05:31 CEST 2008
529 - do not test Coro::LWP for lack of dependencies.
530 - do not test Coro::Select for lack of working perls.
531
5324.73 Thu May 29 2008
533 - fix a bug in Coro::EV which would cause it to block despite
534 there being runnable coroutines.
535 - sprinkle "no warnings" freely over everything, also suppress
536 warnings for some other modules.
537 - fix typo in WSAEWOULDBLOCK.
538
5394.72 Sun May 25 05:14:36 CEST 2008
540 - tweak META.yaml a bit, unfortunately, there is no documented way
541 to have optional dependencies with CPAN. doh :(
542 - avoid running some tests on windows because they would fail due to
543 perl bug (broken fork, broken pipes...).
544 - work around perl on windows bugs where perl returns undocumented
545 error codes for sysread, syswrite etc. by taking advantage
546 of AnyEvent's workaround for that problem.
547 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
548 a common perl implementation bug on windows.
549 - use unix domain sockets in testsuite to work around
550 common perl implementation bugs on widows (they are emulated by
551 tcp sockets on windows. ugh.)
552
5534.71 Sat May 24 20:01:27 CEST 2008
554 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
555 - take advantage of async name resolution of AnyEvent::Util.
556 - work around brutal inet_aton override in Coro::LWP.
557 - take advantage of the readyhook in Coro::EV, for smoother
558 scheduling.
559
5604.7 Sun May 11 00:32:19 CEST 2008
561 - completely reworked the Coro manpage.
562 - added Coro::AnyEvent, generic event loop integration.
563 - implement cancel, ready and kill commands in Coro::Debug.
564 - document find_coro in Coro::Debug.
565 - incompatible change: rename has_stack to has_cctx.
566 - Coro::AIO and Coro::BDB no longer force event model detection,
567 use AnyEvent::AIO and AnyEvent::BDB.
568
5694.6 Sat Apr 26 10:05:14 CEST 2008
570 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
571 by the Coro module. It was a mistake to have it in the first place.
572 - (experimental) support for activestate perl 5.10 (method "w").
573 - (experimental) support for strawberry perl (method "a").
574 - coro_sigelem_set did not return a value although it had to,
575 actual impact unknown.
576
5774.51 Mon Apr 14 13:28:27 CEST 2008
578 - make it compile again on 5.8.
579
5804.50 Thu Apr 10 09:43:17 CEST 2008
581 - I did it twice! (see 4.49).
582
5834.49 Sun Apr 6 21:23:31 CEST 2008
584 - grr, instead of compiling the recent changes on 5.10 only they
585 were compiled on 5.8 only.
586
5874.48 Sun Apr 6 20:36:46 CEST 2008
588 - allow coroutine switches during eval's under 5.10.x, as it seems
589 the parser is a per-interpreter option now, so this is safe
590 (this might fix the odd crashing bug).
591 - drop support for 5.9.x versions: they are dead, jim.
592
5934.47 Sun Apr 6 00:37:52 CEST 2008
594 - force cctx allocation on API calls: we know we need to force one
595 and gcc actually manages to confuse our heuristic nowadays,
596 leading to crashes and worse.
597 - document force_cctx.
598
5994.46 Fri Apr 4 22:05:43 CEST 2008
600 - upgrade libcoro, resulting in pthread-backend (which was only created
601 to fulfill the rules of the programming languages shootout).
602
6034.45 Thu Mar 13 11:55:36 CET 2008
604 - fix a file leaking bug in eg/mhyttpd that would allow
605 downloading of any file (reported by oesi).
606 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
607 (also add testcase).
608 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
609 apparently having loads of fun with that).
610 - implement $state->swap_defsv, ->swap_defav and document ->throw.
611
6124.4 Wed Feb 13 16:44:29 CET 2008
613 - only assume .cfi pseudo isns exist on GNU/Linux.
614 - add get_timed method to Coro::Channels.
615 - fixed Coro::Debug manpage.
616 - perl 5.11 compatbility improvement based on patch by
617 Andreas König.
618
6194.37 Sun Jan 20 11:25:23 CET 2008
620 - soften the check for an active parser for perl 5.10 (avoiding
621 "Coro::State::transfer called while parsing" in many cases).
622
6234.36 Sun Jan 13 10:53:56 CET 2008
624 - reset diehook when terminating from an async_pool as to not
625 trigger any __DIE__ handlers.
626
6274.35 Sun Jan 13 04:14:13 CET 2008
628 - "bt" debug command now displays any exceptions
629 from longmess and also skips the innermost
630 stackframes, giving more useufl output.
631 - allow backtraces in coroutines blocked in Coro::EV,
632 at a <1% speed hit.
633 - handle localising of $SIG{__WARN__} and __DIE__
634 properly (with a proper amount of dirty hacking).
635
6364.34 Sat Dec 22 17:49:53 CET 2007
637 - upgrade to EV version 2.0 API.
638
6394.33 Mon Dec 17 08:36:12 CET 2007
640 - make Coro::AIO etc. loadable in the absence of EV.
641
6424.32 Mon Dec 17 07:46:02 CET 2007
643 - majorly improved Coro::Handle's performance with Coro::EV.
644 - implemented the until now mythical Coro::BDB module.
645 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
646
6474.31 Wed Dec 5 12:32:39 CET 2007
648 - remove warn statement form Coro::Util.
649
6504.3 Tue Dec 4 20:33:14 CET 2007
651 - calls to the idle function could cause stack corruption
652 when the stack changed.
653 - do no longer rely on the presence of EV::DNS if EV is used
654 (because it is gone), but instead take avdantage of EV::ADNS
655 if available.
656 - add ($) prototypes to all functions in Coro::Storable.
657 - use a conventional (safer) idle callback in Coro::EV.
658 - do accept quickly in Coro::Debug to avoid endless loops.
659
6604.22 Fri Nov 30 16:04:04 CET 2007
661 - really use optimised versions for Event and EV in Coro::Util
662 and Coro::Handle.
663
6644.21 Sun Nov 25 10:48:59 CET 2007
665 - fix a spurious memory read.
666 - Coro::EV no longer keeps the eventloop "alive".
667
6684.2 Fri Nov 9 20:47:05 CET 2007
669 - enable/disable tracing from a new coroutine, not a pooled one.
670 - fix a memleak in Coro::Event.
671 - removed killall call from fork_eval.
672 - made sure store_fd is already loaded so that fork_eval
673 does not have to parse autoload in each subprocess.
674 - only use assembly method if -O switch is in $Config{optimize}.
675 - add (optional) Coro::EV module, so far the best event loop module
676 directly supported by Coro.
677 - if the event model is EV, use EV::DNS to resolve
678 stuff in Coro::Util.
679 - don't get confused by multiple event notifications in Coro::Handle.
680 - initial support for EV (libevent interface).
681 - require Event and EV using configure_requires, to force their existance.
682
6834.13 Wed Oct 24 07:26:45 CEST 2007
684 - add Coro::Storable::blocking_thaw.
685 - use a vastly more complicated technique to localise
686 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
687 - use a coroutine for the idle callback Coro::Event,
688 instead of running Event in the current coroutine context.
689 This also catches recursive invocations.
690 - actually report fork errors in gethostbyname and inet_aton.
691
6924.11 Thu Oct 11 02:40:24 CEST 2007
693 - port to threaded perls.
694
6954.1 Thu Oct 11 02:38:16 CEST 2007
696 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
697 be local to each coro (see Coro::State).
698 - incompatible change: for very deep reasons, cede and cede_notself
699 cannot return anything, so nothing will be returned.
700 - possibly bring back 5.10 compatibility (untested).
701 - work around stupid (and wrong) warning on 5.10 :(.
702 - overlay the saved state over the context stack. This saves
703 a few hundred bytes per coroutine on average and also
704 speeds up context switching a bit.
705 - further tune default stack sizes.
706 - (more) correctly calculate stack usage in coro_rss.
707 - Coro::Storable::blocking_* did not properly lock
708 resulting in races between coroutines.
709 - added Coro::Storable::guard.
710 - stopping to trace a coroutine could destroy the cctx of
711 an unrelated coroutine.
712 - explain the relationship between Perl and C coroutines in
713 more detail in Coro::State.
714 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
715 causing a fork per resolve. This also affected Coro::Socket.
716 - switch to a separate stack in $coro->call/eval to avoid
717 invalidating pointers.
718
7194.03 Sat Oct 6 21:24:00 CEST 2007
720 - added Coro::throw method.
721 - minor code cleanups.
722
7234.02 Sat Oct 6 02:36:47 CEST 2007
724 - fix a very minor per-coroutine memleak (a single codereference).
725 - fixed a bug where the currently in-use c context would be freed
726 prematurely (can happen only when programs change the stacksize
727 or use tracing).
728 - tracing can no longer keep a coro alive after it terminated.
729 - do static branch prediction in the common path for gcc. gives
730 about 2-5% speed improvement here.
731
7324.01 Fri Oct 5 22:10:49 CEST 2007
733 - instead of recreating *a* standard output handle we simply
734 use STDOUT, which is faster and hopefully more robust.
735
7364.0 Fri Oct 5 12:56:00 CEST 2007
737 - incompatibly changed Coro::Storable::freeze.
738 - major new feature: added Coro::Debug, for interactive coroutine
739 debugging, tracing and much more.
740 - major bug fix: unbelievable, but true: $_, $/ and many other
741 "saved" variables actually weren't being saved. This has been fixed,
742 of course, while increasing performance while losing all the save
743 flags.
744 - save flags are gone, and all the api functions dealing with them.
745 - added Coro::Semaphore::adjust.
746 - added Coro::Util::fork_eval.
747 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
748 - added Coro::killall.
749 - reduce initial stack sizes to allow for "micro-coroutines".
750 - better async_pool resource management, moved parts of async_pool
751 handling to XS (major speed improvement).
752 - actually croak before modifying important data structures.
753 - refuse to transfer while compiling.
754 - possibly support eval EXPR better now.
755 - enable assembly per default on linux+bsd x86+amd64.
756 - all internal members were renamed _something for easier subclassing.
757 - many minor tweaks.
758
7593.63 Wed May 16 14:10:06 CEST 2007
760 - implement handcoded assembly for x86/amd64 SVR ABI.
761
7623.62 Fri Apr 27 21:36:06 CEST 2007
763 - upgrade libcoro (which might set unwind info correctly).
764 - change default on linux to setjmp/longjmp.
765
7663.61 Thu Apr 19 12:36:18 CEST 2007
767 - Coro::Storable caused an endless loop when thawing invalid
768 pst's sometimes.
769 - use a Semaphore in Coro::Storable, as Storable doesn't
770 seem to be reentrant (although it is documented to
771 be threadsafe...).
772 - fix Coro::Signal to bring back the original unreliable
773 but stateful semantics.
774 - fixed a lot of typos in Coro.pm (patch submitted by David
775 Steinbrunner, which applied flawlessly).
776
7773.6 Sat Apr 14 17:13:31 CEST 2007
778 - added some bugfixes to get eg/myhttpd working again.
779 - added Coro::Storable for often-cede'ing freeze/thaw.
780 - try to do a clean exit when a coroutine calls exit
781 (EXPERIMENTAL).
782 - got rid of indirect call through _coro_init.
783 - updated the partly antique examples in eg/ to
784 work again and be a bit less magic, too.
785 - fixed Coro::Signal semantics to work as documented again.
786
7873.55 Sun Mar 25 01:20:47 CET 2007
788 - add SAVE_DEFFH to save the default filehandle and enable
789 it by default.
790 - finally move socket-operations from Coro::Socket to Coro::Handle
791 to be able to unblock foreign sockets.
792 - add Coro::State::save_also and guarded_save.
793 - add count accessor to Coro::Semaphore.
794 - add Coro::State::cctx_stacksize.
795 - just for the fun of it, do not rely on implicit context,
796 which can dramatically improve performance, but people
797 using windows-process-emulation perls do not care much about
798 performance.
799
8003.51 Sun Mar 4 14:18:04 CET 2007
801 - fixed a problem when you weakened references to Coro::State's
802 (patch by Yuval Kogman).
803
8043.501 Wed Feb 28 12:44:07 CET 2007
805 - rename some global symbols as macosx from hell redefines
806 them without asking.
807
8083.5 Tue Feb 13 20:22:53 CET 2007
809 - do AnyEvent model detection earlier, avoiding problems
810 caused by first using AnyEvent and later Coro::Event.
811 - implement and document Coro::Event event objects.
812 - fix a potential problem in Coro::Event causing crashes.
813 - initialise PL_comppad when creating a new coroutine,
814 avoids crashes on early coro destruction.
815
8163.41 Mon Jan 22 19:19:49 CET 2007
817 - readline on Coro::Handle did not support undefined $/,
818 nor did it deliver partial lines on EOF or error.
819 - implement malloc fallback for stack allocation because
820 stupid broken idiotic OSX has a stupid broken
821 idiotic fits-the-whole-os mmap "implementation" and
822 my dick feels longer if Coro is portable even to
823 obsolete platforms.
824
8253.4 Fri Jan 19 21:52:54 CET 2007
826 - remove t/09_timer.t, as it isn't really testing much
827 but was rather flaky in practise.
828 - async_pool coro would keep arguments and callback alive until
829 it was reused.
830 - cancellation of a coroutine could cause spurious idle calls
831 in cede_notself.
832
8333.3 Sat Jan 6 03:45:00 CET 2007
834 - implement $coro->on_destroy.
835 - Coro::Event blocking semantics have been changed,
836 documented and - hopefully - improved.
837 - fix nice adding, not subtracting, from priority.
838 - fix ->prio and api_is_ready (patch by Mark Hinds).
839 - fixed an assert ("... == prev__cctx->idle_te")
840 that could errronously trigger.
841 - fix various large and small memleaks.
842 - use a (hopefully) more stable cancel implementation
843 that immediately frees the coroutine data.
844 - cede/cede_notself return a status now.
845 - added Coro::guard function.
846 - added a global coroutine pool for jobs (on my machine,
847 I can create and execute 48k simple coros/s with async,
848 and 128k coros with async_pool).
849 - Coro::AIO now uses the coroutine priority as io priority.
850
8513.2 Fri Dec 22 05:07:09 CET 2006
7 - improve portability to slightly older perls. 852 - improve portability to slightly older perls.
8 - use cleaner coroutine destruction. 853 - use cleaner coroutine destruction.
854 - simplify configuration for users.
855 - optionally (unrecommended) prefer perl functions over
856 their coro replacements.
9 857
103.11 Tue Dec 5 13:11:24 CET 2006 8583.11 Tue Dec 5 13:11:24 CET 2006
11 - fixed some bogus assert's, but as perl.h disables assert even 859 - fixed some bogus assert's, but as perl.h disables assert even
12 without NDEBUG (thank you very much), not too many people should 860 without NDEBUG (thank you very much), not too many people should
13 notice (that did include myself). Andreas König noticed, though :) 861 notice (that did include myself). Andreas König noticed, though :)

Diff Legend

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