ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.406
Committed: Thu Nov 13 06:15:37 2008 UTC (15 years, 6 months ago) by root
Branch: MAIN
Changes since 1.405: +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 - minor cleanups.
9 - use much larger stacks on linux and perl < 5.8.8.
10 - Coro::Debug::new_unix_server did not unlink the socket
11 when destroyed.
12
13 4.911 Tue Nov 11 04:26:17 CET 2008
14 - "port" to threaded perls.
15
16 4.91 Mon Nov 10 05:36:38 CET 2008
17 - the ->throw exception object no longer leaks.
18 - creating a new cctx leaked a scopestack entry (memleak).
19 - new coroutines didn't get created with a zero flags field
20 (unknown impact).
21 - calling ->throw on a not-yet-started coroutine should now work
22 instead of being ignored.
23 - ->throw is now supported on Coro::State objects.
24 - clean up cctx creation code a bit.
25 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
26
27 4.9 Sat Nov 8 17:45:27 CET 2008
28 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
29 - (libcoro) no longer leak threads in the experimental pthread backend,
30 also speed it up considerably.
31 - (libcoro) do not rely on makecontext passing void *'s unscathed.
32 - fix compiletime dependencies on libcoro in the Makefile.
33 - cctx_count wasn't always updated properly.
34 - Coro::State::cctx_stacksize wasn't applied correctly.
35 - new function Coro::State::cctx_max_idle.
36 - the default max number of idle C contexts is now 4.
37 - (libcoro) try harder to get _setjmp/_longjmp.
38 - (libcoro) cleanup and extend the libcoro API to officially
39 allow the creation of empty source contexts.
40 - very experimental workaround for the totally broken netbsd platform.
41 - tried to hack around openbsd bugs.
42
43 4.804 Wed Nov 5 16:36:00 CET 2008
44 - Coro::Debug::new_unix_server would not create a non-blocking listening
45 socket, sometimes causing freezes.
46 - (libcoro) fix misaligned stack points for setjmp and assembly
47 methods, which can cause crashes on x86/x86_64 with a sufficiently
48 aggressive compiler.
49 - new function: Coro::Debug::new_tcp_server.
50 - move ->throw into the Coro class because it only works on coro objects.
51
52 4.803 Mon Nov 3 17:16:12 CET 2008
53 - (libcoro) use a global asm statement to become independent of gcc
54 otpimisations for CORO_ASM (thanks to pippijn for the idea).
55 - try to workaround yet another broken bsd, this time dragonfly.
56
57 4.802 Thu Oct 30 10:56:12 CET 2008
58 - support -fno-omit-frame-pointer on x86 with the assembly method.
59 - tune 01_unblock.t tests a bit.
60
61 4.801 Wed Oct 22 18:33:37 CEST 2008
62 - improve readline speed for very long "lines".
63 - backport to 5.8.8.
64
65 4.8 Thu Oct 2 13:34:40 CEST 2008
66 - new function Coro::AIO::aio_wait.
67 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
68 - greatly speed up and reduce memory usage of Coro::AIO requests.
69 - implement some other µ-optimisations.
70
71 4.749 Mon Sep 29 14:40:12 CEST 2008
72 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
73
74 4.748 Sat Sep 27 14:03:19 CEST 2008
75 - implement, but do not document, PerlIO::cede(granularity).
76 - Coro::Storable forgot to wrap Storable::pstore.
77 - work around the multitude of leaks and memory corruption
78 bugs in PerlIO::via by using our own C-level perliol. As a side
79 effect, Coro::Storable is now much, much, much faster.
80
81 4.747 Tue Sep 23 01:59:41 CEST 2008
82 - fix a per-cv memleak (one empty array was leaked per
83 code reference).
84 - avoid a crash in coro->call|rss when the coroutine was already
85 destroyed (most noticably when using Coro::Debug::ps :)
86 - also protect *Storable::FILE.
87 - push up default storable granularity to 20ms.
88
89 4.746 Sun Sep 21 03:22:20 CEST 2008
90 - be more insistent on locking Storable against reentrancy
91 in Coro::Storable.
92 - move swap_def?v and throw to Coro::State, as documented.
93
94 4.745 Thu Jul 24 00:14:38 CEST 2008
95 - remove debugging code related to MgPV_nolen_const, also try to
96 make it compile with perl 5.8.6 (yes, apple apparently loves
97 outdated software). Reported by John S.
98
99 4.744 Tue Jul 8 22:06:35 CEST 2008
100 - correctly provide default DIE/WARN handlers as documented.
101 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
102 implementations inside perl work fine for us.
103
104 4.743 Mon Jun 16 00:21:57 CEST 2008
105 - when using Coro::EV without running EV::loop it could
106 result in busy-waiting for events, this has been fixed.
107 - reduce codesize and improve performance by using EV_DEFAULT_UC.
108
109 4.742 Sat May 31 14:10:21 CEST 2008
110 - implement a workaround for (some) perl <5.8.8 versions.
111 - require EV 3.3+.
112
113 4.741 Fri May 30 23:33:09 CEST 2008
114 - tell netbsd how utterly broken their imitation of an OS is
115 and refuse to build by default if pthreads are in use.
116 - switch to "s" method on all bsd's by default, as their ucontext
117 stuff seems just too broken.
118 - fix a bug in Coro::Select.
119
120 4.74 Thu May 29 20:05:31 CEST 2008
121 - do not test Coro::LWP for lack of dependencies.
122 - do not test Coro::Select for lack of working perls.
123
124 4.73 Thu May 29 2008
125 - fix a bug in Coro::EV which would cause it to block despite
126 there being runnable coroutines.
127 - sprinkle "no warnings" freely over everything, also suppress
128 warnings for some other modules.
129 - fix typo in WSAEWOULDBLOCK.
130
131 4.72 Sun May 25 05:14:36 CEST 2008
132 - tweak META.yaml a bit, unfortunately, there is no documented way
133 to have optional dependencies with CPAN. doh :(
134 - avoid running some tests on windows because they would fail due to
135 perl bug (broken fork, broken pipes...).
136 - work around perl on windows bugs where perl returns undocumented
137 error codes for sysread, syswrite etc. by taking advantage
138 of AnyEvent's workaround for that problem.
139 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
140 a common perl implementation bug on windows.
141 - use unix domain sockets in testsuite to work around
142 common perl implementation bugs on widows (they are emulated by
143 tcp sockets on windows. ugh.)
144
145 4.71 Sat May 24 20:01:27 CEST 2008
146 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
147 - take advantage of async name resolution of AnyEvent::Util.
148 - work around brutal inet_aton override in Coro::LWP.
149 - take advantage of the readyhook in Coro::EV, for smoother
150 scheduling.
151
152 4.7 Sun May 11 00:32:19 CEST 2008
153 - completely reworked the Coro manpage.
154 - added Coro::AnyEvent, generic event loop integration.
155 - implement cancel, ready and kill commands in Coro::Debug.
156 - document find_coro in Coro::Debug.
157 - incompatible change: rename has_stack to has_cctx.
158 - Coro::AIO and Coro::BDB no longer force event model detection,
159 use AnyEvent::AIO and AnyEvent::BDB.
160
161 4.6 Sat Apr 26 10:05:14 CEST 2008
162 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
163 by the Coro module. It was a mistake to have it in the first place.
164 - (experimental) support for activestate perl 5.10 (method "w").
165 - (experimental) support for strawberry perl (method "a").
166 - coro_sigelem_set did not return a value although it had to,
167 actual impact unknown.
168
169 4.51 Mon Apr 14 13:28:27 CEST 2008
170 - make it compile again on 5.8.
171
172 4.50 Thu Apr 10 09:43:17 CEST 2008
173 - I did it twice! (see 4.49).
174
175 4.49 Sun Apr 6 21:23:31 CEST 2008
176 - grr, instead of compiling the recent changes on 5.10 only they
177 were compiled on 5.8 only.
178
179 4.48 Sun Apr 6 20:36:46 CEST 2008
180 - allow coroutine switches during eval's under 5.10.x, as it seems
181 the parser is a per-interpreter option now, so this is safe
182 (this might fix the odd crashing bug).
183 - drop support for 5.9.x versions: they are dead, jim.
184
185 4.47 Sun Apr 6 00:37:52 CEST 2008
186 - force cctx allocation on API calls: we know we need to force one
187 and gcc actually manages to confuse our heuristic nowadays,
188 leading to crashes and worse.
189 - document force_cctx.
190
191 4.46 Fri Apr 4 22:05:43 CEST 2008
192 - upgrade libcoro, resulting in pthread-backend (which was only created
193 to fulfill the rules of the programming languages shootout).
194
195 4.45 Thu Mar 13 11:55:36 CET 2008
196 - fix a file leaking bug in eg/mhyttpd that would allow
197 downloading of any file (reported by oesi).
198 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
199 (also add testcase).
200 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
201 apparently having loads of fun with that).
202 - implement $state->swap_defsv, ->swap_defav and document ->throw.
203
204 4.4 Wed Feb 13 16:44:29 CET 2008
205 - only assume .cfi pseudo isns exist on GNU/Linux.
206 - add get_timed method to Coro::Channels.
207 - fixed Coro::Debug manpage.
208 - perl 5.11 compatbility improvement based on patch by
209 Andreas König.
210
211 4.37 Sun Jan 20 11:25:23 CET 2008
212 - soften the check for an active parser for perl 5.10 (avoiding
213 "Coro::State::transfer called while parsing" in many cases).
214
215 4.36 Sun Jan 13 10:53:56 CET 2008
216 - reset diehook when terminating from an async_pool as to not
217 trigger any __DIE__ handlers.
218
219 4.35 Sun Jan 13 04:14:13 CET 2008
220 - "bt" debug command now displays any exceptions
221 from longmess and also skips the innermost
222 stackframes, giving more useufl output.
223 - allow backtraces in coroutines blocked in Coro::EV,
224 at a <1% speed hit.
225 - handle localising of $SIG{__WARN__} and __DIE__
226 properly (with a proper amount of dirty hacking).
227
228 4.34 Sat Dec 22 17:49:53 CET 2007
229 - upgrade to EV version 2.0 API.
230
231 4.33 Mon Dec 17 08:36:12 CET 2007
232 - make Coro::AIO etc. loadable in the absence of EV.
233
234 4.32 Mon Dec 17 07:46:02 CET 2007
235 - majorly improved Coro::Handle's performance with Coro::EV.
236 - implemented the until now mythical Coro::BDB module.
237 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
238
239 4.31 Wed Dec 5 12:32:39 CET 2007
240 - remove warn statement form Coro::Util.
241
242 4.3 Tue Dec 4 20:33:14 CET 2007
243 - calls to the idle function could cause stack corruption
244 when the stack changed.
245 - do no longer rely on the presence of EV::DNS if EV is used
246 (because it is gone), but instead take avdantage of EV::ADNS
247 if available.
248 - add ($) prototypes to all functions in Coro::Storable.
249 - use a conventional (safer) idle callback in Coro::EV.
250 - do accept quickly in Coro::Debug to avoid endless loops.
251
252 4.22 Fri Nov 30 16:04:04 CET 2007
253 - really use optimised versions for Event and EV in Coro::Util
254 and Coro::Handle.
255
256 4.21 Sun Nov 25 10:48:59 CET 2007
257 - fix a spurious memory read.
258 - Coro::EV no longer keeps the eventloop "alive".
259
260 4.2 Fri Nov 9 20:47:05 CET 2007
261 - enable/disable tracing from a new coroutine, not a pooled one.
262 - fix a memleak in Coro::Event.
263 - removed killall call from fork_eval.
264 - made sure store_fd is already loaded so that fork_eval
265 does not have to parse autoload in each subprocess.
266 - only use assembly method if -O switch is in $Config{optimize}.
267 - add (optional) Coro::EV module, so far the best event loop module
268 directly supported by Coro.
269 - if the event model is EV, use EV::DNS to resolve
270 stuff in Coro::Util.
271 - don't get confused by multiple event notifications in Coro::Handle.
272 - initial support for EV (libevent interface).
273 - require Event and EV using configure_requires, to force their existance.
274
275 4.13 Wed Oct 24 07:26:45 CEST 2007
276 - add Coro::Storable::blocking_thaw.
277 - use a vastly more complicated technique to localise
278 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
279 - use a coroutine for the idle callback Coro::Event,
280 instead of running Event in the current coroutine context.
281 This also catches recursive invocations.
282 - actually report fork errors in gethostbyname and inet_aton.
283
284 4.11 Thu Oct 11 02:40:24 CEST 2007
285 - port to threaded perls.
286
287 4.1 Thu Oct 11 02:38:16 CEST 2007
288 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
289 be local to each coro (see Coro::State).
290 - incompatible change: for very deep reasons, cede and cede_notself
291 cannot return anything, so nothing will be returned.
292 - possibly bring back 5.10 compatibility (untested).
293 - work around stupid (and wrong) warning on 5.10 :(.
294 - overlay the saved state over the context stack. This saves
295 a few hundred bytes per coroutine on average and also
296 speeds up context switching a bit.
297 - further tune default stack sizes.
298 - (more) correctly calculate stack usage in coro_rss.
299 - Coro::Storable::blocking_* did not properly lock
300 resulting in races between coroutines.
301 - added Coro::Storable::guard.
302 - stopping to trace a coroutine could destroy the cctx of
303 an unrelated coroutine.
304 - explain the relationship between Perl and C coroutines in
305 more detail in Coro::State.
306 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
307 causing a fork per resolve. This also affected Coro::Socket.
308 - switch to a separate stack in $coro->call/eval to avoid
309 invalidating pointers.
310
311 4.03 Sat Oct 6 21:24:00 CEST 2007
312 - added Coro::throw method.
313 - minor code cleanups.
314
315 4.02 Sat Oct 6 02:36:47 CEST 2007
316 - fix a very minor per-coroutine memleak (a single codereference).
317 - fixed a bug where the currently in-use c context would be freed
318 prematurely (can happen only when programs change the stacksize
319 or use tracing).
320 - tracing can no longer keep a coro alive after it terminated.
321 - do static branch prediction in the common path for gcc. gives
322 about 2-5% speed improvement here.
323
324 4.01 Fri Oct 5 22:10:49 CEST 2007
325 - instead of recreating *a* standard output handle we simply
326 use STDOUT, which is faster and hopefully more robust.
327
328 4.0 Fri Oct 5 12:56:00 CEST 2007
329 - incompatibly changed Coro::Storable::freeze.
330 - major new feature: added Coro::Debug, for interactive coroutine
331 debugging, tracing and much more.
332 - major bug fix: unbelievable, but true: $_, $/ and many other
333 "saved" variables actually weren't being saved. This has been fixed,
334 of course, while increasing performance while losing all the save
335 flags.
336 - save flags are gone, and all the api functions dealing with them.
337 - added Coro::Semaphore::adjust.
338 - added Coro::Util::fork_eval.
339 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
340 - added Coro::killall.
341 - reduce initial stack sizes to allow for "micro-coroutines".
342 - better async_pool resource management, moved parts of async_pool
343 handling to XS (major speed improvement).
344 - actually croak before modifying important data structures.
345 - refuse to transfer while compiling.
346 - possibly support eval EXPR better now.
347 - enable assembly per default on linux+bsd x86+amd64.
348 - all internal members were renamed _something for easier subclassing.
349 - many minor tweaks.
350
351 3.63 Wed May 16 14:10:06 CEST 2007
352 - implement handcoded assembly for x86/amd64 SVR ABI.
353
354 3.62 Fri Apr 27 21:36:06 CEST 2007
355 - upgrade libcoro (which might set unwind info correctly).
356 - change default on linux to setjmp/longjmp.
357
358 3.61 Thu Apr 19 12:36:18 CEST 2007
359 - Coro::Storable caused an endless loop when thawing invalid
360 pst's sometimes.
361 - use a Semaphore in Coro::Storable, as Storable doesn't
362 seem to be reentrant (although it is documented to
363 be threadsafe...).
364 - fix Coro::Signal to bring back the original unreliable
365 but stateful semantics.
366 - fixed a lot of typos in Coro.pm (patch submitted by David
367 Steinbrunner, which applied flawlessly).
368
369 3.6 Sat Apr 14 17:13:31 CEST 2007
370 - added some bugfixes to get eg/myhttpd working again.
371 - added Coro::Storable for often-cede'ing freeze/thaw.
372 - try to do a clean exit when a coroutine calls exit
373 (EXPERIMENTAL).
374 - got rid of indirect call through _coro_init.
375 - updated the partly antique examples in eg/ to
376 work again and be a bit less magic, too.
377 - fixed Coro::Signal semantics to work as documented again.
378
379 3.55 Sun Mar 25 01:20:47 CET 2007
380 - add SAVE_DEFFH to save the default filehandle and enable
381 it by default.
382 - finally move socket-operations from Coro::Socket to Coro::Handle
383 to be able to unblock foreign sockets.
384 - add Coro::State::save_also and guarded_save.
385 - add count accessor to Coro::Semaphore.
386 - add Coro::State::cctx_stacksize.
387 - just for the fun of it, do not rely on implicit context,
388 which can dramatically improve performance, but people
389 using windows-process-emulation perls do not care much about
390 performance.
391
392 3.51 Sun Mar 4 14:18:04 CET 2007
393 - fixed a problem when you weakened references to Coro::State's
394 (patch by Yuval Kogman).
395
396 3.501 Wed Feb 28 12:44:07 CET 2007
397 - rename some global symbols as macosx from hell redefines
398 them without asking.
399
400 3.5 Tue Feb 13 20:22:53 CET 2007
401 - do AnyEvent model detection earlier, avoiding problems
402 caused by first using AnyEvent and later Coro::Event.
403 - implement and document Coro::Event event objects.
404 - fix a potential problem in Coro::Event causing crashes.
405 - initialise PL_comppad when creating a new coroutine,
406 avoids crashes on early coro destruction.
407
408 3.41 Mon Jan 22 19:19:49 CET 2007
409 - readline on Coro::Handle did not support undefined $/,
410 nor did it deliver partial lines on EOF or error.
411 - implement malloc fallback for stack allocation because
412 stupid broken idiotic OSX has a stupid broken
413 idiotic fits-the-whole-os mmap "implementation" and
414 my dick feels longer if Coro is portable even to
415 obsolete platforms.
416
417 3.4 Fri Jan 19 21:52:54 CET 2007
418 - remove t/09_timer.t, as it isn't really testing much
419 but was rather flaky in practise.
420 - async_pool coro would keep arguments and callback alive until
421 it was reused.
422 - cancellation of a coroutine could cause spurious idle calls
423 in cede_notself.
424
425 3.3 Sat Jan 6 03:45:00 CET 2007
426 - implement $coro->on_destroy.
427 - Coro::Event blocking semantics have been changed,
428 documented and - hopefully - improved.
429 - fix nice adding, not subtracting, from priority.
430 - fix ->prio and api_is_ready (patch by Mark Hinds).
431 - fixed an assert ("... == prev__cctx->idle_te")
432 that could errronously trigger.
433 - fix various large and small memleaks.
434 - use a (hopefully) more stable cancel implementation
435 that immediately frees the coroutine data.
436 - cede/cede_notself return a status now.
437 - added Coro::guard function.
438 - added a global coroutine pool for jobs (on my machine,
439 I can create and execute 48k simple coros/s with async,
440 and 128k coros with async_pool).
441 - Coro::AIO now uses the coroutine priority as io priority.
442
443 3.2 Fri Dec 22 05:07:09 CET 2006
444 - improve portability to slightly older perls.
445 - use cleaner coroutine destruction.
446 - simplify configuration for users.
447 - optionally (unrecommended) prefer perl functions over
448 their coro replacements.
449
450 3.11 Tue Dec 5 13:11:24 CET 2006
451 - fixed some bogus assert's, but as perl.h disables assert even
452 without NDEBUG (thank you very much), not too many people should
453 notice (that did include myself). Andreas König noticed, though :)
454 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
455 it. Also noticed by Andreas König :)
456 - save/restore tainted status.
457 - verified to pass the testsuite on my 5.9.5.
458
459 3.1 Mon Dec 4 23:03:40 CET 2006
460 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
461 - incompatible change: transfer flags are now per-state.
462 - give a better error message on deadlock.
463 - document Coro::nready.
464 - enhanced testsuite.
465
466 3.01 Sun Dec 3 23:47:42 CET 2006
467 - forgot to include Coro::Timer.
468
469 3.0 Sun Dec 3 22:57:25 CET 2006
470 - the "FINALLY COMPLETELY STABLE" release (coming soon:
471 the "FAMOUS LAST WORDS" release).
472 - implement a new stack sharing algorithm, which uses a stack
473 pool (size currently hardcoded).
474 - make stack sharing mandatory (it no longer uses a heuristic).
475 - eval/die no longer cause weird problems under heavy use.
476 - Coro::Event could cause livelocks if it was used but
477 no Coro::Event watchers were used.
478 - Coro::Event now uses asynccheck as prepare does not
479 check for changed watchers.
480 - Coro::Event allows multiple waiting coros and will wake up one
481 per event.
482 - Coro::Event should be cleaner and more efficient now.
483 - new utility function Coro::unblock_sub.
484 - document the sad fact that Event is no longer reentrant.
485 - putting a coroutine into the ready queue twice could under
486 some circumstances lead to stack corruption.
487 - minor incompatibility: subclassing Coro::State is supported
488 directly now without going through a _coro_state member.
489 - state/coro switching is much faster now.
490 - very minor optimisations and code/documentation cleanup.
491 - avoid problems due to compiler inlining.
492 - removed timers from Coro::Timer -> use AnyEvent instead.
493 - replaced Coro::idle coroutine by (cleaner) idle handler.
494 - updated to newest libcoro.
495 - implement enhanced support for valgrind.
496 - implement is_ready and return value for ready.
497 - removed Coro::Cont, it was a misnomer (it's generators),
498 and it was somewhat annoying to get it right. Will come back
499 if somebody asks for it :)
500 - many ->wait methods and Coro::Event could return
501 spuriously without the event having happened.
502
503 2.5 Tue Nov 7 12:22:33 CET 2006
504 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
505 use AnyEvent, moved them to Coro/.
506 - added Coro::LWP which contains all the uglyness required to
507 make LWP non-blocking.
508 - should work with perl 5.9.x now (Andreas König made me do it).
509 - fixed another bug in Coro::Select when the timeout was undef.
510 - reuse PL_start_env for all coros, saves some memory per coroutine.
511 - manage PL_top_env differently, hopefully to avoid panic: top_env.
512 - timeout argument was not properly used in Coro::Socket.
513 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
514 - emulate undocumented(!) functionality of IO::Socket required
515 by LWP(!!).
516 - updated eg/lwp to work with newer lwp's.
517 - remove "FATAL: uncaught exception" prefix. Coroutines that die
518 kill the whole process, just as exceptions in the main "coroutine"
519 did already.
520
521 2.1 Wed Nov 1 23:01:13 CET 2006
522 - fix a long-standing bug in Coro::Select where select with
523 zero timeout would instead change the current default filehandle.
524 - use a simpler and hopefully more robust way to clone padlists
525 (uses less memory and a perl function instead of our own).
526 - coro can now create a stack guard area on many architectures.
527 - Coro::AIO properly reexports additional functions from IO::AIO.
528 - updated libcoro with a workaround for OS X,
529 pach and testing by Michael Schwern.
530
531 2.0 Tue Oct 24 05:47:17 CEST 2006
532 - support additional aio requests in Coro::AIO.
533
534 1.9 (never properly released due to a glitch)
535
536 1.8 Thu Feb 2 00:59:06 CET 2006
537 - applied suggested patch by SAMV to avoid problems during stupid
538 mark & sweep gc run.
539 - applied patch by Scott Walters for 5.9.3 compatibility.
540
541 1.7 Tue Dec 27 01:41:58 CET 2005
542 - added Coro::AIO, a thin wrapper around IO::AIO.
543 - improved Makefile.PL explanations.
544
545 1.6 Mon Dec 12 21:30:05 CET 2005
546 - additionally save PM_curpm might fix as-of-yet
547 unseen problems with regex matches being attributed to the
548 wrong package.
549 - add t/10_bugs.t, which currently checks against imho broken
550 perls that use 0.26MB of stack space per Perl_magic_get
551 invocation (newer linuxes) as opposed to the 0.0002MB perl
552 normally uses.
553 - make stacksize configurable for the ultimate debian experience.
554
555 1.51 Mon Dec 12 18:48:36 CET 2005
556 - remove debugging warn() accidentally left in Coro::Select.
557
558 1.5 Tue Nov 29 13:32:44 CET 2005
559 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
560 - fix Coro::Select 'select'.
561 - strict'ify some modules.
562 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
563 - use XSLoader in selected modules.
564 - remove some 5.6 compatibility cruft.
565
566 1.4 Tue Sep 6 00:11:05 CEST 2005
567 - libcoro did not take into account the trampoline on amd64,
568 when 'l'inux method was used.
569
570 1.31 Tue Aug 30 23:31:33 CEST 2005
571 - some portability fixes/workarounds.
572
573 1.3 Sat Aug 20 03:08:56 CEST 2005
574 - no code changes. module seems to work fine.
575 - improve Coro::State docs, remove reference to nonexisting
576 Coro::State::flush method.
577 - no longer autodetect windows, present it as an option instead.
578
579 1.2 Mon May 16 02:00:55 CEST 2005
580 - included libcoro.c earlier in State.xs, might improve portability.
581 - use a faster and possibly more stable Coro::Cont implementation.
582 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
583 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
584 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
585
586 1.11 Thu Mar 3 18:00:52 CET 2005
587 - change of contact address.
588
589 1.1 Tue Feb 22 20:51:16 CET 2005
590 - support [l]inux method on amd64.
591 - allow some context switches while compiling/eval'ing.
592
593 1.0 Tue Aug 10 03:47:27 CEST 2004
594 - Event 0.89 fixes the bug again.
595 - fix the remaining(??) memleak. I didn't free the cache-AV for
596 padlists, so one AV per created anon-sub leaked.
597
598 0.97 Fri May 14 15:23:32 CEST 2004
599 - remove stupid gcc extension that was an experiment only anyway.
600
601 0.96 Thu May 13 18:09:29 CEST 2004
602 - tell people if they have yet another broken Event variant.
603 - made Coro to work _better_ with threads (it seems that, as long
604 as you use Coros in one thread only everything will work fine.
605 Looking at the perl source the problem might be missing locking
606 between Coro and some perl internal routines. That's not easily
607 fixable).
608 - (hopefully) got rid of the static xnext inside transfer.
609
610 0.95 Sun Feb 15 03:20:28 CET 2004
611 - removed Coro::State::flush, implemented a new and much
612 faster caching mechanism by attaching magic to the CV.
613 - a memleak with (real) closures remains.
614 - dramatically reduced size of stacks to 65536 (linux/x86), after
615 observing that even myhttpd never had more than a kilobyte
616 of stack in use (stack is still only physically allocated in
617 pagesize increments for systems suporting mmap).
618 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
619 - fix for threaded perl.
620
621 0.9 Sun Nov 23 04:32:39 CET 2003
622 - fix a few very hard to track down but (of course) stupid
623 memory leaks.
624 - stringify version number to avoid locale problems :(.
625 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
626 hopefully ignoring it "just works".
627
628 0.8 Wed Nov 5 19:38:40 CET 2003
629 - port to perl5.005_03, but only 5.8.x+ is supported!
630 - honor LocalAddr even without LocalPort in Coro::Socket.
631
632 0.7 Tue May 27 03:12:38 CEST 2003
633 - the version jump indicates some level of testing, not gobs
634 of new features.
635 - uh... I found the prompt function in ExtUtils::MakeMaker.
636 Highly correct stuff, that is...
637 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
638 the fix is not well understood by the author, so beware :(.
639
640 0.652 Thu May 8 02:54:46 CEST 2003
641 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
642 cause version 4 doesn't have ucontext.
643 - Benjamin Reed reported that setjmp works fine on darwin,
644 so preselect it.
645
646 0.651 Sat Mar 29 15:00:23 CET 2003
647 - fix a bug in Coro::Handle where some bytes could get lost
648 on reading, reported by jason@nichego.net.
649
650 0.65 Sun Mar 23 00:08:26 CET 2003
651 - added a README.linux-glibc.
652 - new module Coro::Select.
653 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
654 (I never claimed that I know what I am doing ;).
655 - set default to setjmp/longjmp on non-x86-linux.
656
657 0.6 Thu Nov 21 11:09:06 CET 2002
658 - made lazy context switching the default.
659 - fixed the bug where SemaphoreSet::guard
660 was the same as timed_guard and timed_guard was missing.
661 - fix a memleak in Coro::Timer.
662 - reclassified context sharing as not experimental.
663
664 0.534 Sun Apr 14 03:05:12 CEST 2002
665 - fixed backspace => tab in header parsing (myhttpd).
666 - added eg/readline, for elmex.
667 - Coro::Event now calls ->start on first call instead of
668 ->again, which created wrong timeouts for the first ->next.
669 - fixed a bug where Coro::Socket returned a closed socket instead
670 of nothing on ECONNREFUSED and similar error conditions.
671
672 0.533 Mon Feb 18 18:49:40 CET 2002
673 - adapt to finally working Event-0.85.
674 - fixed buggy event initialization (trapped under windows, but how
675 could it possibly have worked anywhere?)
676
677 0.532 Wed Jan 16 02:45:32 CET 2002
678 - removed the reference to pp_entersub, might work on
679 (some) win32 perls now (testsuite works under cygwin,
680 without Event, which freezes).
681
682 0.531 Mon Dec 10 22:18:44 CET 2001
683 - Fixed a bug in SemaphoreSet::timed_down.
684
685 0.53 Tue Nov 27 21:11:13 CET 2001
686 - some tests for Event.
687 - slightly different internal architecture to get refcnt into the object
688 and not the reference. might fix a few bugs, certainly did introduce
689 new ones.
690 - Coro::Timer (independent of Event).
691 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
692
693 0.52 Tue Nov 6 21:36:18 CET 2001
694 - ported to cygwin (trivial).
695 patch by Gerrit P. Haase <gp@familiehaase.de>.
696 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
697
698 0.51 Thu Nov 1 20:39:01 CET 2001
699 - terminate/cancel now work properly (otherwise termination
700 could cause a "next coroutine is not and contains not..."
701 error.
702 - added Coro::Socket::shutdown.
703 - Coro::Event::loop no is the same as Event::loop.
704 - implemented terminate with args + join.
705
706 0.5 Fri Sep 28 16:15:35 CEST 2001
707 - fixed "print" on a Coro::Handle. The print method worked.
708 - small tweaks (seem to reduce memory consumption a lot)
709 in various modules.
710 - splendid use of "no warnings" scattered throughout.
711 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
712
713 0.49 Sun Sep 16 02:42:45 CEST 2001
714 - changed some method calls to function calls for speed
715 inside Coro::Handle.
716 - make Coro::Handle use an array instead of a hash for
717 speed reasons.
718 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
719 sjlj and ucontext should now work.
720 - IRIX-specific port for libcoro.
721 - swapped order of accept results to match IO::Socket.
722 - changed getsock/peername to sock/peername to match IO::Socket.
723 - fixed a bug that caused segfault when returning to main task
724 under some circumstances.
725 - other bugfixes.
726
727 0.45 Sun Sep 2 02:54:01 CEST 2001
728 - new method Coro::Handle::timeout.
729 - corrected speling of set...name to get...name in Coro::Socket.
730 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
731 - Coro::Handle now supports fileno and FILENO.
732 - added eg/myhttpd, a web server.
733 - Coro::Socket now accepts numeric ports as well as "port(number)"
734 syntax.
735 - moved some scheduling primitives into xs code. more to come.
736 - new simple priority system.
737 - implemented Coro::Event using XS, almost four times faster!
738 - small memory corruption problem fixed (boy that was difficult).
739
740 0.13 Wed Aug 8 16:53:07 CEST 2001
741 - new method Coro::Handle::readline.
742 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
743 (see eg/lwp on how to make LWP non-blocking).
744 - renamed Coro::Socket::new_inet to new.
745 - added Coro::Util, some utility functions.
746 - die/eval now works better (still get annoying (but true ;)
747 "Callback called exit" messages).
748 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
749 - much better Coro::Socket support.
750
751 0.12 Fri Jul 27 04:19:01 CEST 2001
752 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
753 - very experimental stack sharing algorithm. not 100% safe but
754 should work well in practise ;)
755 - added Coro::RWLock.
756 - Coro::Specific now works.
757
758 0.11 Tue Jul 24 22:49:21 CEST 2001
759 - added specialized hack for newer and older linux versions (fast).
760 - renamed Coro::Event::IO to Coro::Handle.
761 - new module Coro::Socket.
762
763 0.1 Tue Jul 24 01:47:53 CEST 2001
764 - release candidate 3. A rather new internal structure :(
765 - the great renaming: Damian Conway gave me a suitable replacement
766 for yield.
767 - added Coro/libcoro, a portable coroutine implementation for C
768 (not even perl-dependent!!!), which can be used by Coro::State.
769 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
770 perl-5.6.
771 - new function Coro::Event::idle.
772 - the idle process is now overriden
773 by default in Coro::Event.
774 - Coro::Channel now enforces the size.
775 - canceling events now works in all cases.
776 - Coro state now include $_ and $@.
777 - yet another bug workaround that I do not really understand :(
778 - new module Coro::Event::IO (very undocumented) to do
779 non-blocking i/o.
780 - performance tuning.
781
782 0.08 Thu Jul 19 06:13:25 CEST 2001
783 - release candidate 2 (new functionality).
784 - add Coro::State::flush function.
785 - transfer now supports a flags value (mostly for speed ATM).
786 - might compile and work in the presence of threads now.
787 - continuations have a saner syntax.
788 - no more memleaks.
789
790 0.07 Tue Jul 17 17:40:18 CEST 2001
791 - release candidate 1 ;)
792 - slightly nicer code.
793 - fixed a scheduling bug in Coro::Event.
794
795 0.06 Tue Jul 17 04:23:24 CEST 2001
796 - ok, I found the showstopper - the same sub must not be
797 re-entered in two different coroutines, otherwise => crash. I
798 see no easy solution to this problem, except by walking the call
799 chaing and saving/restoring the cv's, which is what I do now.
800 - memory leaks still latent, especially at thread termination.
801 - Coro::Event now works (haha).
802
803 0.05 Sun Jul 15 17:32:20 CEST 2001
804 - fixed a few issues in Coro::Event.
805 - I forgot to include Coro::Event in 0.04 :(:(:(
806
807 0.04 Sun Jul 15 05:24:59 CEST 2001
808 - @_ is now properly localized.
809 - Coro::State is now easier subclassable.
810 - Coro::Cont now coroutine-aware.
811 - Coro::Specific is a low-overhead module to create
812 coroutine-specific vars.
813 - Coro::Event provides a simple interface to Event.
814
815 0.03 Fri Jul 13 14:51:52 CEST 2001
816 - transfer() now implemented in XS (beware).
817 - new module Coro::Cont for really faked continuations.
818 - big internal architecture changes: Coro::State is now
819 really low-level and can thus be used to implement other
820 interesting things, While "Coro::" implements a process-like
821 model. Still crude and subject to change.
822 - $_ and $@ are no longer being localized.
823
824 0.02 Tue Jul 10 01:38:17 CEST 2001
825 - implemented "async" attribute.
826 - $_ and $@ are now localized.
827 - added Coro::Channel.
828 - more testcases, still no docs.
829
830 0.01 Tue Jul 3 02:18:41 CEST 2001
831 - original version; copied from Convert::Scalar.
832