ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.455
Committed: Fri Nov 21 02:39:40 2008 UTC (15 years, 6 months ago) by root
Branch: MAIN
Changes since 1.454: +1 -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 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
9 - Coro::Select did errornously dup the file descriptors
10 and didn't work with all AnyEvent backends.
11 - bumped $Coro::POOL_RSS up to 32kb by default.
12
13 5.0 Thu Nov 20 10:35:05 CET 2008
14 - NEW ARCHITECTURE: use the latest 4.x version if you experience
15 stability issues.
16 - bump API version to 7 - all dependents must be recompiled.
17 - removed timed_* functions - they were not being used anyways
18 and should be replaced by a more generic mechanism -
19 and were annoying to support anyways :)
20 - removed SemaphoreSet's waiter method - use sem method instead.
21 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
22 - async_pool did free a scalar value twice
23 ("Attempt to unreference...").
24 - fix a longstanding bug where calling terminate on a coro that
25 was waiting for a semaphore that was just becoming available
26 would cause a deadlock (semaphore would get into a state where
27 it was available but waiters were still blocked).
28 - calling throw on a coroutine that is waiting for a semaphore will
29 no longer make it acquire the semaphore (and thus leak a count).
30 - perl's process emulation is now not even theoretically supported
31 anymore.
32 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
33 conversion of callback-style to blocking-style.
34 - new methods $coro->schedule_to and ->cede_to, to specifically
35 schedule or cede to a specific coroutine.
36 - new function Coro::Semaphore::wait.
37 - use named constants in Coro::Channel (Richard Hundt).
38 - directly patch the entersub opcode calling SLF functions (cede,
39 transfer and so on). this does speed up context switching, but
40 more importanly, it frees us from the hardcoded behaviour of
41 entersub, so we might actually be able to return something from
42 those functions and atcually create new ones.
43 - take advantage of __builtin_frame_address on gcc.
44 - expose THX in coroapi (not sure whether this was a wise decision,
45 as "threaded" perls are running at half speed anyways).
46 - implement execute_slf (schedule-like-function) interface that makes
47 it possible to implement schedule-like-functions in XS.
48 - use new SLF interface to massively speed up Coro::EV by roughly a
49 factor of two.
50 - used new SLF interface to massively speed up Coro::Semaphore by a
51 factor of three.
52 - used new SLF interface to speed up Coro::AIO by roughly a factor of
53 four and reduce its memory usage considerably.
54 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
55 for a nice speedup and vastly more correct behaviour. Also implement
56 a new method "sem" to get at the underlying semaphore object.
57 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
58 (in comparison) 20-40% speedup.
59 - used new SLF interface to reimplement Coro::Signal gaining
60 some unknown (because I was too lazy), but certain, speedup, and also
61 making signals reliable for the first time.
62 - used new SLF interface and other optimisations to speed up async_pool
63 by a factor of two. It also doesn't rely on perl's exception mechanism
64 to exit anymore. The overhead for terminating an async_pool, coro over
65 a normal async is now very small.
66 - sped up coroutine creation/destruction by 40%.
67 - forgot to include Coro/libcoro/README in the dist for all these years.
68 - work around a freebsd pthreads bug (manual testcancel is required as
69 pthread_cond_wait isn't a cancellation point on freebsd).
70 - use new rouse functions to speed up and simplify Coro::BDB.
71 - make "prefer perl native functions" work with threaded perls.
72 - condense Coro::Debug ps output, hint at v and w flags.
73 - (libcoro) lots of minor cleanups and portability improvements.
74
75 4.914 Wed Nov 19 12:54:18 CET 2008
76 - fix a disastrous bug in the readline optimisation
77 introduced in 4.801.
78
79 4.913 Sat Nov 15 07:58:28 CET 2008
80 - async_pool did free a scalar value twice
81 ("Attempt to unreference...").
82
83 4.912 Thu Nov 13 18:31:23 CET 2008
84 - minor cleanups.
85 - use much larger stacks on linux and perl < 5.8.8.
86 - Coro::Debug::new_unix_server did not unlink the socket
87 when destroyed.
88
89 4.911 Tue Nov 11 04:26:17 CET 2008
90 - "port" to threaded perls.
91
92 4.91 Mon Nov 10 05:36:38 CET 2008
93 - the ->throw exception object no longer leaks.
94 - creating a new cctx leaked a scopestack entry (memleak).
95 - new coroutines didn't get created with a zero flags field
96 (unknown impact).
97 - calling ->throw on a not-yet-started coroutine should now work
98 instead of being ignored.
99 - ->throw is now supported on Coro::State objects.
100 - clean up cctx creation code a bit.
101 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
102
103 4.9 Sat Nov 8 17:45:27 CET 2008
104 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
105 - (libcoro) no longer leak threads in the experimental pthread backend,
106 also speed it up considerably.
107 - (libcoro) do not rely on makecontext passing void *'s unscathed.
108 - fix compiletime dependencies on libcoro in the Makefile.
109 - cctx_count wasn't always updated properly.
110 - Coro::State::cctx_stacksize wasn't applied correctly.
111 - new function Coro::State::cctx_max_idle.
112 - the default max number of idle C contexts is now 4.
113 - (libcoro) try harder to get _setjmp/_longjmp.
114 - (libcoro) cleanup and extend the libcoro API to officially
115 allow the creation of empty source contexts.
116 - very experimental workaround for the totally broken netbsd platform.
117 - tried to hack around openbsd bugs.
118
119 4.804 Wed Nov 5 16:36:00 CET 2008
120 - Coro::Debug::new_unix_server would not create a non-blocking listening
121 socket, sometimes causing freezes.
122 - (libcoro) fix misaligned stack points for setjmp and assembly
123 methods, which can cause crashes on x86/x86_64 with a sufficiently
124 aggressive compiler.
125 - new function: Coro::Debug::new_tcp_server.
126 - move ->throw into the Coro class because it only works on coro objects.
127
128 4.803 Mon Nov 3 17:16:12 CET 2008
129 - (libcoro) use a global asm statement to become independent of gcc
130 otpimisations for CORO_ASM (thanks to pippijn for the idea).
131 - try to workaround yet another broken bsd, this time dragonfly.
132
133 4.802 Thu Oct 30 10:56:12 CET 2008
134 - support -fno-omit-frame-pointer on x86 with the assembly method.
135 - tune 01_unblock.t tests a bit.
136
137 4.801 Wed Oct 22 18:33:37 CEST 2008
138 - improve readline speed for very long "lines".
139 - backport to 5.8.8.
140
141 4.8 Thu Oct 2 13:34:40 CEST 2008
142 - new function Coro::AIO::aio_wait.
143 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
144 - greatly speed up and reduce memory usage of Coro::AIO requests.
145 - implement some other µ-optimisations.
146
147 4.749 Mon Sep 29 14:40:12 CEST 2008
148 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
149
150 4.748 Sat Sep 27 14:03:19 CEST 2008
151 - implement, but do not document, PerlIO::cede(granularity).
152 - Coro::Storable forgot to wrap Storable::pstore.
153 - work around the multitude of leaks and memory corruption
154 bugs in PerlIO::via by using our own C-level perliol. As a side
155 effect, Coro::Storable is now much, much, much faster.
156
157 4.747 Tue Sep 23 01:59:41 CEST 2008
158 - fix a per-cv memleak (one empty array was leaked per
159 code reference).
160 - avoid a crash in coro->call|rss when the coroutine was already
161 destroyed (most noticably when using Coro::Debug::ps :)
162 - also protect *Storable::FILE.
163 - push up default storable granularity to 20ms.
164
165 4.746 Sun Sep 21 03:22:20 CEST 2008
166 - be more insistent on locking Storable against reentrancy
167 in Coro::Storable.
168 - move swap_def?v and throw to Coro::State, as documented.
169
170 4.745 Thu Jul 24 00:14:38 CEST 2008
171 - remove debugging code related to MgPV_nolen_const, also try to
172 make it compile with perl 5.8.6 (yes, apple apparently loves
173 outdated software). Reported by John S.
174
175 4.744 Tue Jul 8 22:06:35 CEST 2008
176 - correctly provide default DIE/WARN handlers as documented.
177 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
178 implementations inside perl work fine for us.
179
180 4.743 Mon Jun 16 00:21:57 CEST 2008
181 - when using Coro::EV without running EV::loop it could
182 result in busy-waiting for events, this has been fixed.
183 - reduce codesize and improve performance by using EV_DEFAULT_UC.
184
185 4.742 Sat May 31 14:10:21 CEST 2008
186 - implement a workaround for (some) perl <5.8.8 versions.
187 - require EV 3.3+.
188
189 4.741 Fri May 30 23:33:09 CEST 2008
190 - tell netbsd how utterly broken their imitation of an OS is
191 and refuse to build by default if pthreads are in use.
192 - switch to "s" method on all bsd's by default, as their ucontext
193 stuff seems just too broken.
194 - fix a bug in Coro::Select.
195
196 4.74 Thu May 29 20:05:31 CEST 2008
197 - do not test Coro::LWP for lack of dependencies.
198 - do not test Coro::Select for lack of working perls.
199
200 4.73 Thu May 29 2008
201 - fix a bug in Coro::EV which would cause it to block despite
202 there being runnable coroutines.
203 - sprinkle "no warnings" freely over everything, also suppress
204 warnings for some other modules.
205 - fix typo in WSAEWOULDBLOCK.
206
207 4.72 Sun May 25 05:14:36 CEST 2008
208 - tweak META.yaml a bit, unfortunately, there is no documented way
209 to have optional dependencies with CPAN. doh :(
210 - avoid running some tests on windows because they would fail due to
211 perl bug (broken fork, broken pipes...).
212 - work around perl on windows bugs where perl returns undocumented
213 error codes for sysread, syswrite etc. by taking advantage
214 of AnyEvent's workaround for that problem.
215 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
216 a common perl implementation bug on windows.
217 - use unix domain sockets in testsuite to work around
218 common perl implementation bugs on widows (they are emulated by
219 tcp sockets on windows. ugh.)
220
221 4.71 Sat May 24 20:01:27 CEST 2008
222 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
223 - take advantage of async name resolution of AnyEvent::Util.
224 - work around brutal inet_aton override in Coro::LWP.
225 - take advantage of the readyhook in Coro::EV, for smoother
226 scheduling.
227
228 4.7 Sun May 11 00:32:19 CEST 2008
229 - completely reworked the Coro manpage.
230 - added Coro::AnyEvent, generic event loop integration.
231 - implement cancel, ready and kill commands in Coro::Debug.
232 - document find_coro in Coro::Debug.
233 - incompatible change: rename has_stack to has_cctx.
234 - Coro::AIO and Coro::BDB no longer force event model detection,
235 use AnyEvent::AIO and AnyEvent::BDB.
236
237 4.6 Sat Apr 26 10:05:14 CEST 2008
238 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
239 by the Coro module. It was a mistake to have it in the first place.
240 - (experimental) support for activestate perl 5.10 (method "w").
241 - (experimental) support for strawberry perl (method "a").
242 - coro_sigelem_set did not return a value although it had to,
243 actual impact unknown.
244
245 4.51 Mon Apr 14 13:28:27 CEST 2008
246 - make it compile again on 5.8.
247
248 4.50 Thu Apr 10 09:43:17 CEST 2008
249 - I did it twice! (see 4.49).
250
251 4.49 Sun Apr 6 21:23:31 CEST 2008
252 - grr, instead of compiling the recent changes on 5.10 only they
253 were compiled on 5.8 only.
254
255 4.48 Sun Apr 6 20:36:46 CEST 2008
256 - allow coroutine switches during eval's under 5.10.x, as it seems
257 the parser is a per-interpreter option now, so this is safe
258 (this might fix the odd crashing bug).
259 - drop support for 5.9.x versions: they are dead, jim.
260
261 4.47 Sun Apr 6 00:37:52 CEST 2008
262 - force cctx allocation on API calls: we know we need to force one
263 and gcc actually manages to confuse our heuristic nowadays,
264 leading to crashes and worse.
265 - document force_cctx.
266
267 4.46 Fri Apr 4 22:05:43 CEST 2008
268 - upgrade libcoro, resulting in pthread-backend (which was only created
269 to fulfill the rules of the programming languages shootout).
270
271 4.45 Thu Mar 13 11:55:36 CET 2008
272 - fix a file leaking bug in eg/mhyttpd that would allow
273 downloading of any file (reported by oesi).
274 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
275 (also add testcase).
276 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
277 apparently having loads of fun with that).
278 - implement $state->swap_defsv, ->swap_defav and document ->throw.
279
280 4.4 Wed Feb 13 16:44:29 CET 2008
281 - only assume .cfi pseudo isns exist on GNU/Linux.
282 - add get_timed method to Coro::Channels.
283 - fixed Coro::Debug manpage.
284 - perl 5.11 compatbility improvement based on patch by
285 Andreas König.
286
287 4.37 Sun Jan 20 11:25:23 CET 2008
288 - soften the check for an active parser for perl 5.10 (avoiding
289 "Coro::State::transfer called while parsing" in many cases).
290
291 4.36 Sun Jan 13 10:53:56 CET 2008
292 - reset diehook when terminating from an async_pool as to not
293 trigger any __DIE__ handlers.
294
295 4.35 Sun Jan 13 04:14:13 CET 2008
296 - "bt" debug command now displays any exceptions
297 from longmess and also skips the innermost
298 stackframes, giving more useufl output.
299 - allow backtraces in coroutines blocked in Coro::EV,
300 at a <1% speed hit.
301 - handle localising of $SIG{__WARN__} and __DIE__
302 properly (with a proper amount of dirty hacking).
303
304 4.34 Sat Dec 22 17:49:53 CET 2007
305 - upgrade to EV version 2.0 API.
306
307 4.33 Mon Dec 17 08:36:12 CET 2007
308 - make Coro::AIO etc. loadable in the absence of EV.
309
310 4.32 Mon Dec 17 07:46:02 CET 2007
311 - majorly improved Coro::Handle's performance with Coro::EV.
312 - implemented the until now mythical Coro::BDB module.
313 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
314
315 4.31 Wed Dec 5 12:32:39 CET 2007
316 - remove warn statement form Coro::Util.
317
318 4.3 Tue Dec 4 20:33:14 CET 2007
319 - calls to the idle function could cause stack corruption
320 when the stack changed.
321 - do no longer rely on the presence of EV::DNS if EV is used
322 (because it is gone), but instead take avdantage of EV::ADNS
323 if available.
324 - add ($) prototypes to all functions in Coro::Storable.
325 - use a conventional (safer) idle callback in Coro::EV.
326 - do accept quickly in Coro::Debug to avoid endless loops.
327
328 4.22 Fri Nov 30 16:04:04 CET 2007
329 - really use optimised versions for Event and EV in Coro::Util
330 and Coro::Handle.
331
332 4.21 Sun Nov 25 10:48:59 CET 2007
333 - fix a spurious memory read.
334 - Coro::EV no longer keeps the eventloop "alive".
335
336 4.2 Fri Nov 9 20:47:05 CET 2007
337 - enable/disable tracing from a new coroutine, not a pooled one.
338 - fix a memleak in Coro::Event.
339 - removed killall call from fork_eval.
340 - made sure store_fd is already loaded so that fork_eval
341 does not have to parse autoload in each subprocess.
342 - only use assembly method if -O switch is in $Config{optimize}.
343 - add (optional) Coro::EV module, so far the best event loop module
344 directly supported by Coro.
345 - if the event model is EV, use EV::DNS to resolve
346 stuff in Coro::Util.
347 - don't get confused by multiple event notifications in Coro::Handle.
348 - initial support for EV (libevent interface).
349 - require Event and EV using configure_requires, to force their existance.
350
351 4.13 Wed Oct 24 07:26:45 CEST 2007
352 - add Coro::Storable::blocking_thaw.
353 - use a vastly more complicated technique to localise
354 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
355 - use a coroutine for the idle callback Coro::Event,
356 instead of running Event in the current coroutine context.
357 This also catches recursive invocations.
358 - actually report fork errors in gethostbyname and inet_aton.
359
360 4.11 Thu Oct 11 02:40:24 CEST 2007
361 - port to threaded perls.
362
363 4.1 Thu Oct 11 02:38:16 CEST 2007
364 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
365 be local to each coro (see Coro::State).
366 - incompatible change: for very deep reasons, cede and cede_notself
367 cannot return anything, so nothing will be returned.
368 - possibly bring back 5.10 compatibility (untested).
369 - work around stupid (and wrong) warning on 5.10 :(.
370 - overlay the saved state over the context stack. This saves
371 a few hundred bytes per coroutine on average and also
372 speeds up context switching a bit.
373 - further tune default stack sizes.
374 - (more) correctly calculate stack usage in coro_rss.
375 - Coro::Storable::blocking_* did not properly lock
376 resulting in races between coroutines.
377 - added Coro::Storable::guard.
378 - stopping to trace a coroutine could destroy the cctx of
379 an unrelated coroutine.
380 - explain the relationship between Perl and C coroutines in
381 more detail in Coro::State.
382 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
383 causing a fork per resolve. This also affected Coro::Socket.
384 - switch to a separate stack in $coro->call/eval to avoid
385 invalidating pointers.
386
387 4.03 Sat Oct 6 21:24:00 CEST 2007
388 - added Coro::throw method.
389 - minor code cleanups.
390
391 4.02 Sat Oct 6 02:36:47 CEST 2007
392 - fix a very minor per-coroutine memleak (a single codereference).
393 - fixed a bug where the currently in-use c context would be freed
394 prematurely (can happen only when programs change the stacksize
395 or use tracing).
396 - tracing can no longer keep a coro alive after it terminated.
397 - do static branch prediction in the common path for gcc. gives
398 about 2-5% speed improvement here.
399
400 4.01 Fri Oct 5 22:10:49 CEST 2007
401 - instead of recreating *a* standard output handle we simply
402 use STDOUT, which is faster and hopefully more robust.
403
404 4.0 Fri Oct 5 12:56:00 CEST 2007
405 - incompatibly changed Coro::Storable::freeze.
406 - major new feature: added Coro::Debug, for interactive coroutine
407 debugging, tracing and much more.
408 - major bug fix: unbelievable, but true: $_, $/ and many other
409 "saved" variables actually weren't being saved. This has been fixed,
410 of course, while increasing performance while losing all the save
411 flags.
412 - save flags are gone, and all the api functions dealing with them.
413 - added Coro::Semaphore::adjust.
414 - added Coro::Util::fork_eval.
415 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
416 - added Coro::killall.
417 - reduce initial stack sizes to allow for "micro-coroutines".
418 - better async_pool resource management, moved parts of async_pool
419 handling to XS (major speed improvement).
420 - actually croak before modifying important data structures.
421 - refuse to transfer while compiling.
422 - possibly support eval EXPR better now.
423 - enable assembly per default on linux+bsd x86+amd64.
424 - all internal members were renamed _something for easier subclassing.
425 - many minor tweaks.
426
427 3.63 Wed May 16 14:10:06 CEST 2007
428 - implement handcoded assembly for x86/amd64 SVR ABI.
429
430 3.62 Fri Apr 27 21:36:06 CEST 2007
431 - upgrade libcoro (which might set unwind info correctly).
432 - change default on linux to setjmp/longjmp.
433
434 3.61 Thu Apr 19 12:36:18 CEST 2007
435 - Coro::Storable caused an endless loop when thawing invalid
436 pst's sometimes.
437 - use a Semaphore in Coro::Storable, as Storable doesn't
438 seem to be reentrant (although it is documented to
439 be threadsafe...).
440 - fix Coro::Signal to bring back the original unreliable
441 but stateful semantics.
442 - fixed a lot of typos in Coro.pm (patch submitted by David
443 Steinbrunner, which applied flawlessly).
444
445 3.6 Sat Apr 14 17:13:31 CEST 2007
446 - added some bugfixes to get eg/myhttpd working again.
447 - added Coro::Storable for often-cede'ing freeze/thaw.
448 - try to do a clean exit when a coroutine calls exit
449 (EXPERIMENTAL).
450 - got rid of indirect call through _coro_init.
451 - updated the partly antique examples in eg/ to
452 work again and be a bit less magic, too.
453 - fixed Coro::Signal semantics to work as documented again.
454
455 3.55 Sun Mar 25 01:20:47 CET 2007
456 - add SAVE_DEFFH to save the default filehandle and enable
457 it by default.
458 - finally move socket-operations from Coro::Socket to Coro::Handle
459 to be able to unblock foreign sockets.
460 - add Coro::State::save_also and guarded_save.
461 - add count accessor to Coro::Semaphore.
462 - add Coro::State::cctx_stacksize.
463 - just for the fun of it, do not rely on implicit context,
464 which can dramatically improve performance, but people
465 using windows-process-emulation perls do not care much about
466 performance.
467
468 3.51 Sun Mar 4 14:18:04 CET 2007
469 - fixed a problem when you weakened references to Coro::State's
470 (patch by Yuval Kogman).
471
472 3.501 Wed Feb 28 12:44:07 CET 2007
473 - rename some global symbols as macosx from hell redefines
474 them without asking.
475
476 3.5 Tue Feb 13 20:22:53 CET 2007
477 - do AnyEvent model detection earlier, avoiding problems
478 caused by first using AnyEvent and later Coro::Event.
479 - implement and document Coro::Event event objects.
480 - fix a potential problem in Coro::Event causing crashes.
481 - initialise PL_comppad when creating a new coroutine,
482 avoids crashes on early coro destruction.
483
484 3.41 Mon Jan 22 19:19:49 CET 2007
485 - readline on Coro::Handle did not support undefined $/,
486 nor did it deliver partial lines on EOF or error.
487 - implement malloc fallback for stack allocation because
488 stupid broken idiotic OSX has a stupid broken
489 idiotic fits-the-whole-os mmap "implementation" and
490 my dick feels longer if Coro is portable even to
491 obsolete platforms.
492
493 3.4 Fri Jan 19 21:52:54 CET 2007
494 - remove t/09_timer.t, as it isn't really testing much
495 but was rather flaky in practise.
496 - async_pool coro would keep arguments and callback alive until
497 it was reused.
498 - cancellation of a coroutine could cause spurious idle calls
499 in cede_notself.
500
501 3.3 Sat Jan 6 03:45:00 CET 2007
502 - implement $coro->on_destroy.
503 - Coro::Event blocking semantics have been changed,
504 documented and - hopefully - improved.
505 - fix nice adding, not subtracting, from priority.
506 - fix ->prio and api_is_ready (patch by Mark Hinds).
507 - fixed an assert ("... == prev__cctx->idle_te")
508 that could errronously trigger.
509 - fix various large and small memleaks.
510 - use a (hopefully) more stable cancel implementation
511 that immediately frees the coroutine data.
512 - cede/cede_notself return a status now.
513 - added Coro::guard function.
514 - added a global coroutine pool for jobs (on my machine,
515 I can create and execute 48k simple coros/s with async,
516 and 128k coros with async_pool).
517 - Coro::AIO now uses the coroutine priority as io priority.
518
519 3.2 Fri Dec 22 05:07:09 CET 2006
520 - improve portability to slightly older perls.
521 - use cleaner coroutine destruction.
522 - simplify configuration for users.
523 - optionally (unrecommended) prefer perl functions over
524 their coro replacements.
525
526 3.11 Tue Dec 5 13:11:24 CET 2006
527 - fixed some bogus assert's, but as perl.h disables assert even
528 without NDEBUG (thank you very much), not too many people should
529 notice (that did include myself). Andreas König noticed, though :)
530 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
531 it. Also noticed by Andreas König :)
532 - save/restore tainted status.
533 - verified to pass the testsuite on my 5.9.5.
534
535 3.1 Mon Dec 4 23:03:40 CET 2006
536 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
537 - incompatible change: transfer flags are now per-state.
538 - give a better error message on deadlock.
539 - document Coro::nready.
540 - enhanced testsuite.
541
542 3.01 Sun Dec 3 23:47:42 CET 2006
543 - forgot to include Coro::Timer.
544
545 3.0 Sun Dec 3 22:57:25 CET 2006
546 - the "FINALLY COMPLETELY STABLE" release (coming soon:
547 the "FAMOUS LAST WORDS" release).
548 - implement a new stack sharing algorithm, which uses a stack
549 pool (size currently hardcoded).
550 - make stack sharing mandatory (it no longer uses a heuristic).
551 - eval/die no longer cause weird problems under heavy use.
552 - Coro::Event could cause livelocks if it was used but
553 no Coro::Event watchers were used.
554 - Coro::Event now uses asynccheck as prepare does not
555 check for changed watchers.
556 - Coro::Event allows multiple waiting coros and will wake up one
557 per event.
558 - Coro::Event should be cleaner and more efficient now.
559 - new utility function Coro::unblock_sub.
560 - document the sad fact that Event is no longer reentrant.
561 - putting a coroutine into the ready queue twice could under
562 some circumstances lead to stack corruption.
563 - minor incompatibility: subclassing Coro::State is supported
564 directly now without going through a _coro_state member.
565 - state/coro switching is much faster now.
566 - very minor optimisations and code/documentation cleanup.
567 - avoid problems due to compiler inlining.
568 - removed timers from Coro::Timer -> use AnyEvent instead.
569 - replaced Coro::idle coroutine by (cleaner) idle handler.
570 - updated to newest libcoro.
571 - implement enhanced support for valgrind.
572 - implement is_ready and return value for ready.
573 - removed Coro::Cont, it was a misnomer (it's generators),
574 and it was somewhat annoying to get it right. Will come back
575 if somebody asks for it :)
576 - many ->wait methods and Coro::Event could return
577 spuriously without the event having happened.
578
579 2.5 Tue Nov 7 12:22:33 CET 2006
580 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
581 use AnyEvent, moved them to Coro/.
582 - added Coro::LWP which contains all the uglyness required to
583 make LWP non-blocking.
584 - should work with perl 5.9.x now (Andreas König made me do it).
585 - fixed another bug in Coro::Select when the timeout was undef.
586 - reuse PL_start_env for all coros, saves some memory per coroutine.
587 - manage PL_top_env differently, hopefully to avoid panic: top_env.
588 - timeout argument was not properly used in Coro::Socket.
589 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
590 - emulate undocumented(!) functionality of IO::Socket required
591 by LWP(!!).
592 - updated eg/lwp to work with newer lwp's.
593 - remove "FATAL: uncaught exception" prefix. Coroutines that die
594 kill the whole process, just as exceptions in the main "coroutine"
595 did already.
596
597 2.1 Wed Nov 1 23:01:13 CET 2006
598 - fix a long-standing bug in Coro::Select where select with
599 zero timeout would instead change the current default filehandle.
600 - use a simpler and hopefully more robust way to clone padlists
601 (uses less memory and a perl function instead of our own).
602 - coro can now create a stack guard area on many architectures.
603 - Coro::AIO properly reexports additional functions from IO::AIO.
604 - updated libcoro with a workaround for OS X,
605 pach and testing by Michael Schwern.
606
607 2.0 Tue Oct 24 05:47:17 CEST 2006
608 - support additional aio requests in Coro::AIO.
609
610 1.9 (never properly released due to a glitch)
611
612 1.8 Thu Feb 2 00:59:06 CET 2006
613 - applied suggested patch by SAMV to avoid problems during stupid
614 mark & sweep gc run.
615 - applied patch by Scott Walters for 5.9.3 compatibility.
616
617 1.7 Tue Dec 27 01:41:58 CET 2005
618 - added Coro::AIO, a thin wrapper around IO::AIO.
619 - improved Makefile.PL explanations.
620
621 1.6 Mon Dec 12 21:30:05 CET 2005
622 - additionally save PM_curpm might fix as-of-yet
623 unseen problems with regex matches being attributed to the
624 wrong package.
625 - add t/10_bugs.t, which currently checks against imho broken
626 perls that use 0.26MB of stack space per Perl_magic_get
627 invocation (newer linuxes) as opposed to the 0.0002MB perl
628 normally uses.
629 - make stacksize configurable for the ultimate debian experience.
630
631 1.51 Mon Dec 12 18:48:36 CET 2005
632 - remove debugging warn() accidentally left in Coro::Select.
633
634 1.5 Tue Nov 29 13:32:44 CET 2005
635 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
636 - fix Coro::Select 'select'.
637 - strict'ify some modules.
638 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
639 - use XSLoader in selected modules.
640 - remove some 5.6 compatibility cruft.
641
642 1.4 Tue Sep 6 00:11:05 CEST 2005
643 - libcoro did not take into account the trampoline on amd64,
644 when 'l'inux method was used.
645
646 1.31 Tue Aug 30 23:31:33 CEST 2005
647 - some portability fixes/workarounds.
648
649 1.3 Sat Aug 20 03:08:56 CEST 2005
650 - no code changes. module seems to work fine.
651 - improve Coro::State docs, remove reference to nonexisting
652 Coro::State::flush method.
653 - no longer autodetect windows, present it as an option instead.
654
655 1.2 Mon May 16 02:00:55 CEST 2005
656 - included libcoro.c earlier in State.xs, might improve portability.
657 - use a faster and possibly more stable Coro::Cont implementation.
658 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
659 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
660 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
661
662 1.11 Thu Mar 3 18:00:52 CET 2005
663 - change of contact address.
664
665 1.1 Tue Feb 22 20:51:16 CET 2005
666 - support [l]inux method on amd64.
667 - allow some context switches while compiling/eval'ing.
668
669 1.0 Tue Aug 10 03:47:27 CEST 2004
670 - Event 0.89 fixes the bug again.
671 - fix the remaining(??) memleak. I didn't free the cache-AV for
672 padlists, so one AV per created anon-sub leaked.
673
674 0.97 Fri May 14 15:23:32 CEST 2004
675 - remove stupid gcc extension that was an experiment only anyway.
676
677 0.96 Thu May 13 18:09:29 CEST 2004
678 - tell people if they have yet another broken Event variant.
679 - made Coro to work _better_ with threads (it seems that, as long
680 as you use Coros in one thread only everything will work fine.
681 Looking at the perl source the problem might be missing locking
682 between Coro and some perl internal routines. That's not easily
683 fixable).
684 - (hopefully) got rid of the static xnext inside transfer.
685
686 0.95 Sun Feb 15 03:20:28 CET 2004
687 - removed Coro::State::flush, implemented a new and much
688 faster caching mechanism by attaching magic to the CV.
689 - a memleak with (real) closures remains.
690 - dramatically reduced size of stacks to 65536 (linux/x86), after
691 observing that even myhttpd never had more than a kilobyte
692 of stack in use (stack is still only physically allocated in
693 pagesize increments for systems suporting mmap).
694 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
695 - fix for threaded perl.
696
697 0.9 Sun Nov 23 04:32:39 CET 2003
698 - fix a few very hard to track down but (of course) stupid
699 memory leaks.
700 - stringify version number to avoid locale problems :(.
701 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
702 hopefully ignoring it "just works".
703
704 0.8 Wed Nov 5 19:38:40 CET 2003
705 - port to perl5.005_03, but only 5.8.x+ is supported!
706 - honor LocalAddr even without LocalPort in Coro::Socket.
707
708 0.7 Tue May 27 03:12:38 CEST 2003
709 - the version jump indicates some level of testing, not gobs
710 of new features.
711 - uh... I found the prompt function in ExtUtils::MakeMaker.
712 Highly correct stuff, that is...
713 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
714 the fix is not well understood by the author, so beware :(.
715
716 0.652 Thu May 8 02:54:46 CEST 2003
717 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
718 cause version 4 doesn't have ucontext.
719 - Benjamin Reed reported that setjmp works fine on darwin,
720 so preselect it.
721
722 0.651 Sat Mar 29 15:00:23 CET 2003
723 - fix a bug in Coro::Handle where some bytes could get lost
724 on reading, reported by jason@nichego.net.
725
726 0.65 Sun Mar 23 00:08:26 CET 2003
727 - added a README.linux-glibc.
728 - new module Coro::Select.
729 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
730 (I never claimed that I know what I am doing ;).
731 - set default to setjmp/longjmp on non-x86-linux.
732
733 0.6 Thu Nov 21 11:09:06 CET 2002
734 - made lazy context switching the default.
735 - fixed the bug where SemaphoreSet::guard
736 was the same as timed_guard and timed_guard was missing.
737 - fix a memleak in Coro::Timer.
738 - reclassified context sharing as not experimental.
739
740 0.534 Sun Apr 14 03:05:12 CEST 2002
741 - fixed backspace => tab in header parsing (myhttpd).
742 - added eg/readline, for elmex.
743 - Coro::Event now calls ->start on first call instead of
744 ->again, which created wrong timeouts for the first ->next.
745 - fixed a bug where Coro::Socket returned a closed socket instead
746 of nothing on ECONNREFUSED and similar error conditions.
747
748 0.533 Mon Feb 18 18:49:40 CET 2002
749 - adapt to finally working Event-0.85.
750 - fixed buggy event initialization (trapped under windows, but how
751 could it possibly have worked anywhere?)
752
753 0.532 Wed Jan 16 02:45:32 CET 2002
754 - removed the reference to pp_entersub, might work on
755 (some) win32 perls now (testsuite works under cygwin,
756 without Event, which freezes).
757
758 0.531 Mon Dec 10 22:18:44 CET 2001
759 - Fixed a bug in SemaphoreSet::timed_down.
760
761 0.53 Tue Nov 27 21:11:13 CET 2001
762 - some tests for Event.
763 - slightly different internal architecture to get refcnt into the object
764 and not the reference. might fix a few bugs, certainly did introduce
765 new ones.
766 - Coro::Timer (independent of Event).
767 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
768
769 0.52 Tue Nov 6 21:36:18 CET 2001
770 - ported to cygwin (trivial).
771 patch by Gerrit P. Haase <gp@familiehaase.de>.
772 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
773
774 0.51 Thu Nov 1 20:39:01 CET 2001
775 - terminate/cancel now work properly (otherwise termination
776 could cause a "next coroutine is not and contains not..."
777 error.
778 - added Coro::Socket::shutdown.
779 - Coro::Event::loop no is the same as Event::loop.
780 - implemented terminate with args + join.
781
782 0.5 Fri Sep 28 16:15:35 CEST 2001
783 - fixed "print" on a Coro::Handle. The print method worked.
784 - small tweaks (seem to reduce memory consumption a lot)
785 in various modules.
786 - splendid use of "no warnings" scattered throughout.
787 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
788
789 0.49 Sun Sep 16 02:42:45 CEST 2001
790 - changed some method calls to function calls for speed
791 inside Coro::Handle.
792 - make Coro::Handle use an array instead of a hash for
793 speed reasons.
794 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
795 sjlj and ucontext should now work.
796 - IRIX-specific port for libcoro.
797 - swapped order of accept results to match IO::Socket.
798 - changed getsock/peername to sock/peername to match IO::Socket.
799 - fixed a bug that caused segfault when returning to main task
800 under some circumstances.
801 - other bugfixes.
802
803 0.45 Sun Sep 2 02:54:01 CEST 2001
804 - new method Coro::Handle::timeout.
805 - corrected speling of set...name to get...name in Coro::Socket.
806 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
807 - Coro::Handle now supports fileno and FILENO.
808 - added eg/myhttpd, a web server.
809 - Coro::Socket now accepts numeric ports as well as "port(number)"
810 syntax.
811 - moved some scheduling primitives into xs code. more to come.
812 - new simple priority system.
813 - implemented Coro::Event using XS, almost four times faster!
814 - small memory corruption problem fixed (boy that was difficult).
815
816 0.13 Wed Aug 8 16:53:07 CEST 2001
817 - new method Coro::Handle::readline.
818 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
819 (see eg/lwp on how to make LWP non-blocking).
820 - renamed Coro::Socket::new_inet to new.
821 - added Coro::Util, some utility functions.
822 - die/eval now works better (still get annoying (but true ;)
823 "Callback called exit" messages).
824 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
825 - much better Coro::Socket support.
826
827 0.12 Fri Jul 27 04:19:01 CEST 2001
828 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
829 - very experimental stack sharing algorithm. not 100% safe but
830 should work well in practise ;)
831 - added Coro::RWLock.
832 - Coro::Specific now works.
833
834 0.11 Tue Jul 24 22:49:21 CEST 2001
835 - added specialized hack for newer and older linux versions (fast).
836 - renamed Coro::Event::IO to Coro::Handle.
837 - new module Coro::Socket.
838
839 0.1 Tue Jul 24 01:47:53 CEST 2001
840 - release candidate 3. A rather new internal structure :(
841 - the great renaming: Damian Conway gave me a suitable replacement
842 for yield.
843 - added Coro/libcoro, a portable coroutine implementation for C
844 (not even perl-dependent!!!), which can be used by Coro::State.
845 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
846 perl-5.6.
847 - new function Coro::Event::idle.
848 - the idle process is now overriden
849 by default in Coro::Event.
850 - Coro::Channel now enforces the size.
851 - canceling events now works in all cases.
852 - Coro state now include $_ and $@.
853 - yet another bug workaround that I do not really understand :(
854 - new module Coro::Event::IO (very undocumented) to do
855 non-blocking i/o.
856 - performance tuning.
857
858 0.08 Thu Jul 19 06:13:25 CEST 2001
859 - release candidate 2 (new functionality).
860 - add Coro::State::flush function.
861 - transfer now supports a flags value (mostly for speed ATM).
862 - might compile and work in the presence of threads now.
863 - continuations have a saner syntax.
864 - no more memleaks.
865
866 0.07 Tue Jul 17 17:40:18 CEST 2001
867 - release candidate 1 ;)
868 - slightly nicer code.
869 - fixed a scheduling bug in Coro::Event.
870
871 0.06 Tue Jul 17 04:23:24 CEST 2001
872 - ok, I found the showstopper - the same sub must not be
873 re-entered in two different coroutines, otherwise => crash. I
874 see no easy solution to this problem, except by walking the call
875 chaing and saving/restoring the cv's, which is what I do now.
876 - memory leaks still latent, especially at thread termination.
877 - Coro::Event now works (haha).
878
879 0.05 Sun Jul 15 17:32:20 CEST 2001
880 - fixed a few issues in Coro::Event.
881 - I forgot to include Coro::Event in 0.04 :(:(:(
882
883 0.04 Sun Jul 15 05:24:59 CEST 2001
884 - @_ is now properly localized.
885 - Coro::State is now easier subclassable.
886 - Coro::Cont now coroutine-aware.
887 - Coro::Specific is a low-overhead module to create
888 coroutine-specific vars.
889 - Coro::Event provides a simple interface to Event.
890
891 0.03 Fri Jul 13 14:51:52 CEST 2001
892 - transfer() now implemented in XS (beware).
893 - new module Coro::Cont for really faked continuations.
894 - big internal architecture changes: Coro::State is now
895 really low-level and can thus be used to implement other
896 interesting things, While "Coro::" implements a process-like
897 model. Still crude and subject to change.
898 - $_ and $@ are no longer being localized.
899
900 0.02 Tue Jul 10 01:38:17 CEST 2001
901 - implemented "async" attribute.
902 - $_ and $@ are now localized.
903 - added Coro::Channel.
904 - more testcases, still no docs.
905
906 0.01 Tue Jul 3 02:18:41 CEST 2001
907 - original version; copied from Convert::Scalar.
908