ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.444
Committed: Wed Nov 19 16:01:32 2008 UTC (15 years, 6 months ago) by root
Branch: MAIN
Changes since 1.443: +2 -1 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 Revision history for Perl extension Coro.
2
3 TODO: should explore PerlIO::coroaio (perl leaks like hell).
4 TODO: maybe implement a default message channel, very much like Erlang's
5 actor model (which is cool in a lot of important aspects (failures!),
6 but very lacking in others (higher level ipc)).
7
8 TODO: async_pool to SLF
9 TODO: _cancel to xs
10 5.0
11 - NEW ARCHITECTURE: use the latest 4.x version if you experience
12 stability issues.
13 - bump API version to 7 - all dependents must be recompiled.
14 - removed timed_* functions - they were not being used anyways
15 and should be replaced by a more generic mechanism -
16 and were annoying to support anyways :)
17 - removed SemaphoreSet's waiter method - use sem instead.
18 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
19 - async_pool did free a scalar value twice
20 ("Attempt to unreference...").
21 - fix a disastrous bug in the readline optimisation
22 introduced in 4.801.
23 - fix a longstanding bug where calling terminate on a coro that
24 was waiting for a semaphore that was just becoming available
25 would cause a deadlock (semaphore would get into a state where
26 it was available but waiters were still blocked).
27 - calling throw on a coroutine that is waiting for a semaphore will
28 no longer make it acquire the semaphore (and thus leak a count).
29 - perl's process emulation is now not even theoretically supported
30 anymore.
31 - new functions Coro::rouse_cb and Coro::rouse_wait.
32 - new function Coro::Semaphore::wait.
33 - use named constants in Coro::Channel (Richard Hundt).
34 - directly patch the entersub opcode calling SLF functions (cede,
35 transfer and so on). this does speed up context switching, but
36 more importanly, it frees us from the hardcoded behaviour of
37 entersub, so we might actually be able to return something from
38 those functions and atcually create new ones.
39 - take advantage of __builtin_frame_address on gcc.
40 - expose THX in coroapi (not sure whether this was a wise decision,
41 as "threaded" perls are running at half speed anyways).
42 - implement execute_slf (schedule-like-function) interface that makes
43 it possible to implement schedule-like-functions in XS.
44 - use new SLF interface to massively speed up Coro::EV by roughly a
45 factor of two.
46 - used new SLF interface to massively speed up Coro::Semaphore by a
47 factor of three.
48 - used new SLF interface to speed up Coro::AIO by roughly a factor of
49 four and reduce its memory usage considerably.
50 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
51 for a nice speedup and vastly more correct behaviour. Also implement
52 a new method "sem" to get at the underlying semaphore object.
53 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
54 (in comparison) 20-40% speedup.
55 - used new SLF interface to reimplement Coro::Signal gaining
56 some unknown (because I was too lazy), but certain, speedup, and also
57 making signals reliable for the first time.
58 - sped up coroutine creation/destruction by 40%.
59 - forgot to include Coro/libcoro/README in the dist for all these years.
60 - work around a freebsd pthreads bug (manual testcancel is required as
61 pthread_cond_wait isn't a cancellation point on freebsd).
62 - use new rouse functions to speed up and simplify Coro::BDB.
63 - make "prefer perl native functions" work with threaded perls.
64 - condense Coro::Debug ps output, hint at v and w flags.
65 - (libcoro) lots of minor cleanups and portability improvements.
66
67 4.913 Sat Nov 15 07:58:28 CET 2008
68 - async_pool did free a scalar value twice
69 ("Attempt to unreference...").
70
71 4.912 Thu Nov 13 18:31:23 CET 2008
72 - minor cleanups.
73 - use much larger stacks on linux and perl < 5.8.8.
74 - Coro::Debug::new_unix_server did not unlink the socket
75 when destroyed.
76
77 4.911 Tue Nov 11 04:26:17 CET 2008
78 - "port" to threaded perls.
79
80 4.91 Mon Nov 10 05:36:38 CET 2008
81 - the ->throw exception object no longer leaks.
82 - creating a new cctx leaked a scopestack entry (memleak).
83 - new coroutines didn't get created with a zero flags field
84 (unknown impact).
85 - calling ->throw on a not-yet-started coroutine should now work
86 instead of being ignored.
87 - ->throw is now supported on Coro::State objects.
88 - clean up cctx creation code a bit.
89 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
90
91 4.9 Sat Nov 8 17:45:27 CET 2008
92 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
93 - (libcoro) no longer leak threads in the experimental pthread backend,
94 also speed it up considerably.
95 - (libcoro) do not rely on makecontext passing void *'s unscathed.
96 - fix compiletime dependencies on libcoro in the Makefile.
97 - cctx_count wasn't always updated properly.
98 - Coro::State::cctx_stacksize wasn't applied correctly.
99 - new function Coro::State::cctx_max_idle.
100 - the default max number of idle C contexts is now 4.
101 - (libcoro) try harder to get _setjmp/_longjmp.
102 - (libcoro) cleanup and extend the libcoro API to officially
103 allow the creation of empty source contexts.
104 - very experimental workaround for the totally broken netbsd platform.
105 - tried to hack around openbsd bugs.
106
107 4.804 Wed Nov 5 16:36:00 CET 2008
108 - Coro::Debug::new_unix_server would not create a non-blocking listening
109 socket, sometimes causing freezes.
110 - (libcoro) fix misaligned stack points for setjmp and assembly
111 methods, which can cause crashes on x86/x86_64 with a sufficiently
112 aggressive compiler.
113 - new function: Coro::Debug::new_tcp_server.
114 - move ->throw into the Coro class because it only works on coro objects.
115
116 4.803 Mon Nov 3 17:16:12 CET 2008
117 - (libcoro) use a global asm statement to become independent of gcc
118 otpimisations for CORO_ASM (thanks to pippijn for the idea).
119 - try to workaround yet another broken bsd, this time dragonfly.
120
121 4.802 Thu Oct 30 10:56:12 CET 2008
122 - support -fno-omit-frame-pointer on x86 with the assembly method.
123 - tune 01_unblock.t tests a bit.
124
125 4.801 Wed Oct 22 18:33:37 CEST 2008
126 - improve readline speed for very long "lines".
127 - backport to 5.8.8.
128
129 4.8 Thu Oct 2 13:34:40 CEST 2008
130 - new function Coro::AIO::aio_wait.
131 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
132 - greatly speed up and reduce memory usage of Coro::AIO requests.
133 - implement some other µ-optimisations.
134
135 4.749 Mon Sep 29 14:40:12 CEST 2008
136 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
137
138 4.748 Sat Sep 27 14:03:19 CEST 2008
139 - implement, but do not document, PerlIO::cede(granularity).
140 - Coro::Storable forgot to wrap Storable::pstore.
141 - work around the multitude of leaks and memory corruption
142 bugs in PerlIO::via by using our own C-level perliol. As a side
143 effect, Coro::Storable is now much, much, much faster.
144
145 4.747 Tue Sep 23 01:59:41 CEST 2008
146 - fix a per-cv memleak (one empty array was leaked per
147 code reference).
148 - avoid a crash in coro->call|rss when the coroutine was already
149 destroyed (most noticably when using Coro::Debug::ps :)
150 - also protect *Storable::FILE.
151 - push up default storable granularity to 20ms.
152
153 4.746 Sun Sep 21 03:22:20 CEST 2008
154 - be more insistent on locking Storable against reentrancy
155 in Coro::Storable.
156 - move swap_def?v and throw to Coro::State, as documented.
157
158 4.745 Thu Jul 24 00:14:38 CEST 2008
159 - remove debugging code related to MgPV_nolen_const, also try to
160 make it compile with perl 5.8.6 (yes, apple apparently loves
161 outdated software). Reported by John S.
162
163 4.744 Tue Jul 8 22:06:35 CEST 2008
164 - correctly provide default DIE/WARN handlers as documented.
165 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
166 implementations inside perl work fine for us.
167
168 4.743 Mon Jun 16 00:21:57 CEST 2008
169 - when using Coro::EV without running EV::loop it could
170 result in busy-waiting for events, this has been fixed.
171 - reduce codesize and improve performance by using EV_DEFAULT_UC.
172
173 4.742 Sat May 31 14:10:21 CEST 2008
174 - implement a workaround for (some) perl <5.8.8 versions.
175 - require EV 3.3+.
176
177 4.741 Fri May 30 23:33:09 CEST 2008
178 - tell netbsd how utterly broken their imitation of an OS is
179 and refuse to build by default if pthreads are in use.
180 - switch to "s" method on all bsd's by default, as their ucontext
181 stuff seems just too broken.
182 - fix a bug in Coro::Select.
183
184 4.74 Thu May 29 20:05:31 CEST 2008
185 - do not test Coro::LWP for lack of dependencies.
186 - do not test Coro::Select for lack of working perls.
187
188 4.73 Thu May 29 2008
189 - fix a bug in Coro::EV which would cause it to block despite
190 there being runnable coroutines.
191 - sprinkle "no warnings" freely over everything, also suppress
192 warnings for some other modules.
193 - fix typo in WSAEWOULDBLOCK.
194
195 4.72 Sun May 25 05:14:36 CEST 2008
196 - tweak META.yaml a bit, unfortunately, there is no documented way
197 to have optional dependencies with CPAN. doh :(
198 - avoid running some tests on windows because they would fail due to
199 perl bug (broken fork, broken pipes...).
200 - work around perl on windows bugs where perl returns undocumented
201 error codes for sysread, syswrite etc. by taking advantage
202 of AnyEvent's workaround for that problem.
203 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
204 a common perl implementation bug on windows.
205 - use unix domain sockets in testsuite to work around
206 common perl implementation bugs on widows (they are emulated by
207 tcp sockets on windows. ugh.)
208
209 4.71 Sat May 24 20:01:27 CEST 2008
210 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
211 - take advantage of async name resolution of AnyEvent::Util.
212 - work around brutal inet_aton override in Coro::LWP.
213 - take advantage of the readyhook in Coro::EV, for smoother
214 scheduling.
215
216 4.7 Sun May 11 00:32:19 CEST 2008
217 - completely reworked the Coro manpage.
218 - added Coro::AnyEvent, generic event loop integration.
219 - implement cancel, ready and kill commands in Coro::Debug.
220 - document find_coro in Coro::Debug.
221 - incompatible change: rename has_stack to has_cctx.
222 - Coro::AIO and Coro::BDB no longer force event model detection,
223 use AnyEvent::AIO and AnyEvent::BDB.
224
225 4.6 Sat Apr 26 10:05:14 CEST 2008
226 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
227 by the Coro module. It was a mistake to have it in the first place.
228 - (experimental) support for activestate perl 5.10 (method "w").
229 - (experimental) support for strawberry perl (method "a").
230 - coro_sigelem_set did not return a value although it had to,
231 actual impact unknown.
232
233 4.51 Mon Apr 14 13:28:27 CEST 2008
234 - make it compile again on 5.8.
235
236 4.50 Thu Apr 10 09:43:17 CEST 2008
237 - I did it twice! (see 4.49).
238
239 4.49 Sun Apr 6 21:23:31 CEST 2008
240 - grr, instead of compiling the recent changes on 5.10 only they
241 were compiled on 5.8 only.
242
243 4.48 Sun Apr 6 20:36:46 CEST 2008
244 - allow coroutine switches during eval's under 5.10.x, as it seems
245 the parser is a per-interpreter option now, so this is safe
246 (this might fix the odd crashing bug).
247 - drop support for 5.9.x versions: they are dead, jim.
248
249 4.47 Sun Apr 6 00:37:52 CEST 2008
250 - force cctx allocation on API calls: we know we need to force one
251 and gcc actually manages to confuse our heuristic nowadays,
252 leading to crashes and worse.
253 - document force_cctx.
254
255 4.46 Fri Apr 4 22:05:43 CEST 2008
256 - upgrade libcoro, resulting in pthread-backend (which was only created
257 to fulfill the rules of the programming languages shootout).
258
259 4.45 Thu Mar 13 11:55:36 CET 2008
260 - fix a file leaking bug in eg/mhyttpd that would allow
261 downloading of any file (reported by oesi).
262 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
263 (also add testcase).
264 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
265 apparently having loads of fun with that).
266 - implement $state->swap_defsv, ->swap_defav and document ->throw.
267
268 4.4 Wed Feb 13 16:44:29 CET 2008
269 - only assume .cfi pseudo isns exist on GNU/Linux.
270 - add get_timed method to Coro::Channels.
271 - fixed Coro::Debug manpage.
272 - perl 5.11 compatbility improvement based on patch by
273 Andreas König.
274
275 4.37 Sun Jan 20 11:25:23 CET 2008
276 - soften the check for an active parser for perl 5.10 (avoiding
277 "Coro::State::transfer called while parsing" in many cases).
278
279 4.36 Sun Jan 13 10:53:56 CET 2008
280 - reset diehook when terminating from an async_pool as to not
281 trigger any __DIE__ handlers.
282
283 4.35 Sun Jan 13 04:14:13 CET 2008
284 - "bt" debug command now displays any exceptions
285 from longmess and also skips the innermost
286 stackframes, giving more useufl output.
287 - allow backtraces in coroutines blocked in Coro::EV,
288 at a <1% speed hit.
289 - handle localising of $SIG{__WARN__} and __DIE__
290 properly (with a proper amount of dirty hacking).
291
292 4.34 Sat Dec 22 17:49:53 CET 2007
293 - upgrade to EV version 2.0 API.
294
295 4.33 Mon Dec 17 08:36:12 CET 2007
296 - make Coro::AIO etc. loadable in the absence of EV.
297
298 4.32 Mon Dec 17 07:46:02 CET 2007
299 - majorly improved Coro::Handle's performance with Coro::EV.
300 - implemented the until now mythical Coro::BDB module.
301 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
302
303 4.31 Wed Dec 5 12:32:39 CET 2007
304 - remove warn statement form Coro::Util.
305
306 4.3 Tue Dec 4 20:33:14 CET 2007
307 - calls to the idle function could cause stack corruption
308 when the stack changed.
309 - do no longer rely on the presence of EV::DNS if EV is used
310 (because it is gone), but instead take avdantage of EV::ADNS
311 if available.
312 - add ($) prototypes to all functions in Coro::Storable.
313 - use a conventional (safer) idle callback in Coro::EV.
314 - do accept quickly in Coro::Debug to avoid endless loops.
315
316 4.22 Fri Nov 30 16:04:04 CET 2007
317 - really use optimised versions for Event and EV in Coro::Util
318 and Coro::Handle.
319
320 4.21 Sun Nov 25 10:48:59 CET 2007
321 - fix a spurious memory read.
322 - Coro::EV no longer keeps the eventloop "alive".
323
324 4.2 Fri Nov 9 20:47:05 CET 2007
325 - enable/disable tracing from a new coroutine, not a pooled one.
326 - fix a memleak in Coro::Event.
327 - removed killall call from fork_eval.
328 - made sure store_fd is already loaded so that fork_eval
329 does not have to parse autoload in each subprocess.
330 - only use assembly method if -O switch is in $Config{optimize}.
331 - add (optional) Coro::EV module, so far the best event loop module
332 directly supported by Coro.
333 - if the event model is EV, use EV::DNS to resolve
334 stuff in Coro::Util.
335 - don't get confused by multiple event notifications in Coro::Handle.
336 - initial support for EV (libevent interface).
337 - require Event and EV using configure_requires, to force their existance.
338
339 4.13 Wed Oct 24 07:26:45 CEST 2007
340 - add Coro::Storable::blocking_thaw.
341 - use a vastly more complicated technique to localise
342 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
343 - use a coroutine for the idle callback Coro::Event,
344 instead of running Event in the current coroutine context.
345 This also catches recursive invocations.
346 - actually report fork errors in gethostbyname and inet_aton.
347
348 4.11 Thu Oct 11 02:40:24 CEST 2007
349 - port to threaded perls.
350
351 4.1 Thu Oct 11 02:38:16 CEST 2007
352 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
353 be local to each coro (see Coro::State).
354 - incompatible change: for very deep reasons, cede and cede_notself
355 cannot return anything, so nothing will be returned.
356 - possibly bring back 5.10 compatibility (untested).
357 - work around stupid (and wrong) warning on 5.10 :(.
358 - overlay the saved state over the context stack. This saves
359 a few hundred bytes per coroutine on average and also
360 speeds up context switching a bit.
361 - further tune default stack sizes.
362 - (more) correctly calculate stack usage in coro_rss.
363 - Coro::Storable::blocking_* did not properly lock
364 resulting in races between coroutines.
365 - added Coro::Storable::guard.
366 - stopping to trace a coroutine could destroy the cctx of
367 an unrelated coroutine.
368 - explain the relationship between Perl and C coroutines in
369 more detail in Coro::State.
370 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
371 causing a fork per resolve. This also affected Coro::Socket.
372 - switch to a separate stack in $coro->call/eval to avoid
373 invalidating pointers.
374
375 4.03 Sat Oct 6 21:24:00 CEST 2007
376 - added Coro::throw method.
377 - minor code cleanups.
378
379 4.02 Sat Oct 6 02:36:47 CEST 2007
380 - fix a very minor per-coroutine memleak (a single codereference).
381 - fixed a bug where the currently in-use c context would be freed
382 prematurely (can happen only when programs change the stacksize
383 or use tracing).
384 - tracing can no longer keep a coro alive after it terminated.
385 - do static branch prediction in the common path for gcc. gives
386 about 2-5% speed improvement here.
387
388 4.01 Fri Oct 5 22:10:49 CEST 2007
389 - instead of recreating *a* standard output handle we simply
390 use STDOUT, which is faster and hopefully more robust.
391
392 4.0 Fri Oct 5 12:56:00 CEST 2007
393 - incompatibly changed Coro::Storable::freeze.
394 - major new feature: added Coro::Debug, for interactive coroutine
395 debugging, tracing and much more.
396 - major bug fix: unbelievable, but true: $_, $/ and many other
397 "saved" variables actually weren't being saved. This has been fixed,
398 of course, while increasing performance while losing all the save
399 flags.
400 - save flags are gone, and all the api functions dealing with them.
401 - added Coro::Semaphore::adjust.
402 - added Coro::Util::fork_eval.
403 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
404 - added Coro::killall.
405 - reduce initial stack sizes to allow for "micro-coroutines".
406 - better async_pool resource management, moved parts of async_pool
407 handling to XS (major speed improvement).
408 - actually croak before modifying important data structures.
409 - refuse to transfer while compiling.
410 - possibly support eval EXPR better now.
411 - enable assembly per default on linux+bsd x86+amd64.
412 - all internal members were renamed _something for easier subclassing.
413 - many minor tweaks.
414
415 3.63 Wed May 16 14:10:06 CEST 2007
416 - implement handcoded assembly for x86/amd64 SVR ABI.
417
418 3.62 Fri Apr 27 21:36:06 CEST 2007
419 - upgrade libcoro (which might set unwind info correctly).
420 - change default on linux to setjmp/longjmp.
421
422 3.61 Thu Apr 19 12:36:18 CEST 2007
423 - Coro::Storable caused an endless loop when thawing invalid
424 pst's sometimes.
425 - use a Semaphore in Coro::Storable, as Storable doesn't
426 seem to be reentrant (although it is documented to
427 be threadsafe...).
428 - fix Coro::Signal to bring back the original unreliable
429 but stateful semantics.
430 - fixed a lot of typos in Coro.pm (patch submitted by David
431 Steinbrunner, which applied flawlessly).
432
433 3.6 Sat Apr 14 17:13:31 CEST 2007
434 - added some bugfixes to get eg/myhttpd working again.
435 - added Coro::Storable for often-cede'ing freeze/thaw.
436 - try to do a clean exit when a coroutine calls exit
437 (EXPERIMENTAL).
438 - got rid of indirect call through _coro_init.
439 - updated the partly antique examples in eg/ to
440 work again and be a bit less magic, too.
441 - fixed Coro::Signal semantics to work as documented again.
442
443 3.55 Sun Mar 25 01:20:47 CET 2007
444 - add SAVE_DEFFH to save the default filehandle and enable
445 it by default.
446 - finally move socket-operations from Coro::Socket to Coro::Handle
447 to be able to unblock foreign sockets.
448 - add Coro::State::save_also and guarded_save.
449 - add count accessor to Coro::Semaphore.
450 - add Coro::State::cctx_stacksize.
451 - just for the fun of it, do not rely on implicit context,
452 which can dramatically improve performance, but people
453 using windows-process-emulation perls do not care much about
454 performance.
455
456 3.51 Sun Mar 4 14:18:04 CET 2007
457 - fixed a problem when you weakened references to Coro::State's
458 (patch by Yuval Kogman).
459
460 3.501 Wed Feb 28 12:44:07 CET 2007
461 - rename some global symbols as macosx from hell redefines
462 them without asking.
463
464 3.5 Tue Feb 13 20:22:53 CET 2007
465 - do AnyEvent model detection earlier, avoiding problems
466 caused by first using AnyEvent and later Coro::Event.
467 - implement and document Coro::Event event objects.
468 - fix a potential problem in Coro::Event causing crashes.
469 - initialise PL_comppad when creating a new coroutine,
470 avoids crashes on early coro destruction.
471
472 3.41 Mon Jan 22 19:19:49 CET 2007
473 - readline on Coro::Handle did not support undefined $/,
474 nor did it deliver partial lines on EOF or error.
475 - implement malloc fallback for stack allocation because
476 stupid broken idiotic OSX has a stupid broken
477 idiotic fits-the-whole-os mmap "implementation" and
478 my dick feels longer if Coro is portable even to
479 obsolete platforms.
480
481 3.4 Fri Jan 19 21:52:54 CET 2007
482 - remove t/09_timer.t, as it isn't really testing much
483 but was rather flaky in practise.
484 - async_pool coro would keep arguments and callback alive until
485 it was reused.
486 - cancellation of a coroutine could cause spurious idle calls
487 in cede_notself.
488
489 3.3 Sat Jan 6 03:45:00 CET 2007
490 - implement $coro->on_destroy.
491 - Coro::Event blocking semantics have been changed,
492 documented and - hopefully - improved.
493 - fix nice adding, not subtracting, from priority.
494 - fix ->prio and api_is_ready (patch by Mark Hinds).
495 - fixed an assert ("... == prev__cctx->idle_te")
496 that could errronously trigger.
497 - fix various large and small memleaks.
498 - use a (hopefully) more stable cancel implementation
499 that immediately frees the coroutine data.
500 - cede/cede_notself return a status now.
501 - added Coro::guard function.
502 - added a global coroutine pool for jobs (on my machine,
503 I can create and execute 48k simple coros/s with async,
504 and 128k coros with async_pool).
505 - Coro::AIO now uses the coroutine priority as io priority.
506
507 3.2 Fri Dec 22 05:07:09 CET 2006
508 - improve portability to slightly older perls.
509 - use cleaner coroutine destruction.
510 - simplify configuration for users.
511 - optionally (unrecommended) prefer perl functions over
512 their coro replacements.
513
514 3.11 Tue Dec 5 13:11:24 CET 2006
515 - fixed some bogus assert's, but as perl.h disables assert even
516 without NDEBUG (thank you very much), not too many people should
517 notice (that did include myself). Andreas König noticed, though :)
518 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
519 it. Also noticed by Andreas König :)
520 - save/restore tainted status.
521 - verified to pass the testsuite on my 5.9.5.
522
523 3.1 Mon Dec 4 23:03:40 CET 2006
524 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
525 - incompatible change: transfer flags are now per-state.
526 - give a better error message on deadlock.
527 - document Coro::nready.
528 - enhanced testsuite.
529
530 3.01 Sun Dec 3 23:47:42 CET 2006
531 - forgot to include Coro::Timer.
532
533 3.0 Sun Dec 3 22:57:25 CET 2006
534 - the "FINALLY COMPLETELY STABLE" release (coming soon:
535 the "FAMOUS LAST WORDS" release).
536 - implement a new stack sharing algorithm, which uses a stack
537 pool (size currently hardcoded).
538 - make stack sharing mandatory (it no longer uses a heuristic).
539 - eval/die no longer cause weird problems under heavy use.
540 - Coro::Event could cause livelocks if it was used but
541 no Coro::Event watchers were used.
542 - Coro::Event now uses asynccheck as prepare does not
543 check for changed watchers.
544 - Coro::Event allows multiple waiting coros and will wake up one
545 per event.
546 - Coro::Event should be cleaner and more efficient now.
547 - new utility function Coro::unblock_sub.
548 - document the sad fact that Event is no longer reentrant.
549 - putting a coroutine into the ready queue twice could under
550 some circumstances lead to stack corruption.
551 - minor incompatibility: subclassing Coro::State is supported
552 directly now without going through a _coro_state member.
553 - state/coro switching is much faster now.
554 - very minor optimisations and code/documentation cleanup.
555 - avoid problems due to compiler inlining.
556 - removed timers from Coro::Timer -> use AnyEvent instead.
557 - replaced Coro::idle coroutine by (cleaner) idle handler.
558 - updated to newest libcoro.
559 - implement enhanced support for valgrind.
560 - implement is_ready and return value for ready.
561 - removed Coro::Cont, it was a misnomer (it's generators),
562 and it was somewhat annoying to get it right. Will come back
563 if somebody asks for it :)
564 - many ->wait methods and Coro::Event could return
565 spuriously without the event having happened.
566
567 2.5 Tue Nov 7 12:22:33 CET 2006
568 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
569 use AnyEvent, moved them to Coro/.
570 - added Coro::LWP which contains all the uglyness required to
571 make LWP non-blocking.
572 - should work with perl 5.9.x now (Andreas König made me do it).
573 - fixed another bug in Coro::Select when the timeout was undef.
574 - reuse PL_start_env for all coros, saves some memory per coroutine.
575 - manage PL_top_env differently, hopefully to avoid panic: top_env.
576 - timeout argument was not properly used in Coro::Socket.
577 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
578 - emulate undocumented(!) functionality of IO::Socket required
579 by LWP(!!).
580 - updated eg/lwp to work with newer lwp's.
581 - remove "FATAL: uncaught exception" prefix. Coroutines that die
582 kill the whole process, just as exceptions in the main "coroutine"
583 did already.
584
585 2.1 Wed Nov 1 23:01:13 CET 2006
586 - fix a long-standing bug in Coro::Select where select with
587 zero timeout would instead change the current default filehandle.
588 - use a simpler and hopefully more robust way to clone padlists
589 (uses less memory and a perl function instead of our own).
590 - coro can now create a stack guard area on many architectures.
591 - Coro::AIO properly reexports additional functions from IO::AIO.
592 - updated libcoro with a workaround for OS X,
593 pach and testing by Michael Schwern.
594
595 2.0 Tue Oct 24 05:47:17 CEST 2006
596 - support additional aio requests in Coro::AIO.
597
598 1.9 (never properly released due to a glitch)
599
600 1.8 Thu Feb 2 00:59:06 CET 2006
601 - applied suggested patch by SAMV to avoid problems during stupid
602 mark & sweep gc run.
603 - applied patch by Scott Walters for 5.9.3 compatibility.
604
605 1.7 Tue Dec 27 01:41:58 CET 2005
606 - added Coro::AIO, a thin wrapper around IO::AIO.
607 - improved Makefile.PL explanations.
608
609 1.6 Mon Dec 12 21:30:05 CET 2005
610 - additionally save PM_curpm might fix as-of-yet
611 unseen problems with regex matches being attributed to the
612 wrong package.
613 - add t/10_bugs.t, which currently checks against imho broken
614 perls that use 0.26MB of stack space per Perl_magic_get
615 invocation (newer linuxes) as opposed to the 0.0002MB perl
616 normally uses.
617 - make stacksize configurable for the ultimate debian experience.
618
619 1.51 Mon Dec 12 18:48:36 CET 2005
620 - remove debugging warn() accidentally left in Coro::Select.
621
622 1.5 Tue Nov 29 13:32:44 CET 2005
623 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
624 - fix Coro::Select 'select'.
625 - strict'ify some modules.
626 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
627 - use XSLoader in selected modules.
628 - remove some 5.6 compatibility cruft.
629
630 1.4 Tue Sep 6 00:11:05 CEST 2005
631 - libcoro did not take into account the trampoline on amd64,
632 when 'l'inux method was used.
633
634 1.31 Tue Aug 30 23:31:33 CEST 2005
635 - some portability fixes/workarounds.
636
637 1.3 Sat Aug 20 03:08:56 CEST 2005
638 - no code changes. module seems to work fine.
639 - improve Coro::State docs, remove reference to nonexisting
640 Coro::State::flush method.
641 - no longer autodetect windows, present it as an option instead.
642
643 1.2 Mon May 16 02:00:55 CEST 2005
644 - included libcoro.c earlier in State.xs, might improve portability.
645 - use a faster and possibly more stable Coro::Cont implementation.
646 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
647 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
648 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
649
650 1.11 Thu Mar 3 18:00:52 CET 2005
651 - change of contact address.
652
653 1.1 Tue Feb 22 20:51:16 CET 2005
654 - support [l]inux method on amd64.
655 - allow some context switches while compiling/eval'ing.
656
657 1.0 Tue Aug 10 03:47:27 CEST 2004
658 - Event 0.89 fixes the bug again.
659 - fix the remaining(??) memleak. I didn't free the cache-AV for
660 padlists, so one AV per created anon-sub leaked.
661
662 0.97 Fri May 14 15:23:32 CEST 2004
663 - remove stupid gcc extension that was an experiment only anyway.
664
665 0.96 Thu May 13 18:09:29 CEST 2004
666 - tell people if they have yet another broken Event variant.
667 - made Coro to work _better_ with threads (it seems that, as long
668 as you use Coros in one thread only everything will work fine.
669 Looking at the perl source the problem might be missing locking
670 between Coro and some perl internal routines. That's not easily
671 fixable).
672 - (hopefully) got rid of the static xnext inside transfer.
673
674 0.95 Sun Feb 15 03:20:28 CET 2004
675 - removed Coro::State::flush, implemented a new and much
676 faster caching mechanism by attaching magic to the CV.
677 - a memleak with (real) closures remains.
678 - dramatically reduced size of stacks to 65536 (linux/x86), after
679 observing that even myhttpd never had more than a kilobyte
680 of stack in use (stack is still only physically allocated in
681 pagesize increments for systems suporting mmap).
682 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
683 - fix for threaded perl.
684
685 0.9 Sun Nov 23 04:32:39 CET 2003
686 - fix a few very hard to track down but (of course) stupid
687 memory leaks.
688 - stringify version number to avoid locale problems :(.
689 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
690 hopefully ignoring it "just works".
691
692 0.8 Wed Nov 5 19:38:40 CET 2003
693 - port to perl5.005_03, but only 5.8.x+ is supported!
694 - honor LocalAddr even without LocalPort in Coro::Socket.
695
696 0.7 Tue May 27 03:12:38 CEST 2003
697 - the version jump indicates some level of testing, not gobs
698 of new features.
699 - uh... I found the prompt function in ExtUtils::MakeMaker.
700 Highly correct stuff, that is...
701 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
702 the fix is not well understood by the author, so beware :(.
703
704 0.652 Thu May 8 02:54:46 CEST 2003
705 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
706 cause version 4 doesn't have ucontext.
707 - Benjamin Reed reported that setjmp works fine on darwin,
708 so preselect it.
709
710 0.651 Sat Mar 29 15:00:23 CET 2003
711 - fix a bug in Coro::Handle where some bytes could get lost
712 on reading, reported by jason@nichego.net.
713
714 0.65 Sun Mar 23 00:08:26 CET 2003
715 - added a README.linux-glibc.
716 - new module Coro::Select.
717 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
718 (I never claimed that I know what I am doing ;).
719 - set default to setjmp/longjmp on non-x86-linux.
720
721 0.6 Thu Nov 21 11:09:06 CET 2002
722 - made lazy context switching the default.
723 - fixed the bug where SemaphoreSet::guard
724 was the same as timed_guard and timed_guard was missing.
725 - fix a memleak in Coro::Timer.
726 - reclassified context sharing as not experimental.
727
728 0.534 Sun Apr 14 03:05:12 CEST 2002
729 - fixed backspace => tab in header parsing (myhttpd).
730 - added eg/readline, for elmex.
731 - Coro::Event now calls ->start on first call instead of
732 ->again, which created wrong timeouts for the first ->next.
733 - fixed a bug where Coro::Socket returned a closed socket instead
734 of nothing on ECONNREFUSED and similar error conditions.
735
736 0.533 Mon Feb 18 18:49:40 CET 2002
737 - adapt to finally working Event-0.85.
738 - fixed buggy event initialization (trapped under windows, but how
739 could it possibly have worked anywhere?)
740
741 0.532 Wed Jan 16 02:45:32 CET 2002
742 - removed the reference to pp_entersub, might work on
743 (some) win32 perls now (testsuite works under cygwin,
744 without Event, which freezes).
745
746 0.531 Mon Dec 10 22:18:44 CET 2001
747 - Fixed a bug in SemaphoreSet::timed_down.
748
749 0.53 Tue Nov 27 21:11:13 CET 2001
750 - some tests for Event.
751 - slightly different internal architecture to get refcnt into the object
752 and not the reference. might fix a few bugs, certainly did introduce
753 new ones.
754 - Coro::Timer (independent of Event).
755 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
756
757 0.52 Tue Nov 6 21:36:18 CET 2001
758 - ported to cygwin (trivial).
759 patch by Gerrit P. Haase <gp@familiehaase.de>.
760 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
761
762 0.51 Thu Nov 1 20:39:01 CET 2001
763 - terminate/cancel now work properly (otherwise termination
764 could cause a "next coroutine is not and contains not..."
765 error.
766 - added Coro::Socket::shutdown.
767 - Coro::Event::loop no is the same as Event::loop.
768 - implemented terminate with args + join.
769
770 0.5 Fri Sep 28 16:15:35 CEST 2001
771 - fixed "print" on a Coro::Handle. The print method worked.
772 - small tweaks (seem to reduce memory consumption a lot)
773 in various modules.
774 - splendid use of "no warnings" scattered throughout.
775 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
776
777 0.49 Sun Sep 16 02:42:45 CEST 2001
778 - changed some method calls to function calls for speed
779 inside Coro::Handle.
780 - make Coro::Handle use an array instead of a hash for
781 speed reasons.
782 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
783 sjlj and ucontext should now work.
784 - IRIX-specific port for libcoro.
785 - swapped order of accept results to match IO::Socket.
786 - changed getsock/peername to sock/peername to match IO::Socket.
787 - fixed a bug that caused segfault when returning to main task
788 under some circumstances.
789 - other bugfixes.
790
791 0.45 Sun Sep 2 02:54:01 CEST 2001
792 - new method Coro::Handle::timeout.
793 - corrected speling of set...name to get...name in Coro::Socket.
794 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
795 - Coro::Handle now supports fileno and FILENO.
796 - added eg/myhttpd, a web server.
797 - Coro::Socket now accepts numeric ports as well as "port(number)"
798 syntax.
799 - moved some scheduling primitives into xs code. more to come.
800 - new simple priority system.
801 - implemented Coro::Event using XS, almost four times faster!
802 - small memory corruption problem fixed (boy that was difficult).
803
804 0.13 Wed Aug 8 16:53:07 CEST 2001
805 - new method Coro::Handle::readline.
806 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
807 (see eg/lwp on how to make LWP non-blocking).
808 - renamed Coro::Socket::new_inet to new.
809 - added Coro::Util, some utility functions.
810 - die/eval now works better (still get annoying (but true ;)
811 "Callback called exit" messages).
812 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
813 - much better Coro::Socket support.
814
815 0.12 Fri Jul 27 04:19:01 CEST 2001
816 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
817 - very experimental stack sharing algorithm. not 100% safe but
818 should work well in practise ;)
819 - added Coro::RWLock.
820 - Coro::Specific now works.
821
822 0.11 Tue Jul 24 22:49:21 CEST 2001
823 - added specialized hack for newer and older linux versions (fast).
824 - renamed Coro::Event::IO to Coro::Handle.
825 - new module Coro::Socket.
826
827 0.1 Tue Jul 24 01:47:53 CEST 2001
828 - release candidate 3. A rather new internal structure :(
829 - the great renaming: Damian Conway gave me a suitable replacement
830 for yield.
831 - added Coro/libcoro, a portable coroutine implementation for C
832 (not even perl-dependent!!!), which can be used by Coro::State.
833 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
834 perl-5.6.
835 - new function Coro::Event::idle.
836 - the idle process is now overriden
837 by default in Coro::Event.
838 - Coro::Channel now enforces the size.
839 - canceling events now works in all cases.
840 - Coro state now include $_ and $@.
841 - yet another bug workaround that I do not really understand :(
842 - new module Coro::Event::IO (very undocumented) to do
843 non-blocking i/o.
844 - performance tuning.
845
846 0.08 Thu Jul 19 06:13:25 CEST 2001
847 - release candidate 2 (new functionality).
848 - add Coro::State::flush function.
849 - transfer now supports a flags value (mostly for speed ATM).
850 - might compile and work in the presence of threads now.
851 - continuations have a saner syntax.
852 - no more memleaks.
853
854 0.07 Tue Jul 17 17:40:18 CEST 2001
855 - release candidate 1 ;)
856 - slightly nicer code.
857 - fixed a scheduling bug in Coro::Event.
858
859 0.06 Tue Jul 17 04:23:24 CEST 2001
860 - ok, I found the showstopper - the same sub must not be
861 re-entered in two different coroutines, otherwise => crash. I
862 see no easy solution to this problem, except by walking the call
863 chaing and saving/restoring the cv's, which is what I do now.
864 - memory leaks still latent, especially at thread termination.
865 - Coro::Event now works (haha).
866
867 0.05 Sun Jul 15 17:32:20 CEST 2001
868 - fixed a few issues in Coro::Event.
869 - I forgot to include Coro::Event in 0.04 :(:(:(
870
871 0.04 Sun Jul 15 05:24:59 CEST 2001
872 - @_ is now properly localized.
873 - Coro::State is now easier subclassable.
874 - Coro::Cont now coroutine-aware.
875 - Coro::Specific is a low-overhead module to create
876 coroutine-specific vars.
877 - Coro::Event provides a simple interface to Event.
878
879 0.03 Fri Jul 13 14:51:52 CEST 2001
880 - transfer() now implemented in XS (beware).
881 - new module Coro::Cont for really faked continuations.
882 - big internal architecture changes: Coro::State is now
883 really low-level and can thus be used to implement other
884 interesting things, While "Coro::" implements a process-like
885 model. Still crude and subject to change.
886 - $_ and $@ are no longer being localized.
887
888 0.02 Tue Jul 10 01:38:17 CEST 2001
889 - implemented "async" attribute.
890 - $_ and $@ are now localized.
891 - added Coro::Channel.
892 - more testcases, still no docs.
893
894 0.01 Tue Jul 3 02:18:41 CEST 2001
895 - original version; copied from Convert::Scalar.
896