ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.420
Committed: Sun Nov 16 09:43:18 2008 UTC (15 years, 6 months ago) by root
Branch: MAIN
Changes since 1.419: +4 -0 lines
Log Message:
be good citizen

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