ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.441
Committed: Wed Nov 19 11:57:55 2008 UTC (15 years, 6 months ago) by root
Branch: MAIN
Changes since 1.440: +2 -0 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: testcancel
9 TODO: guard as SLF - to avoid deadlocks.
10 - avoid deadlock in Coro::Channel if maxsize == 1 (Richard Hundt).
11
12 5.0
13 - NEW ARCHITECTURE: use the latest 4.x version if you experience
14 stability issues.
15 - bump API version to 7 - all dependents must be recompiled.
16 - removed timed_* functions - they were not being used anyways
17 and should be replaced by a more generic mechanism -
18 and were annoying to support anyways :)
19 - removed SemaphoreSet's waiter method - use sem instead.
20 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
21 - async_pool did free a scalar value twice
22 ("Attempt to unreference...").
23 - fix a disastrous bug in the readline optimisation
24 introduced in 4.801.
25 - fix a longstanding bug where calling terminate on a coro that
26 was waiting for a semaphore that was just becoming available
27 would cause a deadlock (semaphore would get into a state where
28 it was available but waiters were still blocked).
29 - calling throw on a coroutine that is waiting for a semaphore will
30 no longer make it acquire the semaphore (and thus leak a count).
31 - perl's process emulation is now not even theoretically supported
32 anymore.
33 - new functions Coro::rouse_cb and Coro::rouse_wait.
34 - new function Coro::Semaphore::wait.
35 - use named constants in Coro::Channel (Richard Hundt).
36 - directly patch the entersub opcode calling SLF functions (cede,
37 transfer and so on). this does speed up context switching, but
38 more importanly, it frees us from the hardcoded behaviour of
39 entersub, so we might actually be able to return something from
40 those functions and atcually create new ones.
41 - take advantage of __builtin_frame_address on gcc.
42 - expose THX in coroapi (not sure whether this was a wise decision,
43 as "threaded" perls are running at half speed anyways).
44 - implement execute_slf (schedule-like-function) interface that makes
45 it possible to implement schedule-like-functions in XS.
46 - use new SLF interface to massively speed up Coro::EV by roughly a
47 factor of two.
48 - used new SLF interface to massively speed up Coro::Semaphore by a
49 factor of three.
50 - used new SLF interface to speed up Coro::AIO by roughly a factor of
51 four and reduce its memory usage considerably.
52 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
53 for a nice speedup and vastly more correct behaviour. Also implement
54 a new method "sem" to get at the underlying semaphore object.
55 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
56 (in comparison) 20-40% speedup.
57 - used new SLF interface to reimplement Coro::Signal gaining
58 some unknown (because I was too lazy), but certain, speedup, and also
59 making signals reliable for the first time.
60 - forgot to include Coro/libcoro/README in the dist for all these years.
61 - work around a freebsd pthreads bug (manual testcancel is required as
62 pthread_cond_wait isn't a cancellation point on freebsd).
63 - use new rouse functions to speed up and simplify Coro::BDB.
64 - make "prefer perl native functions" work with threaded perls.
65 - condense Coro::Debug ps output, hint at v and w flags.
66 - (libcoro) lots of minor cleanups and portability improvements.
67
68 4.913 Sat Nov 15 07:58:28 CET 2008
69 - async_pool did free a scalar value twice
70 ("Attempt to unreference...").
71
72 4.912 Thu Nov 13 18:31:23 CET 2008
73 - minor cleanups.
74 - use much larger stacks on linux and perl < 5.8.8.
75 - Coro::Debug::new_unix_server did not unlink the socket
76 when destroyed.
77
78 4.911 Tue Nov 11 04:26:17 CET 2008
79 - "port" to threaded perls.
80
81 4.91 Mon Nov 10 05:36:38 CET 2008
82 - the ->throw exception object no longer leaks.
83 - creating a new cctx leaked a scopestack entry (memleak).
84 - new coroutines didn't get created with a zero flags field
85 (unknown impact).
86 - calling ->throw on a not-yet-started coroutine should now work
87 instead of being ignored.
88 - ->throw is now supported on Coro::State objects.
89 - clean up cctx creation code a bit.
90 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
91
92 4.9 Sat Nov 8 17:45:27 CET 2008
93 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
94 - (libcoro) no longer leak threads in the experimental pthread backend,
95 also speed it up considerably.
96 - (libcoro) do not rely on makecontext passing void *'s unscathed.
97 - fix compiletime dependencies on libcoro in the Makefile.
98 - cctx_count wasn't always updated properly.
99 - Coro::State::cctx_stacksize wasn't applied correctly.
100 - new function Coro::State::cctx_max_idle.
101 - the default max number of idle C contexts is now 4.
102 - (libcoro) try harder to get _setjmp/_longjmp.
103 - (libcoro) cleanup and extend the libcoro API to officially
104 allow the creation of empty source contexts.
105 - very experimental workaround for the totally broken netbsd platform.
106 - tried to hack around openbsd bugs.
107
108 4.804 Wed Nov 5 16:36:00 CET 2008
109 - Coro::Debug::new_unix_server would not create a non-blocking listening
110 socket, sometimes causing freezes.
111 - (libcoro) fix misaligned stack points for setjmp and assembly
112 methods, which can cause crashes on x86/x86_64 with a sufficiently
113 aggressive compiler.
114 - new function: Coro::Debug::new_tcp_server.
115 - move ->throw into the Coro class because it only works on coro objects.
116
117 4.803 Mon Nov 3 17:16:12 CET 2008
118 - (libcoro) use a global asm statement to become independent of gcc
119 otpimisations for CORO_ASM (thanks to pippijn for the idea).
120 - try to workaround yet another broken bsd, this time dragonfly.
121
122 4.802 Thu Oct 30 10:56:12 CET 2008
123 - support -fno-omit-frame-pointer on x86 with the assembly method.
124 - tune 01_unblock.t tests a bit.
125
126 4.801 Wed Oct 22 18:33:37 CEST 2008
127 - improve readline speed for very long "lines".
128 - backport to 5.8.8.
129
130 4.8 Thu Oct 2 13:34:40 CEST 2008
131 - new function Coro::AIO::aio_wait.
132 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
133 - greatly speed up and reduce memory usage of Coro::AIO requests.
134 - implement some other µ-optimisations.
135
136 4.749 Mon Sep 29 14:40:12 CEST 2008
137 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
138
139 4.748 Sat Sep 27 14:03:19 CEST 2008
140 - implement, but do not document, PerlIO::cede(granularity).
141 - Coro::Storable forgot to wrap Storable::pstore.
142 - work around the multitude of leaks and memory corruption
143 bugs in PerlIO::via by using our own C-level perliol. As a side
144 effect, Coro::Storable is now much, much, much faster.
145
146 4.747 Tue Sep 23 01:59:41 CEST 2008
147 - fix a per-cv memleak (one empty array was leaked per
148 code reference).
149 - avoid a crash in coro->call|rss when the coroutine was already
150 destroyed (most noticably when using Coro::Debug::ps :)
151 - also protect *Storable::FILE.
152 - push up default storable granularity to 20ms.
153
154 4.746 Sun Sep 21 03:22:20 CEST 2008
155 - be more insistent on locking Storable against reentrancy
156 in Coro::Storable.
157 - move swap_def?v and throw to Coro::State, as documented.
158
159 4.745 Thu Jul 24 00:14:38 CEST 2008
160 - remove debugging code related to MgPV_nolen_const, also try to
161 make it compile with perl 5.8.6 (yes, apple apparently loves
162 outdated software). Reported by John S.
163
164 4.744 Tue Jul 8 22:06:35 CEST 2008
165 - correctly provide default DIE/WARN handlers as documented.
166 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
167 implementations inside perl work fine for us.
168
169 4.743 Mon Jun 16 00:21:57 CEST 2008
170 - when using Coro::EV without running EV::loop it could
171 result in busy-waiting for events, this has been fixed.
172 - reduce codesize and improve performance by using EV_DEFAULT_UC.
173
174 4.742 Sat May 31 14:10:21 CEST 2008
175 - implement a workaround for (some) perl <5.8.8 versions.
176 - require EV 3.3+.
177
178 4.741 Fri May 30 23:33:09 CEST 2008
179 - tell netbsd how utterly broken their imitation of an OS is
180 and refuse to build by default if pthreads are in use.
181 - switch to "s" method on all bsd's by default, as their ucontext
182 stuff seems just too broken.
183 - fix a bug in Coro::Select.
184
185 4.74 Thu May 29 20:05:31 CEST 2008
186 - do not test Coro::LWP for lack of dependencies.
187 - do not test Coro::Select for lack of working perls.
188
189 4.73 Thu May 29 2008
190 - fix a bug in Coro::EV which would cause it to block despite
191 there being runnable coroutines.
192 - sprinkle "no warnings" freely over everything, also suppress
193 warnings for some other modules.
194 - fix typo in WSAEWOULDBLOCK.
195
196 4.72 Sun May 25 05:14:36 CEST 2008
197 - tweak META.yaml a bit, unfortunately, there is no documented way
198 to have optional dependencies with CPAN. doh :(
199 - avoid running some tests on windows because they would fail due to
200 perl bug (broken fork, broken pipes...).
201 - work around perl on windows bugs where perl returns undocumented
202 error codes for sysread, syswrite etc. by taking advantage
203 of AnyEvent's workaround for that problem.
204 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
205 a common perl implementation bug on windows.
206 - use unix domain sockets in testsuite to work around
207 common perl implementation bugs on widows (they are emulated by
208 tcp sockets on windows. ugh.)
209
210 4.71 Sat May 24 20:01:27 CEST 2008
211 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
212 - take advantage of async name resolution of AnyEvent::Util.
213 - work around brutal inet_aton override in Coro::LWP.
214 - take advantage of the readyhook in Coro::EV, for smoother
215 scheduling.
216
217 4.7 Sun May 11 00:32:19 CEST 2008
218 - completely reworked the Coro manpage.
219 - added Coro::AnyEvent, generic event loop integration.
220 - implement cancel, ready and kill commands in Coro::Debug.
221 - document find_coro in Coro::Debug.
222 - incompatible change: rename has_stack to has_cctx.
223 - Coro::AIO and Coro::BDB no longer force event model detection,
224 use AnyEvent::AIO and AnyEvent::BDB.
225
226 4.6 Sat Apr 26 10:05:14 CEST 2008
227 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
228 by the Coro module. It was a mistake to have it in the first place.
229 - (experimental) support for activestate perl 5.10 (method "w").
230 - (experimental) support for strawberry perl (method "a").
231 - coro_sigelem_set did not return a value although it had to,
232 actual impact unknown.
233
234 4.51 Mon Apr 14 13:28:27 CEST 2008
235 - make it compile again on 5.8.
236
237 4.50 Thu Apr 10 09:43:17 CEST 2008
238 - I did it twice! (see 4.49).
239
240 4.49 Sun Apr 6 21:23:31 CEST 2008
241 - grr, instead of compiling the recent changes on 5.10 only they
242 were compiled on 5.8 only.
243
244 4.48 Sun Apr 6 20:36:46 CEST 2008
245 - allow coroutine switches during eval's under 5.10.x, as it seems
246 the parser is a per-interpreter option now, so this is safe
247 (this might fix the odd crashing bug).
248 - drop support for 5.9.x versions: they are dead, jim.
249
250 4.47 Sun Apr 6 00:37:52 CEST 2008
251 - force cctx allocation on API calls: we know we need to force one
252 and gcc actually manages to confuse our heuristic nowadays,
253 leading to crashes and worse.
254 - document force_cctx.
255
256 4.46 Fri Apr 4 22:05:43 CEST 2008
257 - upgrade libcoro, resulting in pthread-backend (which was only created
258 to fulfill the rules of the programming languages shootout).
259
260 4.45 Thu Mar 13 11:55:36 CET 2008
261 - fix a file leaking bug in eg/mhyttpd that would allow
262 downloading of any file (reported by oesi).
263 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
264 (also add testcase).
265 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
266 apparently having loads of fun with that).
267 - implement $state->swap_defsv, ->swap_defav and document ->throw.
268
269 4.4 Wed Feb 13 16:44:29 CET 2008
270 - only assume .cfi pseudo isns exist on GNU/Linux.
271 - add get_timed method to Coro::Channels.
272 - fixed Coro::Debug manpage.
273 - perl 5.11 compatbility improvement based on patch by
274 Andreas König.
275
276 4.37 Sun Jan 20 11:25:23 CET 2008
277 - soften the check for an active parser for perl 5.10 (avoiding
278 "Coro::State::transfer called while parsing" in many cases).
279
280 4.36 Sun Jan 13 10:53:56 CET 2008
281 - reset diehook when terminating from an async_pool as to not
282 trigger any __DIE__ handlers.
283
284 4.35 Sun Jan 13 04:14:13 CET 2008
285 - "bt" debug command now displays any exceptions
286 from longmess and also skips the innermost
287 stackframes, giving more useufl output.
288 - allow backtraces in coroutines blocked in Coro::EV,
289 at a <1% speed hit.
290 - handle localising of $SIG{__WARN__} and __DIE__
291 properly (with a proper amount of dirty hacking).
292
293 4.34 Sat Dec 22 17:49:53 CET 2007
294 - upgrade to EV version 2.0 API.
295
296 4.33 Mon Dec 17 08:36:12 CET 2007
297 - make Coro::AIO etc. loadable in the absence of EV.
298
299 4.32 Mon Dec 17 07:46:02 CET 2007
300 - majorly improved Coro::Handle's performance with Coro::EV.
301 - implemented the until now mythical Coro::BDB module.
302 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
303
304 4.31 Wed Dec 5 12:32:39 CET 2007
305 - remove warn statement form Coro::Util.
306
307 4.3 Tue Dec 4 20:33:14 CET 2007
308 - calls to the idle function could cause stack corruption
309 when the stack changed.
310 - do no longer rely on the presence of EV::DNS if EV is used
311 (because it is gone), but instead take avdantage of EV::ADNS
312 if available.
313 - add ($) prototypes to all functions in Coro::Storable.
314 - use a conventional (safer) idle callback in Coro::EV.
315 - do accept quickly in Coro::Debug to avoid endless loops.
316
317 4.22 Fri Nov 30 16:04:04 CET 2007
318 - really use optimised versions for Event and EV in Coro::Util
319 and Coro::Handle.
320
321 4.21 Sun Nov 25 10:48:59 CET 2007
322 - fix a spurious memory read.
323 - Coro::EV no longer keeps the eventloop "alive".
324
325 4.2 Fri Nov 9 20:47:05 CET 2007
326 - enable/disable tracing from a new coroutine, not a pooled one.
327 - fix a memleak in Coro::Event.
328 - removed killall call from fork_eval.
329 - made sure store_fd is already loaded so that fork_eval
330 does not have to parse autoload in each subprocess.
331 - only use assembly method if -O switch is in $Config{optimize}.
332 - add (optional) Coro::EV module, so far the best event loop module
333 directly supported by Coro.
334 - if the event model is EV, use EV::DNS to resolve
335 stuff in Coro::Util.
336 - don't get confused by multiple event notifications in Coro::Handle.
337 - initial support for EV (libevent interface).
338 - require Event and EV using configure_requires, to force their existance.
339
340 4.13 Wed Oct 24 07:26:45 CEST 2007
341 - add Coro::Storable::blocking_thaw.
342 - use a vastly more complicated technique to localise
343 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
344 - use a coroutine for the idle callback Coro::Event,
345 instead of running Event in the current coroutine context.
346 This also catches recursive invocations.
347 - actually report fork errors in gethostbyname and inet_aton.
348
349 4.11 Thu Oct 11 02:40:24 CEST 2007
350 - port to threaded perls.
351
352 4.1 Thu Oct 11 02:38:16 CEST 2007
353 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
354 be local to each coro (see Coro::State).
355 - incompatible change: for very deep reasons, cede and cede_notself
356 cannot return anything, so nothing will be returned.
357 - possibly bring back 5.10 compatibility (untested).
358 - work around stupid (and wrong) warning on 5.10 :(.
359 - overlay the saved state over the context stack. This saves
360 a few hundred bytes per coroutine on average and also
361 speeds up context switching a bit.
362 - further tune default stack sizes.
363 - (more) correctly calculate stack usage in coro_rss.
364 - Coro::Storable::blocking_* did not properly lock
365 resulting in races between coroutines.
366 - added Coro::Storable::guard.
367 - stopping to trace a coroutine could destroy the cctx of
368 an unrelated coroutine.
369 - explain the relationship between Perl and C coroutines in
370 more detail in Coro::State.
371 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
372 causing a fork per resolve. This also affected Coro::Socket.
373 - switch to a separate stack in $coro->call/eval to avoid
374 invalidating pointers.
375
376 4.03 Sat Oct 6 21:24:00 CEST 2007
377 - added Coro::throw method.
378 - minor code cleanups.
379
380 4.02 Sat Oct 6 02:36:47 CEST 2007
381 - fix a very minor per-coroutine memleak (a single codereference).
382 - fixed a bug where the currently in-use c context would be freed
383 prematurely (can happen only when programs change the stacksize
384 or use tracing).
385 - tracing can no longer keep a coro alive after it terminated.
386 - do static branch prediction in the common path for gcc. gives
387 about 2-5% speed improvement here.
388
389 4.01 Fri Oct 5 22:10:49 CEST 2007
390 - instead of recreating *a* standard output handle we simply
391 use STDOUT, which is faster and hopefully more robust.
392
393 4.0 Fri Oct 5 12:56:00 CEST 2007
394 - incompatibly changed Coro::Storable::freeze.
395 - major new feature: added Coro::Debug, for interactive coroutine
396 debugging, tracing and much more.
397 - major bug fix: unbelievable, but true: $_, $/ and many other
398 "saved" variables actually weren't being saved. This has been fixed,
399 of course, while increasing performance while losing all the save
400 flags.
401 - save flags are gone, and all the api functions dealing with them.
402 - added Coro::Semaphore::adjust.
403 - added Coro::Util::fork_eval.
404 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
405 - added Coro::killall.
406 - reduce initial stack sizes to allow for "micro-coroutines".
407 - better async_pool resource management, moved parts of async_pool
408 handling to XS (major speed improvement).
409 - actually croak before modifying important data structures.
410 - refuse to transfer while compiling.
411 - possibly support eval EXPR better now.
412 - enable assembly per default on linux+bsd x86+amd64.
413 - all internal members were renamed _something for easier subclassing.
414 - many minor tweaks.
415
416 3.63 Wed May 16 14:10:06 CEST 2007
417 - implement handcoded assembly for x86/amd64 SVR ABI.
418
419 3.62 Fri Apr 27 21:36:06 CEST 2007
420 - upgrade libcoro (which might set unwind info correctly).
421 - change default on linux to setjmp/longjmp.
422
423 3.61 Thu Apr 19 12:36:18 CEST 2007
424 - Coro::Storable caused an endless loop when thawing invalid
425 pst's sometimes.
426 - use a Semaphore in Coro::Storable, as Storable doesn't
427 seem to be reentrant (although it is documented to
428 be threadsafe...).
429 - fix Coro::Signal to bring back the original unreliable
430 but stateful semantics.
431 - fixed a lot of typos in Coro.pm (patch submitted by David
432 Steinbrunner, which applied flawlessly).
433
434 3.6 Sat Apr 14 17:13:31 CEST 2007
435 - added some bugfixes to get eg/myhttpd working again.
436 - added Coro::Storable for often-cede'ing freeze/thaw.
437 - try to do a clean exit when a coroutine calls exit
438 (EXPERIMENTAL).
439 - got rid of indirect call through _coro_init.
440 - updated the partly antique examples in eg/ to
441 work again and be a bit less magic, too.
442 - fixed Coro::Signal semantics to work as documented again.
443
444 3.55 Sun Mar 25 01:20:47 CET 2007
445 - add SAVE_DEFFH to save the default filehandle and enable
446 it by default.
447 - finally move socket-operations from Coro::Socket to Coro::Handle
448 to be able to unblock foreign sockets.
449 - add Coro::State::save_also and guarded_save.
450 - add count accessor to Coro::Semaphore.
451 - add Coro::State::cctx_stacksize.
452 - just for the fun of it, do not rely on implicit context,
453 which can dramatically improve performance, but people
454 using windows-process-emulation perls do not care much about
455 performance.
456
457 3.51 Sun Mar 4 14:18:04 CET 2007
458 - fixed a problem when you weakened references to Coro::State's
459 (patch by Yuval Kogman).
460
461 3.501 Wed Feb 28 12:44:07 CET 2007
462 - rename some global symbols as macosx from hell redefines
463 them without asking.
464
465 3.5 Tue Feb 13 20:22:53 CET 2007
466 - do AnyEvent model detection earlier, avoiding problems
467 caused by first using AnyEvent and later Coro::Event.
468 - implement and document Coro::Event event objects.
469 - fix a potential problem in Coro::Event causing crashes.
470 - initialise PL_comppad when creating a new coroutine,
471 avoids crashes on early coro destruction.
472
473 3.41 Mon Jan 22 19:19:49 CET 2007
474 - readline on Coro::Handle did not support undefined $/,
475 nor did it deliver partial lines on EOF or error.
476 - implement malloc fallback for stack allocation because
477 stupid broken idiotic OSX has a stupid broken
478 idiotic fits-the-whole-os mmap "implementation" and
479 my dick feels longer if Coro is portable even to
480 obsolete platforms.
481
482 3.4 Fri Jan 19 21:52:54 CET 2007
483 - remove t/09_timer.t, as it isn't really testing much
484 but was rather flaky in practise.
485 - async_pool coro would keep arguments and callback alive until
486 it was reused.
487 - cancellation of a coroutine could cause spurious idle calls
488 in cede_notself.
489
490 3.3 Sat Jan 6 03:45:00 CET 2007
491 - implement $coro->on_destroy.
492 - Coro::Event blocking semantics have been changed,
493 documented and - hopefully - improved.
494 - fix nice adding, not subtracting, from priority.
495 - fix ->prio and api_is_ready (patch by Mark Hinds).
496 - fixed an assert ("... == prev__cctx->idle_te")
497 that could errronously trigger.
498 - fix various large and small memleaks.
499 - use a (hopefully) more stable cancel implementation
500 that immediately frees the coroutine data.
501 - cede/cede_notself return a status now.
502 - added Coro::guard function.
503 - added a global coroutine pool for jobs (on my machine,
504 I can create and execute 48k simple coros/s with async,
505 and 128k coros with async_pool).
506 - Coro::AIO now uses the coroutine priority as io priority.
507
508 3.2 Fri Dec 22 05:07:09 CET 2006
509 - improve portability to slightly older perls.
510 - use cleaner coroutine destruction.
511 - simplify configuration for users.
512 - optionally (unrecommended) prefer perl functions over
513 their coro replacements.
514
515 3.11 Tue Dec 5 13:11:24 CET 2006
516 - fixed some bogus assert's, but as perl.h disables assert even
517 without NDEBUG (thank you very much), not too many people should
518 notice (that did include myself). Andreas König noticed, though :)
519 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
520 it. Also noticed by Andreas König :)
521 - save/restore tainted status.
522 - verified to pass the testsuite on my 5.9.5.
523
524 3.1 Mon Dec 4 23:03:40 CET 2006
525 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
526 - incompatible change: transfer flags are now per-state.
527 - give a better error message on deadlock.
528 - document Coro::nready.
529 - enhanced testsuite.
530
531 3.01 Sun Dec 3 23:47:42 CET 2006
532 - forgot to include Coro::Timer.
533
534 3.0 Sun Dec 3 22:57:25 CET 2006
535 - the "FINALLY COMPLETELY STABLE" release (coming soon:
536 the "FAMOUS LAST WORDS" release).
537 - implement a new stack sharing algorithm, which uses a stack
538 pool (size currently hardcoded).
539 - make stack sharing mandatory (it no longer uses a heuristic).
540 - eval/die no longer cause weird problems under heavy use.
541 - Coro::Event could cause livelocks if it was used but
542 no Coro::Event watchers were used.
543 - Coro::Event now uses asynccheck as prepare does not
544 check for changed watchers.
545 - Coro::Event allows multiple waiting coros and will wake up one
546 per event.
547 - Coro::Event should be cleaner and more efficient now.
548 - new utility function Coro::unblock_sub.
549 - document the sad fact that Event is no longer reentrant.
550 - putting a coroutine into the ready queue twice could under
551 some circumstances lead to stack corruption.
552 - minor incompatibility: subclassing Coro::State is supported
553 directly now without going through a _coro_state member.
554 - state/coro switching is much faster now.
555 - very minor optimisations and code/documentation cleanup.
556 - avoid problems due to compiler inlining.
557 - removed timers from Coro::Timer -> use AnyEvent instead.
558 - replaced Coro::idle coroutine by (cleaner) idle handler.
559 - updated to newest libcoro.
560 - implement enhanced support for valgrind.
561 - implement is_ready and return value for ready.
562 - removed Coro::Cont, it was a misnomer (it's generators),
563 and it was somewhat annoying to get it right. Will come back
564 if somebody asks for it :)
565 - many ->wait methods and Coro::Event could return
566 spuriously without the event having happened.
567
568 2.5 Tue Nov 7 12:22:33 CET 2006
569 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
570 use AnyEvent, moved them to Coro/.
571 - added Coro::LWP which contains all the uglyness required to
572 make LWP non-blocking.
573 - should work with perl 5.9.x now (Andreas König made me do it).
574 - fixed another bug in Coro::Select when the timeout was undef.
575 - reuse PL_start_env for all coros, saves some memory per coroutine.
576 - manage PL_top_env differently, hopefully to avoid panic: top_env.
577 - timeout argument was not properly used in Coro::Socket.
578 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
579 - emulate undocumented(!) functionality of IO::Socket required
580 by LWP(!!).
581 - updated eg/lwp to work with newer lwp's.
582 - remove "FATAL: uncaught exception" prefix. Coroutines that die
583 kill the whole process, just as exceptions in the main "coroutine"
584 did already.
585
586 2.1 Wed Nov 1 23:01:13 CET 2006
587 - fix a long-standing bug in Coro::Select where select with
588 zero timeout would instead change the current default filehandle.
589 - use a simpler and hopefully more robust way to clone padlists
590 (uses less memory and a perl function instead of our own).
591 - coro can now create a stack guard area on many architectures.
592 - Coro::AIO properly reexports additional functions from IO::AIO.
593 - updated libcoro with a workaround for OS X,
594 pach and testing by Michael Schwern.
595
596 2.0 Tue Oct 24 05:47:17 CEST 2006
597 - support additional aio requests in Coro::AIO.
598
599 1.9 (never properly released due to a glitch)
600
601 1.8 Thu Feb 2 00:59:06 CET 2006
602 - applied suggested patch by SAMV to avoid problems during stupid
603 mark & sweep gc run.
604 - applied patch by Scott Walters for 5.9.3 compatibility.
605
606 1.7 Tue Dec 27 01:41:58 CET 2005
607 - added Coro::AIO, a thin wrapper around IO::AIO.
608 - improved Makefile.PL explanations.
609
610 1.6 Mon Dec 12 21:30:05 CET 2005
611 - additionally save PM_curpm might fix as-of-yet
612 unseen problems with regex matches being attributed to the
613 wrong package.
614 - add t/10_bugs.t, which currently checks against imho broken
615 perls that use 0.26MB of stack space per Perl_magic_get
616 invocation (newer linuxes) as opposed to the 0.0002MB perl
617 normally uses.
618 - make stacksize configurable for the ultimate debian experience.
619
620 1.51 Mon Dec 12 18:48:36 CET 2005
621 - remove debugging warn() accidentally left in Coro::Select.
622
623 1.5 Tue Nov 29 13:32:44 CET 2005
624 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
625 - fix Coro::Select 'select'.
626 - strict'ify some modules.
627 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
628 - use XSLoader in selected modules.
629 - remove some 5.6 compatibility cruft.
630
631 1.4 Tue Sep 6 00:11:05 CEST 2005
632 - libcoro did not take into account the trampoline on amd64,
633 when 'l'inux method was used.
634
635 1.31 Tue Aug 30 23:31:33 CEST 2005
636 - some portability fixes/workarounds.
637
638 1.3 Sat Aug 20 03:08:56 CEST 2005
639 - no code changes. module seems to work fine.
640 - improve Coro::State docs, remove reference to nonexisting
641 Coro::State::flush method.
642 - no longer autodetect windows, present it as an option instead.
643
644 1.2 Mon May 16 02:00:55 CEST 2005
645 - included libcoro.c earlier in State.xs, might improve portability.
646 - use a faster and possibly more stable Coro::Cont implementation.
647 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
648 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
649 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
650
651 1.11 Thu Mar 3 18:00:52 CET 2005
652 - change of contact address.
653
654 1.1 Tue Feb 22 20:51:16 CET 2005
655 - support [l]inux method on amd64.
656 - allow some context switches while compiling/eval'ing.
657
658 1.0 Tue Aug 10 03:47:27 CEST 2004
659 - Event 0.89 fixes the bug again.
660 - fix the remaining(??) memleak. I didn't free the cache-AV for
661 padlists, so one AV per created anon-sub leaked.
662
663 0.97 Fri May 14 15:23:32 CEST 2004
664 - remove stupid gcc extension that was an experiment only anyway.
665
666 0.96 Thu May 13 18:09:29 CEST 2004
667 - tell people if they have yet another broken Event variant.
668 - made Coro to work _better_ with threads (it seems that, as long
669 as you use Coros in one thread only everything will work fine.
670 Looking at the perl source the problem might be missing locking
671 between Coro and some perl internal routines. That's not easily
672 fixable).
673 - (hopefully) got rid of the static xnext inside transfer.
674
675 0.95 Sun Feb 15 03:20:28 CET 2004
676 - removed Coro::State::flush, implemented a new and much
677 faster caching mechanism by attaching magic to the CV.
678 - a memleak with (real) closures remains.
679 - dramatically reduced size of stacks to 65536 (linux/x86), after
680 observing that even myhttpd never had more than a kilobyte
681 of stack in use (stack is still only physically allocated in
682 pagesize increments for systems suporting mmap).
683 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
684 - fix for threaded perl.
685
686 0.9 Sun Nov 23 04:32:39 CET 2003
687 - fix a few very hard to track down but (of course) stupid
688 memory leaks.
689 - stringify version number to avoid locale problems :(.
690 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
691 hopefully ignoring it "just works".
692
693 0.8 Wed Nov 5 19:38:40 CET 2003
694 - port to perl5.005_03, but only 5.8.x+ is supported!
695 - honor LocalAddr even without LocalPort in Coro::Socket.
696
697 0.7 Tue May 27 03:12:38 CEST 2003
698 - the version jump indicates some level of testing, not gobs
699 of new features.
700 - uh... I found the prompt function in ExtUtils::MakeMaker.
701 Highly correct stuff, that is...
702 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
703 the fix is not well understood by the author, so beware :(.
704
705 0.652 Thu May 8 02:54:46 CEST 2003
706 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
707 cause version 4 doesn't have ucontext.
708 - Benjamin Reed reported that setjmp works fine on darwin,
709 so preselect it.
710
711 0.651 Sat Mar 29 15:00:23 CET 2003
712 - fix a bug in Coro::Handle where some bytes could get lost
713 on reading, reported by jason@nichego.net.
714
715 0.65 Sun Mar 23 00:08:26 CET 2003
716 - added a README.linux-glibc.
717 - new module Coro::Select.
718 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
719 (I never claimed that I know what I am doing ;).
720 - set default to setjmp/longjmp on non-x86-linux.
721
722 0.6 Thu Nov 21 11:09:06 CET 2002
723 - made lazy context switching the default.
724 - fixed the bug where SemaphoreSet::guard
725 was the same as timed_guard and timed_guard was missing.
726 - fix a memleak in Coro::Timer.
727 - reclassified context sharing as not experimental.
728
729 0.534 Sun Apr 14 03:05:12 CEST 2002
730 - fixed backspace => tab in header parsing (myhttpd).
731 - added eg/readline, for elmex.
732 - Coro::Event now calls ->start on first call instead of
733 ->again, which created wrong timeouts for the first ->next.
734 - fixed a bug where Coro::Socket returned a closed socket instead
735 of nothing on ECONNREFUSED and similar error conditions.
736
737 0.533 Mon Feb 18 18:49:40 CET 2002
738 - adapt to finally working Event-0.85.
739 - fixed buggy event initialization (trapped under windows, but how
740 could it possibly have worked anywhere?)
741
742 0.532 Wed Jan 16 02:45:32 CET 2002
743 - removed the reference to pp_entersub, might work on
744 (some) win32 perls now (testsuite works under cygwin,
745 without Event, which freezes).
746
747 0.531 Mon Dec 10 22:18:44 CET 2001
748 - Fixed a bug in SemaphoreSet::timed_down.
749
750 0.53 Tue Nov 27 21:11:13 CET 2001
751 - some tests for Event.
752 - slightly different internal architecture to get refcnt into the object
753 and not the reference. might fix a few bugs, certainly did introduce
754 new ones.
755 - Coro::Timer (independent of Event).
756 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
757
758 0.52 Tue Nov 6 21:36:18 CET 2001
759 - ported to cygwin (trivial).
760 patch by Gerrit P. Haase <gp@familiehaase.de>.
761 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
762
763 0.51 Thu Nov 1 20:39:01 CET 2001
764 - terminate/cancel now work properly (otherwise termination
765 could cause a "next coroutine is not and contains not..."
766 error.
767 - added Coro::Socket::shutdown.
768 - Coro::Event::loop no is the same as Event::loop.
769 - implemented terminate with args + join.
770
771 0.5 Fri Sep 28 16:15:35 CEST 2001
772 - fixed "print" on a Coro::Handle. The print method worked.
773 - small tweaks (seem to reduce memory consumption a lot)
774 in various modules.
775 - splendid use of "no warnings" scattered throughout.
776 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
777
778 0.49 Sun Sep 16 02:42:45 CEST 2001
779 - changed some method calls to function calls for speed
780 inside Coro::Handle.
781 - make Coro::Handle use an array instead of a hash for
782 speed reasons.
783 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
784 sjlj and ucontext should now work.
785 - IRIX-specific port for libcoro.
786 - swapped order of accept results to match IO::Socket.
787 - changed getsock/peername to sock/peername to match IO::Socket.
788 - fixed a bug that caused segfault when returning to main task
789 under some circumstances.
790 - other bugfixes.
791
792 0.45 Sun Sep 2 02:54:01 CEST 2001
793 - new method Coro::Handle::timeout.
794 - corrected speling of set...name to get...name in Coro::Socket.
795 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
796 - Coro::Handle now supports fileno and FILENO.
797 - added eg/myhttpd, a web server.
798 - Coro::Socket now accepts numeric ports as well as "port(number)"
799 syntax.
800 - moved some scheduling primitives into xs code. more to come.
801 - new simple priority system.
802 - implemented Coro::Event using XS, almost four times faster!
803 - small memory corruption problem fixed (boy that was difficult).
804
805 0.13 Wed Aug 8 16:53:07 CEST 2001
806 - new method Coro::Handle::readline.
807 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
808 (see eg/lwp on how to make LWP non-blocking).
809 - renamed Coro::Socket::new_inet to new.
810 - added Coro::Util, some utility functions.
811 - die/eval now works better (still get annoying (but true ;)
812 "Callback called exit" messages).
813 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
814 - much better Coro::Socket support.
815
816 0.12 Fri Jul 27 04:19:01 CEST 2001
817 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
818 - very experimental stack sharing algorithm. not 100% safe but
819 should work well in practise ;)
820 - added Coro::RWLock.
821 - Coro::Specific now works.
822
823 0.11 Tue Jul 24 22:49:21 CEST 2001
824 - added specialized hack for newer and older linux versions (fast).
825 - renamed Coro::Event::IO to Coro::Handle.
826 - new module Coro::Socket.
827
828 0.1 Tue Jul 24 01:47:53 CEST 2001
829 - release candidate 3. A rather new internal structure :(
830 - the great renaming: Damian Conway gave me a suitable replacement
831 for yield.
832 - added Coro/libcoro, a portable coroutine implementation for C
833 (not even perl-dependent!!!), which can be used by Coro::State.
834 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
835 perl-5.6.
836 - new function Coro::Event::idle.
837 - the idle process is now overriden
838 by default in Coro::Event.
839 - Coro::Channel now enforces the size.
840 - canceling events now works in all cases.
841 - Coro state now include $_ and $@.
842 - yet another bug workaround that I do not really understand :(
843 - new module Coro::Event::IO (very undocumented) to do
844 non-blocking i/o.
845 - performance tuning.
846
847 0.08 Thu Jul 19 06:13:25 CEST 2001
848 - release candidate 2 (new functionality).
849 - add Coro::State::flush function.
850 - transfer now supports a flags value (mostly for speed ATM).
851 - might compile and work in the presence of threads now.
852 - continuations have a saner syntax.
853 - no more memleaks.
854
855 0.07 Tue Jul 17 17:40:18 CEST 2001
856 - release candidate 1 ;)
857 - slightly nicer code.
858 - fixed a scheduling bug in Coro::Event.
859
860 0.06 Tue Jul 17 04:23:24 CEST 2001
861 - ok, I found the showstopper - the same sub must not be
862 re-entered in two different coroutines, otherwise => crash. I
863 see no easy solution to this problem, except by walking the call
864 chaing and saving/restoring the cv's, which is what I do now.
865 - memory leaks still latent, especially at thread termination.
866 - Coro::Event now works (haha).
867
868 0.05 Sun Jul 15 17:32:20 CEST 2001
869 - fixed a few issues in Coro::Event.
870 - I forgot to include Coro::Event in 0.04 :(:(:(
871
872 0.04 Sun Jul 15 05:24:59 CEST 2001
873 - @_ is now properly localized.
874 - Coro::State is now easier subclassable.
875 - Coro::Cont now coroutine-aware.
876 - Coro::Specific is a low-overhead module to create
877 coroutine-specific vars.
878 - Coro::Event provides a simple interface to Event.
879
880 0.03 Fri Jul 13 14:51:52 CEST 2001
881 - transfer() now implemented in XS (beware).
882 - new module Coro::Cont for really faked continuations.
883 - big internal architecture changes: Coro::State is now
884 really low-level and can thus be used to implement other
885 interesting things, While "Coro::" implements a process-like
886 model. Still crude and subject to change.
887 - $_ and $@ are no longer being localized.
888
889 0.02 Tue Jul 10 01:38:17 CEST 2001
890 - implemented "async" attribute.
891 - $_ and $@ are now localized.
892 - added Coro::Channel.
893 - more testcases, still no docs.
894
895 0.01 Tue Jul 3 02:18:41 CEST 2001
896 - original version; copied from Convert::Scalar.
897