ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.410
Committed: Fri Nov 14 06:41:41 2008 UTC (15 years, 6 months ago) by root
Branch: MAIN
Changes since 1.409: +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 5.0
9 - NEW ARCHITECTURE: use 4.912 for the stable version.
10 - directly patch the entersub opcode calling ssl functions (cede,
11 transfer and so on). this does speed up context switching by
12 up to 10%. More importanly, it frees us from the hardcoded
13 behaviour of entersub, so we might actually be able to return
14 something from those functions.
15 - take advanatge of __builtin_frame_address on gcc.
16
17 4.912 Thu Nov 13 18:31:23 CET 2008
18 - minor cleanups.
19 - use much larger stacks on linux and perl < 5.8.8.
20 - Coro::Debug::new_unix_server did not unlink the socket
21 when destroyed.
22
23 4.911 Tue Nov 11 04:26:17 CET 2008
24 - "port" to threaded perls.
25
26 4.91 Mon Nov 10 05:36:38 CET 2008
27 - the ->throw exception object no longer leaks.
28 - creating a new cctx leaked a scopestack entry (memleak).
29 - new coroutines didn't get created with a zero flags field
30 (unknown impact).
31 - calling ->throw on a not-yet-started coroutine should now work
32 instead of being ignored.
33 - ->throw is now supported on Coro::State objects.
34 - clean up cctx creation code a bit.
35 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
36
37 4.9 Sat Nov 8 17:45:27 CET 2008
38 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
39 - (libcoro) no longer leak threads in the experimental pthread backend,
40 also speed it up considerably.
41 - (libcoro) do not rely on makecontext passing void *'s unscathed.
42 - fix compiletime dependencies on libcoro in the Makefile.
43 - cctx_count wasn't always updated properly.
44 - Coro::State::cctx_stacksize wasn't applied correctly.
45 - new function Coro::State::cctx_max_idle.
46 - the default max number of idle C contexts is now 4.
47 - (libcoro) try harder to get _setjmp/_longjmp.
48 - (libcoro) cleanup and extend the libcoro API to officially
49 allow the creation of empty source contexts.
50 - very experimental workaround for the totally broken netbsd platform.
51 - tried to hack around openbsd bugs.
52
53 4.804 Wed Nov 5 16:36:00 CET 2008
54 - Coro::Debug::new_unix_server would not create a non-blocking listening
55 socket, sometimes causing freezes.
56 - (libcoro) fix misaligned stack points for setjmp and assembly
57 methods, which can cause crashes on x86/x86_64 with a sufficiently
58 aggressive compiler.
59 - new function: Coro::Debug::new_tcp_server.
60 - move ->throw into the Coro class because it only works on coro objects.
61
62 4.803 Mon Nov 3 17:16:12 CET 2008
63 - (libcoro) use a global asm statement to become independent of gcc
64 otpimisations for CORO_ASM (thanks to pippijn for the idea).
65 - try to workaround yet another broken bsd, this time dragonfly.
66
67 4.802 Thu Oct 30 10:56:12 CET 2008
68 - support -fno-omit-frame-pointer on x86 with the assembly method.
69 - tune 01_unblock.t tests a bit.
70
71 4.801 Wed Oct 22 18:33:37 CEST 2008
72 - improve readline speed for very long "lines".
73 - backport to 5.8.8.
74
75 4.8 Thu Oct 2 13:34:40 CEST 2008
76 - new function Coro::AIO::aio_wait.
77 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
78 - greatly speed up and reduce memory usage of Coro::AIO requests.
79 - implement some other µ-optimisations.
80
81 4.749 Mon Sep 29 14:40:12 CEST 2008
82 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
83
84 4.748 Sat Sep 27 14:03:19 CEST 2008
85 - implement, but do not document, PerlIO::cede(granularity).
86 - Coro::Storable forgot to wrap Storable::pstore.
87 - work around the multitude of leaks and memory corruption
88 bugs in PerlIO::via by using our own C-level perliol. As a side
89 effect, Coro::Storable is now much, much, much faster.
90
91 4.747 Tue Sep 23 01:59:41 CEST 2008
92 - fix a per-cv memleak (one empty array was leaked per
93 code reference).
94 - avoid a crash in coro->call|rss when the coroutine was already
95 destroyed (most noticably when using Coro::Debug::ps :)
96 - also protect *Storable::FILE.
97 - push up default storable granularity to 20ms.
98
99 4.746 Sun Sep 21 03:22:20 CEST 2008
100 - be more insistent on locking Storable against reentrancy
101 in Coro::Storable.
102 - move swap_def?v and throw to Coro::State, as documented.
103
104 4.745 Thu Jul 24 00:14:38 CEST 2008
105 - remove debugging code related to MgPV_nolen_const, also try to
106 make it compile with perl 5.8.6 (yes, apple apparently loves
107 outdated software). Reported by John S.
108
109 4.744 Tue Jul 8 22:06:35 CEST 2008
110 - correctly provide default DIE/WARN handlers as documented.
111 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
112 implementations inside perl work fine for us.
113
114 4.743 Mon Jun 16 00:21:57 CEST 2008
115 - when using Coro::EV without running EV::loop it could
116 result in busy-waiting for events, this has been fixed.
117 - reduce codesize and improve performance by using EV_DEFAULT_UC.
118
119 4.742 Sat May 31 14:10:21 CEST 2008
120 - implement a workaround for (some) perl <5.8.8 versions.
121 - require EV 3.3+.
122
123 4.741 Fri May 30 23:33:09 CEST 2008
124 - tell netbsd how utterly broken their imitation of an OS is
125 and refuse to build by default if pthreads are in use.
126 - switch to "s" method on all bsd's by default, as their ucontext
127 stuff seems just too broken.
128 - fix a bug in Coro::Select.
129
130 4.74 Thu May 29 20:05:31 CEST 2008
131 - do not test Coro::LWP for lack of dependencies.
132 - do not test Coro::Select for lack of working perls.
133
134 4.73 Thu May 29 2008
135 - fix a bug in Coro::EV which would cause it to block despite
136 there being runnable coroutines.
137 - sprinkle "no warnings" freely over everything, also suppress
138 warnings for some other modules.
139 - fix typo in WSAEWOULDBLOCK.
140
141 4.72 Sun May 25 05:14:36 CEST 2008
142 - tweak META.yaml a bit, unfortunately, there is no documented way
143 to have optional dependencies with CPAN. doh :(
144 - avoid running some tests on windows because they would fail due to
145 perl bug (broken fork, broken pipes...).
146 - work around perl on windows bugs where perl returns undocumented
147 error codes for sysread, syswrite etc. by taking advantage
148 of AnyEvent's workaround for that problem.
149 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
150 a common perl implementation bug on windows.
151 - use unix domain sockets in testsuite to work around
152 common perl implementation bugs on widows (they are emulated by
153 tcp sockets on windows. ugh.)
154
155 4.71 Sat May 24 20:01:27 CEST 2008
156 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
157 - take advantage of async name resolution of AnyEvent::Util.
158 - work around brutal inet_aton override in Coro::LWP.
159 - take advantage of the readyhook in Coro::EV, for smoother
160 scheduling.
161
162 4.7 Sun May 11 00:32:19 CEST 2008
163 - completely reworked the Coro manpage.
164 - added Coro::AnyEvent, generic event loop integration.
165 - implement cancel, ready and kill commands in Coro::Debug.
166 - document find_coro in Coro::Debug.
167 - incompatible change: rename has_stack to has_cctx.
168 - Coro::AIO and Coro::BDB no longer force event model detection,
169 use AnyEvent::AIO and AnyEvent::BDB.
170
171 4.6 Sat Apr 26 10:05:14 CEST 2008
172 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
173 by the Coro module. It was a mistake to have it in the first place.
174 - (experimental) support for activestate perl 5.10 (method "w").
175 - (experimental) support for strawberry perl (method "a").
176 - coro_sigelem_set did not return a value although it had to,
177 actual impact unknown.
178
179 4.51 Mon Apr 14 13:28:27 CEST 2008
180 - make it compile again on 5.8.
181
182 4.50 Thu Apr 10 09:43:17 CEST 2008
183 - I did it twice! (see 4.49).
184
185 4.49 Sun Apr 6 21:23:31 CEST 2008
186 - grr, instead of compiling the recent changes on 5.10 only they
187 were compiled on 5.8 only.
188
189 4.48 Sun Apr 6 20:36:46 CEST 2008
190 - allow coroutine switches during eval's under 5.10.x, as it seems
191 the parser is a per-interpreter option now, so this is safe
192 (this might fix the odd crashing bug).
193 - drop support for 5.9.x versions: they are dead, jim.
194
195 4.47 Sun Apr 6 00:37:52 CEST 2008
196 - force cctx allocation on API calls: we know we need to force one
197 and gcc actually manages to confuse our heuristic nowadays,
198 leading to crashes and worse.
199 - document force_cctx.
200
201 4.46 Fri Apr 4 22:05:43 CEST 2008
202 - upgrade libcoro, resulting in pthread-backend (which was only created
203 to fulfill the rules of the programming languages shootout).
204
205 4.45 Thu Mar 13 11:55:36 CET 2008
206 - fix a file leaking bug in eg/mhyttpd that would allow
207 downloading of any file (reported by oesi).
208 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
209 (also add testcase).
210 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
211 apparently having loads of fun with that).
212 - implement $state->swap_defsv, ->swap_defav and document ->throw.
213
214 4.4 Wed Feb 13 16:44:29 CET 2008
215 - only assume .cfi pseudo isns exist on GNU/Linux.
216 - add get_timed method to Coro::Channels.
217 - fixed Coro::Debug manpage.
218 - perl 5.11 compatbility improvement based on patch by
219 Andreas König.
220
221 4.37 Sun Jan 20 11:25:23 CET 2008
222 - soften the check for an active parser for perl 5.10 (avoiding
223 "Coro::State::transfer called while parsing" in many cases).
224
225 4.36 Sun Jan 13 10:53:56 CET 2008
226 - reset diehook when terminating from an async_pool as to not
227 trigger any __DIE__ handlers.
228
229 4.35 Sun Jan 13 04:14:13 CET 2008
230 - "bt" debug command now displays any exceptions
231 from longmess and also skips the innermost
232 stackframes, giving more useufl output.
233 - allow backtraces in coroutines blocked in Coro::EV,
234 at a <1% speed hit.
235 - handle localising of $SIG{__WARN__} and __DIE__
236 properly (with a proper amount of dirty hacking).
237
238 4.34 Sat Dec 22 17:49:53 CET 2007
239 - upgrade to EV version 2.0 API.
240
241 4.33 Mon Dec 17 08:36:12 CET 2007
242 - make Coro::AIO etc. loadable in the absence of EV.
243
244 4.32 Mon Dec 17 07:46:02 CET 2007
245 - majorly improved Coro::Handle's performance with Coro::EV.
246 - implemented the until now mythical Coro::BDB module.
247 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
248
249 4.31 Wed Dec 5 12:32:39 CET 2007
250 - remove warn statement form Coro::Util.
251
252 4.3 Tue Dec 4 20:33:14 CET 2007
253 - calls to the idle function could cause stack corruption
254 when the stack changed.
255 - do no longer rely on the presence of EV::DNS if EV is used
256 (because it is gone), but instead take avdantage of EV::ADNS
257 if available.
258 - add ($) prototypes to all functions in Coro::Storable.
259 - use a conventional (safer) idle callback in Coro::EV.
260 - do accept quickly in Coro::Debug to avoid endless loops.
261
262 4.22 Fri Nov 30 16:04:04 CET 2007
263 - really use optimised versions for Event and EV in Coro::Util
264 and Coro::Handle.
265
266 4.21 Sun Nov 25 10:48:59 CET 2007
267 - fix a spurious memory read.
268 - Coro::EV no longer keeps the eventloop "alive".
269
270 4.2 Fri Nov 9 20:47:05 CET 2007
271 - enable/disable tracing from a new coroutine, not a pooled one.
272 - fix a memleak in Coro::Event.
273 - removed killall call from fork_eval.
274 - made sure store_fd is already loaded so that fork_eval
275 does not have to parse autoload in each subprocess.
276 - only use assembly method if -O switch is in $Config{optimize}.
277 - add (optional) Coro::EV module, so far the best event loop module
278 directly supported by Coro.
279 - if the event model is EV, use EV::DNS to resolve
280 stuff in Coro::Util.
281 - don't get confused by multiple event notifications in Coro::Handle.
282 - initial support for EV (libevent interface).
283 - require Event and EV using configure_requires, to force their existance.
284
285 4.13 Wed Oct 24 07:26:45 CEST 2007
286 - add Coro::Storable::blocking_thaw.
287 - use a vastly more complicated technique to localise
288 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
289 - use a coroutine for the idle callback Coro::Event,
290 instead of running Event in the current coroutine context.
291 This also catches recursive invocations.
292 - actually report fork errors in gethostbyname and inet_aton.
293
294 4.11 Thu Oct 11 02:40:24 CEST 2007
295 - port to threaded perls.
296
297 4.1 Thu Oct 11 02:38:16 CEST 2007
298 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
299 be local to each coro (see Coro::State).
300 - incompatible change: for very deep reasons, cede and cede_notself
301 cannot return anything, so nothing will be returned.
302 - possibly bring back 5.10 compatibility (untested).
303 - work around stupid (and wrong) warning on 5.10 :(.
304 - overlay the saved state over the context stack. This saves
305 a few hundred bytes per coroutine on average and also
306 speeds up context switching a bit.
307 - further tune default stack sizes.
308 - (more) correctly calculate stack usage in coro_rss.
309 - Coro::Storable::blocking_* did not properly lock
310 resulting in races between coroutines.
311 - added Coro::Storable::guard.
312 - stopping to trace a coroutine could destroy the cctx of
313 an unrelated coroutine.
314 - explain the relationship between Perl and C coroutines in
315 more detail in Coro::State.
316 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
317 causing a fork per resolve. This also affected Coro::Socket.
318 - switch to a separate stack in $coro->call/eval to avoid
319 invalidating pointers.
320
321 4.03 Sat Oct 6 21:24:00 CEST 2007
322 - added Coro::throw method.
323 - minor code cleanups.
324
325 4.02 Sat Oct 6 02:36:47 CEST 2007
326 - fix a very minor per-coroutine memleak (a single codereference).
327 - fixed a bug where the currently in-use c context would be freed
328 prematurely (can happen only when programs change the stacksize
329 or use tracing).
330 - tracing can no longer keep a coro alive after it terminated.
331 - do static branch prediction in the common path for gcc. gives
332 about 2-5% speed improvement here.
333
334 4.01 Fri Oct 5 22:10:49 CEST 2007
335 - instead of recreating *a* standard output handle we simply
336 use STDOUT, which is faster and hopefully more robust.
337
338 4.0 Fri Oct 5 12:56:00 CEST 2007
339 - incompatibly changed Coro::Storable::freeze.
340 - major new feature: added Coro::Debug, for interactive coroutine
341 debugging, tracing and much more.
342 - major bug fix: unbelievable, but true: $_, $/ and many other
343 "saved" variables actually weren't being saved. This has been fixed,
344 of course, while increasing performance while losing all the save
345 flags.
346 - save flags are gone, and all the api functions dealing with them.
347 - added Coro::Semaphore::adjust.
348 - added Coro::Util::fork_eval.
349 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
350 - added Coro::killall.
351 - reduce initial stack sizes to allow for "micro-coroutines".
352 - better async_pool resource management, moved parts of async_pool
353 handling to XS (major speed improvement).
354 - actually croak before modifying important data structures.
355 - refuse to transfer while compiling.
356 - possibly support eval EXPR better now.
357 - enable assembly per default on linux+bsd x86+amd64.
358 - all internal members were renamed _something for easier subclassing.
359 - many minor tweaks.
360
361 3.63 Wed May 16 14:10:06 CEST 2007
362 - implement handcoded assembly for x86/amd64 SVR ABI.
363
364 3.62 Fri Apr 27 21:36:06 CEST 2007
365 - upgrade libcoro (which might set unwind info correctly).
366 - change default on linux to setjmp/longjmp.
367
368 3.61 Thu Apr 19 12:36:18 CEST 2007
369 - Coro::Storable caused an endless loop when thawing invalid
370 pst's sometimes.
371 - use a Semaphore in Coro::Storable, as Storable doesn't
372 seem to be reentrant (although it is documented to
373 be threadsafe...).
374 - fix Coro::Signal to bring back the original unreliable
375 but stateful semantics.
376 - fixed a lot of typos in Coro.pm (patch submitted by David
377 Steinbrunner, which applied flawlessly).
378
379 3.6 Sat Apr 14 17:13:31 CEST 2007
380 - added some bugfixes to get eg/myhttpd working again.
381 - added Coro::Storable for often-cede'ing freeze/thaw.
382 - try to do a clean exit when a coroutine calls exit
383 (EXPERIMENTAL).
384 - got rid of indirect call through _coro_init.
385 - updated the partly antique examples in eg/ to
386 work again and be a bit less magic, too.
387 - fixed Coro::Signal semantics to work as documented again.
388
389 3.55 Sun Mar 25 01:20:47 CET 2007
390 - add SAVE_DEFFH to save the default filehandle and enable
391 it by default.
392 - finally move socket-operations from Coro::Socket to Coro::Handle
393 to be able to unblock foreign sockets.
394 - add Coro::State::save_also and guarded_save.
395 - add count accessor to Coro::Semaphore.
396 - add Coro::State::cctx_stacksize.
397 - just for the fun of it, do not rely on implicit context,
398 which can dramatically improve performance, but people
399 using windows-process-emulation perls do not care much about
400 performance.
401
402 3.51 Sun Mar 4 14:18:04 CET 2007
403 - fixed a problem when you weakened references to Coro::State's
404 (patch by Yuval Kogman).
405
406 3.501 Wed Feb 28 12:44:07 CET 2007
407 - rename some global symbols as macosx from hell redefines
408 them without asking.
409
410 3.5 Tue Feb 13 20:22:53 CET 2007
411 - do AnyEvent model detection earlier, avoiding problems
412 caused by first using AnyEvent and later Coro::Event.
413 - implement and document Coro::Event event objects.
414 - fix a potential problem in Coro::Event causing crashes.
415 - initialise PL_comppad when creating a new coroutine,
416 avoids crashes on early coro destruction.
417
418 3.41 Mon Jan 22 19:19:49 CET 2007
419 - readline on Coro::Handle did not support undefined $/,
420 nor did it deliver partial lines on EOF or error.
421 - implement malloc fallback for stack allocation because
422 stupid broken idiotic OSX has a stupid broken
423 idiotic fits-the-whole-os mmap "implementation" and
424 my dick feels longer if Coro is portable even to
425 obsolete platforms.
426
427 3.4 Fri Jan 19 21:52:54 CET 2007
428 - remove t/09_timer.t, as it isn't really testing much
429 but was rather flaky in practise.
430 - async_pool coro would keep arguments and callback alive until
431 it was reused.
432 - cancellation of a coroutine could cause spurious idle calls
433 in cede_notself.
434
435 3.3 Sat Jan 6 03:45:00 CET 2007
436 - implement $coro->on_destroy.
437 - Coro::Event blocking semantics have been changed,
438 documented and - hopefully - improved.
439 - fix nice adding, not subtracting, from priority.
440 - fix ->prio and api_is_ready (patch by Mark Hinds).
441 - fixed an assert ("... == prev__cctx->idle_te")
442 that could errronously trigger.
443 - fix various large and small memleaks.
444 - use a (hopefully) more stable cancel implementation
445 that immediately frees the coroutine data.
446 - cede/cede_notself return a status now.
447 - added Coro::guard function.
448 - added a global coroutine pool for jobs (on my machine,
449 I can create and execute 48k simple coros/s with async,
450 and 128k coros with async_pool).
451 - Coro::AIO now uses the coroutine priority as io priority.
452
453 3.2 Fri Dec 22 05:07:09 CET 2006
454 - improve portability to slightly older perls.
455 - use cleaner coroutine destruction.
456 - simplify configuration for users.
457 - optionally (unrecommended) prefer perl functions over
458 their coro replacements.
459
460 3.11 Tue Dec 5 13:11:24 CET 2006
461 - fixed some bogus assert's, but as perl.h disables assert even
462 without NDEBUG (thank you very much), not too many people should
463 notice (that did include myself). Andreas König noticed, though :)
464 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
465 it. Also noticed by Andreas König :)
466 - save/restore tainted status.
467 - verified to pass the testsuite on my 5.9.5.
468
469 3.1 Mon Dec 4 23:03:40 CET 2006
470 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
471 - incompatible change: transfer flags are now per-state.
472 - give a better error message on deadlock.
473 - document Coro::nready.
474 - enhanced testsuite.
475
476 3.01 Sun Dec 3 23:47:42 CET 2006
477 - forgot to include Coro::Timer.
478
479 3.0 Sun Dec 3 22:57:25 CET 2006
480 - the "FINALLY COMPLETELY STABLE" release (coming soon:
481 the "FAMOUS LAST WORDS" release).
482 - implement a new stack sharing algorithm, which uses a stack
483 pool (size currently hardcoded).
484 - make stack sharing mandatory (it no longer uses a heuristic).
485 - eval/die no longer cause weird problems under heavy use.
486 - Coro::Event could cause livelocks if it was used but
487 no Coro::Event watchers were used.
488 - Coro::Event now uses asynccheck as prepare does not
489 check for changed watchers.
490 - Coro::Event allows multiple waiting coros and will wake up one
491 per event.
492 - Coro::Event should be cleaner and more efficient now.
493 - new utility function Coro::unblock_sub.
494 - document the sad fact that Event is no longer reentrant.
495 - putting a coroutine into the ready queue twice could under
496 some circumstances lead to stack corruption.
497 - minor incompatibility: subclassing Coro::State is supported
498 directly now without going through a _coro_state member.
499 - state/coro switching is much faster now.
500 - very minor optimisations and code/documentation cleanup.
501 - avoid problems due to compiler inlining.
502 - removed timers from Coro::Timer -> use AnyEvent instead.
503 - replaced Coro::idle coroutine by (cleaner) idle handler.
504 - updated to newest libcoro.
505 - implement enhanced support for valgrind.
506 - implement is_ready and return value for ready.
507 - removed Coro::Cont, it was a misnomer (it's generators),
508 and it was somewhat annoying to get it right. Will come back
509 if somebody asks for it :)
510 - many ->wait methods and Coro::Event could return
511 spuriously without the event having happened.
512
513 2.5 Tue Nov 7 12:22:33 CET 2006
514 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
515 use AnyEvent, moved them to Coro/.
516 - added Coro::LWP which contains all the uglyness required to
517 make LWP non-blocking.
518 - should work with perl 5.9.x now (Andreas König made me do it).
519 - fixed another bug in Coro::Select when the timeout was undef.
520 - reuse PL_start_env for all coros, saves some memory per coroutine.
521 - manage PL_top_env differently, hopefully to avoid panic: top_env.
522 - timeout argument was not properly used in Coro::Socket.
523 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
524 - emulate undocumented(!) functionality of IO::Socket required
525 by LWP(!!).
526 - updated eg/lwp to work with newer lwp's.
527 - remove "FATAL: uncaught exception" prefix. Coroutines that die
528 kill the whole process, just as exceptions in the main "coroutine"
529 did already.
530
531 2.1 Wed Nov 1 23:01:13 CET 2006
532 - fix a long-standing bug in Coro::Select where select with
533 zero timeout would instead change the current default filehandle.
534 - use a simpler and hopefully more robust way to clone padlists
535 (uses less memory and a perl function instead of our own).
536 - coro can now create a stack guard area on many architectures.
537 - Coro::AIO properly reexports additional functions from IO::AIO.
538 - updated libcoro with a workaround for OS X,
539 pach and testing by Michael Schwern.
540
541 2.0 Tue Oct 24 05:47:17 CEST 2006
542 - support additional aio requests in Coro::AIO.
543
544 1.9 (never properly released due to a glitch)
545
546 1.8 Thu Feb 2 00:59:06 CET 2006
547 - applied suggested patch by SAMV to avoid problems during stupid
548 mark & sweep gc run.
549 - applied patch by Scott Walters for 5.9.3 compatibility.
550
551 1.7 Tue Dec 27 01:41:58 CET 2005
552 - added Coro::AIO, a thin wrapper around IO::AIO.
553 - improved Makefile.PL explanations.
554
555 1.6 Mon Dec 12 21:30:05 CET 2005
556 - additionally save PM_curpm might fix as-of-yet
557 unseen problems with regex matches being attributed to the
558 wrong package.
559 - add t/10_bugs.t, which currently checks against imho broken
560 perls that use 0.26MB of stack space per Perl_magic_get
561 invocation (newer linuxes) as opposed to the 0.0002MB perl
562 normally uses.
563 - make stacksize configurable for the ultimate debian experience.
564
565 1.51 Mon Dec 12 18:48:36 CET 2005
566 - remove debugging warn() accidentally left in Coro::Select.
567
568 1.5 Tue Nov 29 13:32:44 CET 2005
569 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
570 - fix Coro::Select 'select'.
571 - strict'ify some modules.
572 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
573 - use XSLoader in selected modules.
574 - remove some 5.6 compatibility cruft.
575
576 1.4 Tue Sep 6 00:11:05 CEST 2005
577 - libcoro did not take into account the trampoline on amd64,
578 when 'l'inux method was used.
579
580 1.31 Tue Aug 30 23:31:33 CEST 2005
581 - some portability fixes/workarounds.
582
583 1.3 Sat Aug 20 03:08:56 CEST 2005
584 - no code changes. module seems to work fine.
585 - improve Coro::State docs, remove reference to nonexisting
586 Coro::State::flush method.
587 - no longer autodetect windows, present it as an option instead.
588
589 1.2 Mon May 16 02:00:55 CEST 2005
590 - included libcoro.c earlier in State.xs, might improve portability.
591 - use a faster and possibly more stable Coro::Cont implementation.
592 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
593 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
594 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
595
596 1.11 Thu Mar 3 18:00:52 CET 2005
597 - change of contact address.
598
599 1.1 Tue Feb 22 20:51:16 CET 2005
600 - support [l]inux method on amd64.
601 - allow some context switches while compiling/eval'ing.
602
603 1.0 Tue Aug 10 03:47:27 CEST 2004
604 - Event 0.89 fixes the bug again.
605 - fix the remaining(??) memleak. I didn't free the cache-AV for
606 padlists, so one AV per created anon-sub leaked.
607
608 0.97 Fri May 14 15:23:32 CEST 2004
609 - remove stupid gcc extension that was an experiment only anyway.
610
611 0.96 Thu May 13 18:09:29 CEST 2004
612 - tell people if they have yet another broken Event variant.
613 - made Coro to work _better_ with threads (it seems that, as long
614 as you use Coros in one thread only everything will work fine.
615 Looking at the perl source the problem might be missing locking
616 between Coro and some perl internal routines. That's not easily
617 fixable).
618 - (hopefully) got rid of the static xnext inside transfer.
619
620 0.95 Sun Feb 15 03:20:28 CET 2004
621 - removed Coro::State::flush, implemented a new and much
622 faster caching mechanism by attaching magic to the CV.
623 - a memleak with (real) closures remains.
624 - dramatically reduced size of stacks to 65536 (linux/x86), after
625 observing that even myhttpd never had more than a kilobyte
626 of stack in use (stack is still only physically allocated in
627 pagesize increments for systems suporting mmap).
628 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
629 - fix for threaded perl.
630
631 0.9 Sun Nov 23 04:32:39 CET 2003
632 - fix a few very hard to track down but (of course) stupid
633 memory leaks.
634 - stringify version number to avoid locale problems :(.
635 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
636 hopefully ignoring it "just works".
637
638 0.8 Wed Nov 5 19:38:40 CET 2003
639 - port to perl5.005_03, but only 5.8.x+ is supported!
640 - honor LocalAddr even without LocalPort in Coro::Socket.
641
642 0.7 Tue May 27 03:12:38 CEST 2003
643 - the version jump indicates some level of testing, not gobs
644 of new features.
645 - uh... I found the prompt function in ExtUtils::MakeMaker.
646 Highly correct stuff, that is...
647 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
648 the fix is not well understood by the author, so beware :(.
649
650 0.652 Thu May 8 02:54:46 CEST 2003
651 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
652 cause version 4 doesn't have ucontext.
653 - Benjamin Reed reported that setjmp works fine on darwin,
654 so preselect it.
655
656 0.651 Sat Mar 29 15:00:23 CET 2003
657 - fix a bug in Coro::Handle where some bytes could get lost
658 on reading, reported by jason@nichego.net.
659
660 0.65 Sun Mar 23 00:08:26 CET 2003
661 - added a README.linux-glibc.
662 - new module Coro::Select.
663 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
664 (I never claimed that I know what I am doing ;).
665 - set default to setjmp/longjmp on non-x86-linux.
666
667 0.6 Thu Nov 21 11:09:06 CET 2002
668 - made lazy context switching the default.
669 - fixed the bug where SemaphoreSet::guard
670 was the same as timed_guard and timed_guard was missing.
671 - fix a memleak in Coro::Timer.
672 - reclassified context sharing as not experimental.
673
674 0.534 Sun Apr 14 03:05:12 CEST 2002
675 - fixed backspace => tab in header parsing (myhttpd).
676 - added eg/readline, for elmex.
677 - Coro::Event now calls ->start on first call instead of
678 ->again, which created wrong timeouts for the first ->next.
679 - fixed a bug where Coro::Socket returned a closed socket instead
680 of nothing on ECONNREFUSED and similar error conditions.
681
682 0.533 Mon Feb 18 18:49:40 CET 2002
683 - adapt to finally working Event-0.85.
684 - fixed buggy event initialization (trapped under windows, but how
685 could it possibly have worked anywhere?)
686
687 0.532 Wed Jan 16 02:45:32 CET 2002
688 - removed the reference to pp_entersub, might work on
689 (some) win32 perls now (testsuite works under cygwin,
690 without Event, which freezes).
691
692 0.531 Mon Dec 10 22:18:44 CET 2001
693 - Fixed a bug in SemaphoreSet::timed_down.
694
695 0.53 Tue Nov 27 21:11:13 CET 2001
696 - some tests for Event.
697 - slightly different internal architecture to get refcnt into the object
698 and not the reference. might fix a few bugs, certainly did introduce
699 new ones.
700 - Coro::Timer (independent of Event).
701 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
702
703 0.52 Tue Nov 6 21:36:18 CET 2001
704 - ported to cygwin (trivial).
705 patch by Gerrit P. Haase <gp@familiehaase.de>.
706 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
707
708 0.51 Thu Nov 1 20:39:01 CET 2001
709 - terminate/cancel now work properly (otherwise termination
710 could cause a "next coroutine is not and contains not..."
711 error.
712 - added Coro::Socket::shutdown.
713 - Coro::Event::loop no is the same as Event::loop.
714 - implemented terminate with args + join.
715
716 0.5 Fri Sep 28 16:15:35 CEST 2001
717 - fixed "print" on a Coro::Handle. The print method worked.
718 - small tweaks (seem to reduce memory consumption a lot)
719 in various modules.
720 - splendid use of "no warnings" scattered throughout.
721 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
722
723 0.49 Sun Sep 16 02:42:45 CEST 2001
724 - changed some method calls to function calls for speed
725 inside Coro::Handle.
726 - make Coro::Handle use an array instead of a hash for
727 speed reasons.
728 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
729 sjlj and ucontext should now work.
730 - IRIX-specific port for libcoro.
731 - swapped order of accept results to match IO::Socket.
732 - changed getsock/peername to sock/peername to match IO::Socket.
733 - fixed a bug that caused segfault when returning to main task
734 under some circumstances.
735 - other bugfixes.
736
737 0.45 Sun Sep 2 02:54:01 CEST 2001
738 - new method Coro::Handle::timeout.
739 - corrected speling of set...name to get...name in Coro::Socket.
740 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
741 - Coro::Handle now supports fileno and FILENO.
742 - added eg/myhttpd, a web server.
743 - Coro::Socket now accepts numeric ports as well as "port(number)"
744 syntax.
745 - moved some scheduling primitives into xs code. more to come.
746 - new simple priority system.
747 - implemented Coro::Event using XS, almost four times faster!
748 - small memory corruption problem fixed (boy that was difficult).
749
750 0.13 Wed Aug 8 16:53:07 CEST 2001
751 - new method Coro::Handle::readline.
752 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
753 (see eg/lwp on how to make LWP non-blocking).
754 - renamed Coro::Socket::new_inet to new.
755 - added Coro::Util, some utility functions.
756 - die/eval now works better (still get annoying (but true ;)
757 "Callback called exit" messages).
758 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
759 - much better Coro::Socket support.
760
761 0.12 Fri Jul 27 04:19:01 CEST 2001
762 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
763 - very experimental stack sharing algorithm. not 100% safe but
764 should work well in practise ;)
765 - added Coro::RWLock.
766 - Coro::Specific now works.
767
768 0.11 Tue Jul 24 22:49:21 CEST 2001
769 - added specialized hack for newer and older linux versions (fast).
770 - renamed Coro::Event::IO to Coro::Handle.
771 - new module Coro::Socket.
772
773 0.1 Tue Jul 24 01:47:53 CEST 2001
774 - release candidate 3. A rather new internal structure :(
775 - the great renaming: Damian Conway gave me a suitable replacement
776 for yield.
777 - added Coro/libcoro, a portable coroutine implementation for C
778 (not even perl-dependent!!!), which can be used by Coro::State.
779 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
780 perl-5.6.
781 - new function Coro::Event::idle.
782 - the idle process is now overriden
783 by default in Coro::Event.
784 - Coro::Channel now enforces the size.
785 - canceling events now works in all cases.
786 - Coro state now include $_ and $@.
787 - yet another bug workaround that I do not really understand :(
788 - new module Coro::Event::IO (very undocumented) to do
789 non-blocking i/o.
790 - performance tuning.
791
792 0.08 Thu Jul 19 06:13:25 CEST 2001
793 - release candidate 2 (new functionality).
794 - add Coro::State::flush function.
795 - transfer now supports a flags value (mostly for speed ATM).
796 - might compile and work in the presence of threads now.
797 - continuations have a saner syntax.
798 - no more memleaks.
799
800 0.07 Tue Jul 17 17:40:18 CEST 2001
801 - release candidate 1 ;)
802 - slightly nicer code.
803 - fixed a scheduling bug in Coro::Event.
804
805 0.06 Tue Jul 17 04:23:24 CEST 2001
806 - ok, I found the showstopper - the same sub must not be
807 re-entered in two different coroutines, otherwise => crash. I
808 see no easy solution to this problem, except by walking the call
809 chaing and saving/restoring the cv's, which is what I do now.
810 - memory leaks still latent, especially at thread termination.
811 - Coro::Event now works (haha).
812
813 0.05 Sun Jul 15 17:32:20 CEST 2001
814 - fixed a few issues in Coro::Event.
815 - I forgot to include Coro::Event in 0.04 :(:(:(
816
817 0.04 Sun Jul 15 05:24:59 CEST 2001
818 - @_ is now properly localized.
819 - Coro::State is now easier subclassable.
820 - Coro::Cont now coroutine-aware.
821 - Coro::Specific is a low-overhead module to create
822 coroutine-specific vars.
823 - Coro::Event provides a simple interface to Event.
824
825 0.03 Fri Jul 13 14:51:52 CEST 2001
826 - transfer() now implemented in XS (beware).
827 - new module Coro::Cont for really faked continuations.
828 - big internal architecture changes: Coro::State is now
829 really low-level and can thus be used to implement other
830 interesting things, While "Coro::" implements a process-like
831 model. Still crude and subject to change.
832 - $_ and $@ are no longer being localized.
833
834 0.02 Tue Jul 10 01:38:17 CEST 2001
835 - implemented "async" attribute.
836 - $_ and $@ are now localized.
837 - added Coro::Channel.
838 - more testcases, still no docs.
839
840 0.01 Tue Jul 3 02:18:41 CEST 2001
841 - original version; copied from Convert::Scalar.
842