ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.407
Committed: Thu Nov 13 17:32:01 2008 UTC (15 years, 6 months ago) by root
Branch: MAIN
CVS Tags: rel-4_912
Changes since 1.406: +1 -0 lines
Log Message:
4.912

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