ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
Revision: 1.458
Committed: Sat Nov 22 05:13:34 2008 UTC (15 years, 5 months ago) by root
Branch: MAIN
Changes since 1.457: +4 -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 - disassociate c contexts from coro objects - this is agruably more
12 correct, but mostly allows sharing of cctxs between coro and state
13 objects, for added memory savings and speed increases.
14 - bumped $Coro::POOL_RSS up to 32kb by default.
15 - no longer set the optype to OP_CUSTOM, as B::* understandably
16 doesn't like this very much (and we *are* a type of entersub).
17 - implement state cloning, just to prove that call/cc can be done.
18 - better 5.8.6 compatibility.
19
20 5.0 Thu Nov 20 10:35:05 CET 2008
21 - NEW ARCHITECTURE: use the latest 4.x version if you experience
22 stability issues.
23 - bump API version to 7 - all dependents must be recompiled.
24 - removed timed_* functions - they were not being used anyways
25 and should be replaced by a more generic mechanism -
26 and were annoying to support anyways :)
27 - removed SemaphoreSet's waiter method - use sem method instead.
28 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
29 - async_pool did free a scalar value twice
30 ("Attempt to unreference...").
31 - fix a longstanding bug where calling terminate on a coro that
32 was waiting for a semaphore that was just becoming available
33 would cause a deadlock (semaphore would get into a state where
34 it was available but waiters were still blocked).
35 - calling throw on a coroutine that is waiting for a semaphore will
36 no longer make it acquire the semaphore (and thus leak a count).
37 - perl's process emulation is now not even theoretically supported
38 anymore.
39 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
40 conversion of callback-style to blocking-style.
41 - new methods $coro->schedule_to and ->cede_to, to specifically
42 schedule or cede to a specific coroutine.
43 - new function Coro::Semaphore::wait.
44 - use named constants in Coro::Channel (Richard Hundt).
45 - directly patch the entersub opcode calling SLF functions (cede,
46 transfer and so on). this does speed up context switching, but
47 more importanly, it frees us from the hardcoded behaviour of
48 entersub, so we might actually be able to return something from
49 those functions and atcually create new ones.
50 - take advantage of __builtin_frame_address on gcc.
51 - expose THX in coroapi (not sure whether this was a wise decision,
52 as "threaded" perls are running at half speed anyways).
53 - implement execute_slf (schedule-like-function) interface that makes
54 it possible to implement schedule-like-functions in XS.
55 - use new SLF interface to massively speed up Coro::EV by roughly a
56 factor of two.
57 - used new SLF interface to massively speed up Coro::Semaphore by a
58 factor of three.
59 - used new SLF interface to speed up Coro::AIO by roughly a factor of
60 four and reduce its memory usage considerably.
61 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
62 for a nice speedup and vastly more correct behaviour. Also implement
63 a new method "sem" to get at the underlying semaphore object.
64 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
65 (in comparison) 20-40% speedup.
66 - used new SLF interface to reimplement Coro::Signal gaining
67 some unknown (because I was too lazy), but certain, speedup, and also
68 making signals reliable for the first time.
69 - used new SLF interface and other optimisations to speed up async_pool
70 by a factor of two. It also doesn't rely on perl's exception mechanism
71 to exit anymore. The overhead for terminating an async_pool, coro over
72 a normal async is now very small.
73 - sped up coroutine creation/destruction by 40%.
74 - forgot to include Coro/libcoro/README in the dist for all these years.
75 - work around a freebsd pthreads bug (manual testcancel is required as
76 pthread_cond_wait isn't a cancellation point on freebsd).
77 - use new rouse functions to speed up and simplify Coro::BDB.
78 - make "prefer perl native functions" work with threaded perls.
79 - condense Coro::Debug ps output, hint at v and w flags.
80 - (libcoro) lots of minor cleanups and portability improvements.
81
82 4.914 Wed Nov 19 12:54:18 CET 2008
83 - fix a disastrous bug in the readline optimisation
84 introduced in 4.801.
85
86 4.913 Sat Nov 15 07:58:28 CET 2008
87 - async_pool did free a scalar value twice
88 ("Attempt to unreference...").
89
90 4.912 Thu Nov 13 18:31:23 CET 2008
91 - minor cleanups.
92 - use much larger stacks on linux and perl < 5.8.8.
93 - Coro::Debug::new_unix_server did not unlink the socket
94 when destroyed.
95
96 4.911 Tue Nov 11 04:26:17 CET 2008
97 - "port" to threaded perls.
98
99 4.91 Mon Nov 10 05:36:38 CET 2008
100 - the ->throw exception object no longer leaks.
101 - creating a new cctx leaked a scopestack entry (memleak).
102 - new coroutines didn't get created with a zero flags field
103 (unknown impact).
104 - calling ->throw on a not-yet-started coroutine should now work
105 instead of being ignored.
106 - ->throw is now supported on Coro::State objects.
107 - clean up cctx creation code a bit.
108 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
109
110 4.9 Sat Nov 8 17:45:27 CET 2008
111 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
112 - (libcoro) no longer leak threads in the experimental pthread backend,
113 also speed it up considerably.
114 - (libcoro) do not rely on makecontext passing void *'s unscathed.
115 - fix compiletime dependencies on libcoro in the Makefile.
116 - cctx_count wasn't always updated properly.
117 - Coro::State::cctx_stacksize wasn't applied correctly.
118 - new function Coro::State::cctx_max_idle.
119 - the default max number of idle C contexts is now 4.
120 - (libcoro) try harder to get _setjmp/_longjmp.
121 - (libcoro) cleanup and extend the libcoro API to officially
122 allow the creation of empty source contexts.
123 - very experimental workaround for the totally broken netbsd platform.
124 - tried to hack around openbsd bugs.
125
126 4.804 Wed Nov 5 16:36:00 CET 2008
127 - Coro::Debug::new_unix_server would not create a non-blocking listening
128 socket, sometimes causing freezes.
129 - (libcoro) fix misaligned stack points for setjmp and assembly
130 methods, which can cause crashes on x86/x86_64 with a sufficiently
131 aggressive compiler.
132 - new function: Coro::Debug::new_tcp_server.
133 - move ->throw into the Coro class because it only works on coro objects.
134
135 4.803 Mon Nov 3 17:16:12 CET 2008
136 - (libcoro) use a global asm statement to become independent of gcc
137 otpimisations for CORO_ASM (thanks to pippijn for the idea).
138 - try to workaround yet another broken bsd, this time dragonfly.
139
140 4.802 Thu Oct 30 10:56:12 CET 2008
141 - support -fno-omit-frame-pointer on x86 with the assembly method.
142 - tune 01_unblock.t tests a bit.
143
144 4.801 Wed Oct 22 18:33:37 CEST 2008
145 - improve readline speed for very long "lines".
146 - backport to 5.8.8.
147
148 4.8 Thu Oct 2 13:34:40 CEST 2008
149 - new function Coro::AIO::aio_wait.
150 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
151 - greatly speed up and reduce memory usage of Coro::AIO requests.
152 - implement some other ยต-optimisations.
153
154 4.749 Mon Sep 29 14:40:12 CEST 2008
155 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
156
157 4.748 Sat Sep 27 14:03:19 CEST 2008
158 - implement, but do not document, PerlIO::cede(granularity).
159 - Coro::Storable forgot to wrap Storable::pstore.
160 - work around the multitude of leaks and memory corruption
161 bugs in PerlIO::via by using our own C-level perliol. As a side
162 effect, Coro::Storable is now much, much, much faster.
163
164 4.747 Tue Sep 23 01:59:41 CEST 2008
165 - fix a per-cv memleak (one empty array was leaked per
166 code reference).
167 - avoid a crash in coro->call|rss when the coroutine was already
168 destroyed (most noticably when using Coro::Debug::ps :)
169 - also protect *Storable::FILE.
170 - push up default storable granularity to 20ms.
171
172 4.746 Sun Sep 21 03:22:20 CEST 2008
173 - be more insistent on locking Storable against reentrancy
174 in Coro::Storable.
175 - move swap_def?v and throw to Coro::State, as documented.
176
177 4.745 Thu Jul 24 00:14:38 CEST 2008
178 - remove debugging code related to MgPV_nolen_const, also try to
179 make it compile with perl 5.8.6 (yes, apple apparently loves
180 outdated software). Reported by John S.
181
182 4.744 Tue Jul 8 22:06:35 CEST 2008
183 - correctly provide default DIE/WARN handlers as documented.
184 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
185 implementations inside perl work fine for us.
186
187 4.743 Mon Jun 16 00:21:57 CEST 2008
188 - when using Coro::EV without running EV::loop it could
189 result in busy-waiting for events, this has been fixed.
190 - reduce codesize and improve performance by using EV_DEFAULT_UC.
191
192 4.742 Sat May 31 14:10:21 CEST 2008
193 - implement a workaround for (some) perl <5.8.8 versions.
194 - require EV 3.3+.
195
196 4.741 Fri May 30 23:33:09 CEST 2008
197 - tell netbsd how utterly broken their imitation of an OS is
198 and refuse to build by default if pthreads are in use.
199 - switch to "s" method on all bsd's by default, as their ucontext
200 stuff seems just too broken.
201 - fix a bug in Coro::Select.
202
203 4.74 Thu May 29 20:05:31 CEST 2008
204 - do not test Coro::LWP for lack of dependencies.
205 - do not test Coro::Select for lack of working perls.
206
207 4.73 Thu May 29 2008
208 - fix a bug in Coro::EV which would cause it to block despite
209 there being runnable coroutines.
210 - sprinkle "no warnings" freely over everything, also suppress
211 warnings for some other modules.
212 - fix typo in WSAEWOULDBLOCK.
213
214 4.72 Sun May 25 05:14:36 CEST 2008
215 - tweak META.yaml a bit, unfortunately, there is no documented way
216 to have optional dependencies with CPAN. doh :(
217 - avoid running some tests on windows because they would fail due to
218 perl bug (broken fork, broken pipes...).
219 - work around perl on windows bugs where perl returns undocumented
220 error codes for sysread, syswrite etc. by taking advantage
221 of AnyEvent's workaround for that problem.
222 - use AnyEvent::Util::fh_nonblocking in Coro::Handle to work around
223 a common perl implementation bug on windows.
224 - use unix domain sockets in testsuite to work around
225 common perl implementation bugs on widows (they are emulated by
226 tcp sockets on windows. ugh.)
227
228 4.71 Sat May 24 20:01:27 CEST 2008
229 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
230 - take advantage of async name resolution of AnyEvent::Util.
231 - work around brutal inet_aton override in Coro::LWP.
232 - take advantage of the readyhook in Coro::EV, for smoother
233 scheduling.
234
235 4.7 Sun May 11 00:32:19 CEST 2008
236 - completely reworked the Coro manpage.
237 - added Coro::AnyEvent, generic event loop integration.
238 - implement cancel, ready and kill commands in Coro::Debug.
239 - document find_coro in Coro::Debug.
240 - incompatible change: rename has_stack to has_cctx.
241 - Coro::AIO and Coro::BDB no longer force event model detection,
242 use AnyEvent::AIO and AnyEvent::BDB.
243
244 4.6 Sat Apr 26 10:05:14 CEST 2008
245 - INCOMPATIBLE CHANGE: sub/code attributes are no longer supported
246 by the Coro module. It was a mistake to have it in the first place.
247 - (experimental) support for activestate perl 5.10 (method "w").
248 - (experimental) support for strawberry perl (method "a").
249 - coro_sigelem_set did not return a value although it had to,
250 actual impact unknown.
251
252 4.51 Mon Apr 14 13:28:27 CEST 2008
253 - make it compile again on 5.8.
254
255 4.50 Thu Apr 10 09:43:17 CEST 2008
256 - I did it twice! (see 4.49).
257
258 4.49 Sun Apr 6 21:23:31 CEST 2008
259 - grr, instead of compiling the recent changes on 5.10 only they
260 were compiled on 5.8 only.
261
262 4.48 Sun Apr 6 20:36:46 CEST 2008
263 - allow coroutine switches during eval's under 5.10.x, as it seems
264 the parser is a per-interpreter option now, so this is safe
265 (this might fix the odd crashing bug).
266 - drop support for 5.9.x versions: they are dead, jim.
267
268 4.47 Sun Apr 6 00:37:52 CEST 2008
269 - force cctx allocation on API calls: we know we need to force one
270 and gcc actually manages to confuse our heuristic nowadays,
271 leading to crashes and worse.
272 - document force_cctx.
273
274 4.46 Fri Apr 4 22:05:43 CEST 2008
275 - upgrade libcoro, resulting in pthread-backend (which was only created
276 to fulfill the rules of the programming languages shootout).
277
278 4.45 Thu Mar 13 11:55:36 CET 2008
279 - fix a file leaking bug in eg/mhyttpd that would allow
280 downloading of any file (reported by oesi).
281 - fix deadlock bug in Coro::Channel (reported by Richard Hundt)
282 (also add testcase).
283 - support Broadcast option in Coro::Socket (patch by Richard Hundt,
284 apparently having loads of fun with that).
285 - implement $state->swap_defsv, ->swap_defav and document ->throw.
286
287 4.4 Wed Feb 13 16:44:29 CET 2008
288 - only assume .cfi pseudo isns exist on GNU/Linux.
289 - add get_timed method to Coro::Channels.
290 - fixed Coro::Debug manpage.
291 - perl 5.11 compatbility improvement based on patch by
292 Andreas Kรถnig.
293
294 4.37 Sun Jan 20 11:25:23 CET 2008
295 - soften the check for an active parser for perl 5.10 (avoiding
296 "Coro::State::transfer called while parsing" in many cases).
297
298 4.36 Sun Jan 13 10:53:56 CET 2008
299 - reset diehook when terminating from an async_pool as to not
300 trigger any __DIE__ handlers.
301
302 4.35 Sun Jan 13 04:14:13 CET 2008
303 - "bt" debug command now displays any exceptions
304 from longmess and also skips the innermost
305 stackframes, giving more useufl output.
306 - allow backtraces in coroutines blocked in Coro::EV,
307 at a <1% speed hit.
308 - handle localising of $SIG{__WARN__} and __DIE__
309 properly (with a proper amount of dirty hacking).
310
311 4.34 Sat Dec 22 17:49:53 CET 2007
312 - upgrade to EV version 2.0 API.
313
314 4.33 Mon Dec 17 08:36:12 CET 2007
315 - make Coro::AIO etc. loadable in the absence of EV.
316
317 4.32 Mon Dec 17 07:46:02 CET 2007
318 - majorly improved Coro::Handle's performance with Coro::EV.
319 - implemented the until now mythical Coro::BDB module.
320 - specialcase EV in Coro::AIO and Coro::BDB for extra speed.
321
322 4.31 Wed Dec 5 12:32:39 CET 2007
323 - remove warn statement form Coro::Util.
324
325 4.3 Tue Dec 4 20:33:14 CET 2007
326 - calls to the idle function could cause stack corruption
327 when the stack changed.
328 - do no longer rely on the presence of EV::DNS if EV is used
329 (because it is gone), but instead take avdantage of EV::ADNS
330 if available.
331 - add ($) prototypes to all functions in Coro::Storable.
332 - use a conventional (safer) idle callback in Coro::EV.
333 - do accept quickly in Coro::Debug to avoid endless loops.
334
335 4.22 Fri Nov 30 16:04:04 CET 2007
336 - really use optimised versions for Event and EV in Coro::Util
337 and Coro::Handle.
338
339 4.21 Sun Nov 25 10:48:59 CET 2007
340 - fix a spurious memory read.
341 - Coro::EV no longer keeps the eventloop "alive".
342
343 4.2 Fri Nov 9 20:47:05 CET 2007
344 - enable/disable tracing from a new coroutine, not a pooled one.
345 - fix a memleak in Coro::Event.
346 - removed killall call from fork_eval.
347 - made sure store_fd is already loaded so that fork_eval
348 does not have to parse autoload in each subprocess.
349 - only use assembly method if -O switch is in $Config{optimize}.
350 - add (optional) Coro::EV module, so far the best event loop module
351 directly supported by Coro.
352 - if the event model is EV, use EV::DNS to resolve
353 stuff in Coro::Util.
354 - don't get confused by multiple event notifications in Coro::Handle.
355 - initial support for EV (libevent interface).
356 - require Event and EV using configure_requires, to force their existance.
357
358 4.13 Wed Oct 24 07:26:45 CEST 2007
359 - add Coro::Storable::blocking_thaw.
360 - use a vastly more complicated technique to localise
361 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
362 - use a coroutine for the idle callback Coro::Event,
363 instead of running Event in the current coroutine context.
364 This also catches recursive invocations.
365 - actually report fork errors in gethostbyname and inet_aton.
366
367 4.11 Thu Oct 11 02:40:24 CEST 2007
368 - port to threaded perls.
369
370 4.1 Thu Oct 11 02:38:16 CEST 2007
371 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
372 be local to each coro (see Coro::State).
373 - incompatible change: for very deep reasons, cede and cede_notself
374 cannot return anything, so nothing will be returned.
375 - possibly bring back 5.10 compatibility (untested).
376 - work around stupid (and wrong) warning on 5.10 :(.
377 - overlay the saved state over the context stack. This saves
378 a few hundred bytes per coroutine on average and also
379 speeds up context switching a bit.
380 - further tune default stack sizes.
381 - (more) correctly calculate stack usage in coro_rss.
382 - Coro::Storable::blocking_* did not properly lock
383 resulting in races between coroutines.
384 - added Coro::Storable::guard.
385 - stopping to trace a coroutine could destroy the cctx of
386 an unrelated coroutine.
387 - explain the relationship between Perl and C coroutines in
388 more detail in Coro::State.
389 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
390 causing a fork per resolve. This also affected Coro::Socket.
391 - switch to a separate stack in $coro->call/eval to avoid
392 invalidating pointers.
393
394 4.03 Sat Oct 6 21:24:00 CEST 2007
395 - added Coro::throw method.
396 - minor code cleanups.
397
398 4.02 Sat Oct 6 02:36:47 CEST 2007
399 - fix a very minor per-coroutine memleak (a single codereference).
400 - fixed a bug where the currently in-use c context would be freed
401 prematurely (can happen only when programs change the stacksize
402 or use tracing).
403 - tracing can no longer keep a coro alive after it terminated.
404 - do static branch prediction in the common path for gcc. gives
405 about 2-5% speed improvement here.
406
407 4.01 Fri Oct 5 22:10:49 CEST 2007
408 - instead of recreating *a* standard output handle we simply
409 use STDOUT, which is faster and hopefully more robust.
410
411 4.0 Fri Oct 5 12:56:00 CEST 2007
412 - incompatibly changed Coro::Storable::freeze.
413 - major new feature: added Coro::Debug, for interactive coroutine
414 debugging, tracing and much more.
415 - major bug fix: unbelievable, but true: $_, $/ and many other
416 "saved" variables actually weren't being saved. This has been fixed,
417 of course, while increasing performance while losing all the save
418 flags.
419 - save flags are gone, and all the api functions dealing with them.
420 - added Coro::Semaphore::adjust.
421 - added Coro::Util::fork_eval.
422 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
423 - added Coro::killall.
424 - reduce initial stack sizes to allow for "micro-coroutines".
425 - better async_pool resource management, moved parts of async_pool
426 handling to XS (major speed improvement).
427 - actually croak before modifying important data structures.
428 - refuse to transfer while compiling.
429 - possibly support eval EXPR better now.
430 - enable assembly per default on linux+bsd x86+amd64.
431 - all internal members were renamed _something for easier subclassing.
432 - many minor tweaks.
433
434 3.63 Wed May 16 14:10:06 CEST 2007
435 - implement handcoded assembly for x86/amd64 SVR ABI.
436
437 3.62 Fri Apr 27 21:36:06 CEST 2007
438 - upgrade libcoro (which might set unwind info correctly).
439 - change default on linux to setjmp/longjmp.
440
441 3.61 Thu Apr 19 12:36:18 CEST 2007
442 - Coro::Storable caused an endless loop when thawing invalid
443 pst's sometimes.
444 - use a Semaphore in Coro::Storable, as Storable doesn't
445 seem to be reentrant (although it is documented to
446 be threadsafe...).
447 - fix Coro::Signal to bring back the original unreliable
448 but stateful semantics.
449 - fixed a lot of typos in Coro.pm (patch submitted by David
450 Steinbrunner, which applied flawlessly).
451
452 3.6 Sat Apr 14 17:13:31 CEST 2007
453 - added some bugfixes to get eg/myhttpd working again.
454 - added Coro::Storable for often-cede'ing freeze/thaw.
455 - try to do a clean exit when a coroutine calls exit
456 (EXPERIMENTAL).
457 - got rid of indirect call through _coro_init.
458 - updated the partly antique examples in eg/ to
459 work again and be a bit less magic, too.
460 - fixed Coro::Signal semantics to work as documented again.
461
462 3.55 Sun Mar 25 01:20:47 CET 2007
463 - add SAVE_DEFFH to save the default filehandle and enable
464 it by default.
465 - finally move socket-operations from Coro::Socket to Coro::Handle
466 to be able to unblock foreign sockets.
467 - add Coro::State::save_also and guarded_save.
468 - add count accessor to Coro::Semaphore.
469 - add Coro::State::cctx_stacksize.
470 - just for the fun of it, do not rely on implicit context,
471 which can dramatically improve performance, but people
472 using windows-process-emulation perls do not care much about
473 performance.
474
475 3.51 Sun Mar 4 14:18:04 CET 2007
476 - fixed a problem when you weakened references to Coro::State's
477 (patch by Yuval Kogman).
478
479 3.501 Wed Feb 28 12:44:07 CET 2007
480 - rename some global symbols as macosx from hell redefines
481 them without asking.
482
483 3.5 Tue Feb 13 20:22:53 CET 2007
484 - do AnyEvent model detection earlier, avoiding problems
485 caused by first using AnyEvent and later Coro::Event.
486 - implement and document Coro::Event event objects.
487 - fix a potential problem in Coro::Event causing crashes.
488 - initialise PL_comppad when creating a new coroutine,
489 avoids crashes on early coro destruction.
490
491 3.41 Mon Jan 22 19:19:49 CET 2007
492 - readline on Coro::Handle did not support undefined $/,
493 nor did it deliver partial lines on EOF or error.
494 - implement malloc fallback for stack allocation because
495 stupid broken idiotic OSX has a stupid broken
496 idiotic fits-the-whole-os mmap "implementation" and
497 my dick feels longer if Coro is portable even to
498 obsolete platforms.
499
500 3.4 Fri Jan 19 21:52:54 CET 2007
501 - remove t/09_timer.t, as it isn't really testing much
502 but was rather flaky in practise.
503 - async_pool coro would keep arguments and callback alive until
504 it was reused.
505 - cancellation of a coroutine could cause spurious idle calls
506 in cede_notself.
507
508 3.3 Sat Jan 6 03:45:00 CET 2007
509 - implement $coro->on_destroy.
510 - Coro::Event blocking semantics have been changed,
511 documented and - hopefully - improved.
512 - fix nice adding, not subtracting, from priority.
513 - fix ->prio and api_is_ready (patch by Mark Hinds).
514 - fixed an assert ("... == prev__cctx->idle_te")
515 that could errronously trigger.
516 - fix various large and small memleaks.
517 - use a (hopefully) more stable cancel implementation
518 that immediately frees the coroutine data.
519 - cede/cede_notself return a status now.
520 - added Coro::guard function.
521 - added a global coroutine pool for jobs (on my machine,
522 I can create and execute 48k simple coros/s with async,
523 and 128k coros with async_pool).
524 - Coro::AIO now uses the coroutine priority as io priority.
525
526 3.2 Fri Dec 22 05:07:09 CET 2006
527 - improve portability to slightly older perls.
528 - use cleaner coroutine destruction.
529 - simplify configuration for users.
530 - optionally (unrecommended) prefer perl functions over
531 their coro replacements.
532
533 3.11 Tue Dec 5 13:11:24 CET 2006
534 - fixed some bogus assert's, but as perl.h disables assert even
535 without NDEBUG (thank you very much), not too many people should
536 notice (that did include myself). Andreas Kรถnig noticed, though :)
537 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
538 it. Also noticed by Andreas Kรถnig :)
539 - save/restore tainted status.
540 - verified to pass the testsuite on my 5.9.5.
541
542 3.1 Mon Dec 4 23:03:40 CET 2006
543 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
544 - incompatible change: transfer flags are now per-state.
545 - give a better error message on deadlock.
546 - document Coro::nready.
547 - enhanced testsuite.
548
549 3.01 Sun Dec 3 23:47:42 CET 2006
550 - forgot to include Coro::Timer.
551
552 3.0 Sun Dec 3 22:57:25 CET 2006
553 - the "FINALLY COMPLETELY STABLE" release (coming soon:
554 the "FAMOUS LAST WORDS" release).
555 - implement a new stack sharing algorithm, which uses a stack
556 pool (size currently hardcoded).
557 - make stack sharing mandatory (it no longer uses a heuristic).
558 - eval/die no longer cause weird problems under heavy use.
559 - Coro::Event could cause livelocks if it was used but
560 no Coro::Event watchers were used.
561 - Coro::Event now uses asynccheck as prepare does not
562 check for changed watchers.
563 - Coro::Event allows multiple waiting coros and will wake up one
564 per event.
565 - Coro::Event should be cleaner and more efficient now.
566 - new utility function Coro::unblock_sub.
567 - document the sad fact that Event is no longer reentrant.
568 - putting a coroutine into the ready queue twice could under
569 some circumstances lead to stack corruption.
570 - minor incompatibility: subclassing Coro::State is supported
571 directly now without going through a _coro_state member.
572 - state/coro switching is much faster now.
573 - very minor optimisations and code/documentation cleanup.
574 - avoid problems due to compiler inlining.
575 - removed timers from Coro::Timer -> use AnyEvent instead.
576 - replaced Coro::idle coroutine by (cleaner) idle handler.
577 - updated to newest libcoro.
578 - implement enhanced support for valgrind.
579 - implement is_ready and return value for ready.
580 - removed Coro::Cont, it was a misnomer (it's generators),
581 and it was somewhat annoying to get it right. Will come back
582 if somebody asks for it :)
583 - many ->wait methods and Coro::Event could return
584 spuriously without the event having happened.
585
586 2.5 Tue Nov 7 12:22:33 CET 2006
587 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
588 use AnyEvent, moved them to Coro/.
589 - added Coro::LWP which contains all the uglyness required to
590 make LWP non-blocking.
591 - should work with perl 5.9.x now (Andreas Kรถnig made me do it).
592 - fixed another bug in Coro::Select when the timeout was undef.
593 - reuse PL_start_env for all coros, saves some memory per coroutine.
594 - manage PL_top_env differently, hopefully to avoid panic: top_env.
595 - timeout argument was not properly used in Coro::Socket.
596 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
597 - emulate undocumented(!) functionality of IO::Socket required
598 by LWP(!!).
599 - updated eg/lwp to work with newer lwp's.
600 - remove "FATAL: uncaught exception" prefix. Coroutines that die
601 kill the whole process, just as exceptions in the main "coroutine"
602 did already.
603
604 2.1 Wed Nov 1 23:01:13 CET 2006
605 - fix a long-standing bug in Coro::Select where select with
606 zero timeout would instead change the current default filehandle.
607 - use a simpler and hopefully more robust way to clone padlists
608 (uses less memory and a perl function instead of our own).
609 - coro can now create a stack guard area on many architectures.
610 - Coro::AIO properly reexports additional functions from IO::AIO.
611 - updated libcoro with a workaround for OS X,
612 pach and testing by Michael Schwern.
613
614 2.0 Tue Oct 24 05:47:17 CEST 2006
615 - support additional aio requests in Coro::AIO.
616
617 1.9 (never properly released due to a glitch)
618
619 1.8 Thu Feb 2 00:59:06 CET 2006
620 - applied suggested patch by SAMV to avoid problems during stupid
621 mark & sweep gc run.
622 - applied patch by Scott Walters for 5.9.3 compatibility.
623
624 1.7 Tue Dec 27 01:41:58 CET 2005
625 - added Coro::AIO, a thin wrapper around IO::AIO.
626 - improved Makefile.PL explanations.
627
628 1.6 Mon Dec 12 21:30:05 CET 2005
629 - additionally save PM_curpm might fix as-of-yet
630 unseen problems with regex matches being attributed to the
631 wrong package.
632 - add t/10_bugs.t, which currently checks against imho broken
633 perls that use 0.26MB of stack space per Perl_magic_get
634 invocation (newer linuxes) as opposed to the 0.0002MB perl
635 normally uses.
636 - make stacksize configurable for the ultimate debian experience.
637
638 1.51 Mon Dec 12 18:48:36 CET 2005
639 - remove debugging warn() accidentally left in Coro::Select.
640
641 1.5 Tue Nov 29 13:32:44 CET 2005
642 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
643 - fix Coro::Select 'select'.
644 - strict'ify some modules.
645 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
646 - use XSLoader in selected modules.
647 - remove some 5.6 compatibility cruft.
648
649 1.4 Tue Sep 6 00:11:05 CEST 2005
650 - libcoro did not take into account the trampoline on amd64,
651 when 'l'inux method was used.
652
653 1.31 Tue Aug 30 23:31:33 CEST 2005
654 - some portability fixes/workarounds.
655
656 1.3 Sat Aug 20 03:08:56 CEST 2005
657 - no code changes. module seems to work fine.
658 - improve Coro::State docs, remove reference to nonexisting
659 Coro::State::flush method.
660 - no longer autodetect windows, present it as an option instead.
661
662 1.2 Mon May 16 02:00:55 CEST 2005
663 - included libcoro.c earlier in State.xs, might improve portability.
664 - use a faster and possibly more stable Coro::Cont implementation.
665 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
666 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
667 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
668
669 1.11 Thu Mar 3 18:00:52 CET 2005
670 - change of contact address.
671
672 1.1 Tue Feb 22 20:51:16 CET 2005
673 - support [l]inux method on amd64.
674 - allow some context switches while compiling/eval'ing.
675
676 1.0 Tue Aug 10 03:47:27 CEST 2004
677 - Event 0.89 fixes the bug again.
678 - fix the remaining(??) memleak. I didn't free the cache-AV for
679 padlists, so one AV per created anon-sub leaked.
680
681 0.97 Fri May 14 15:23:32 CEST 2004
682 - remove stupid gcc extension that was an experiment only anyway.
683
684 0.96 Thu May 13 18:09:29 CEST 2004
685 - tell people if they have yet another broken Event variant.
686 - made Coro to work _better_ with threads (it seems that, as long
687 as you use Coros in one thread only everything will work fine.
688 Looking at the perl source the problem might be missing locking
689 between Coro and some perl internal routines. That's not easily
690 fixable).
691 - (hopefully) got rid of the static xnext inside transfer.
692
693 0.95 Sun Feb 15 03:20:28 CET 2004
694 - removed Coro::State::flush, implemented a new and much
695 faster caching mechanism by attaching magic to the CV.
696 - a memleak with (real) closures remains.
697 - dramatically reduced size of stacks to 65536 (linux/x86), after
698 observing that even myhttpd never had more than a kilobyte
699 of stack in use (stack is still only physically allocated in
700 pagesize increments for systems suporting mmap).
701 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
702 - fix for threaded perl.
703
704 0.9 Sun Nov 23 04:32:39 CET 2003
705 - fix a few very hard to track down but (of course) stupid
706 memory leaks.
707 - stringify version number to avoid locale problems :(.
708 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
709 hopefully ignoring it "just works".
710
711 0.8 Wed Nov 5 19:38:40 CET 2003
712 - port to perl5.005_03, but only 5.8.x+ is supported!
713 - honor LocalAddr even without LocalPort in Coro::Socket.
714
715 0.7 Tue May 27 03:12:38 CEST 2003
716 - the version jump indicates some level of testing, not gobs
717 of new features.
718 - uh... I found the prompt function in ExtUtils::MakeMaker.
719 Highly correct stuff, that is...
720 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
721 the fix is not well understood by the author, so beware :(.
722
723 0.652 Thu May 8 02:54:46 CEST 2003
724 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
725 cause version 4 doesn't have ucontext.
726 - Benjamin Reed reported that setjmp works fine on darwin,
727 so preselect it.
728
729 0.651 Sat Mar 29 15:00:23 CET 2003
730 - fix a bug in Coro::Handle where some bytes could get lost
731 on reading, reported by jason@nichego.net.
732
733 0.65 Sun Mar 23 00:08:26 CET 2003
734 - added a README.linux-glibc.
735 - new module Coro::Select.
736 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
737 (I never claimed that I know what I am doing ;).
738 - set default to setjmp/longjmp on non-x86-linux.
739
740 0.6 Thu Nov 21 11:09:06 CET 2002
741 - made lazy context switching the default.
742 - fixed the bug where SemaphoreSet::guard
743 was the same as timed_guard and timed_guard was missing.
744 - fix a memleak in Coro::Timer.
745 - reclassified context sharing as not experimental.
746
747 0.534 Sun Apr 14 03:05:12 CEST 2002
748 - fixed backspace => tab in header parsing (myhttpd).
749 - added eg/readline, for elmex.
750 - Coro::Event now calls ->start on first call instead of
751 ->again, which created wrong timeouts for the first ->next.
752 - fixed a bug where Coro::Socket returned a closed socket instead
753 of nothing on ECONNREFUSED and similar error conditions.
754
755 0.533 Mon Feb 18 18:49:40 CET 2002
756 - adapt to finally working Event-0.85.
757 - fixed buggy event initialization (trapped under windows, but how
758 could it possibly have worked anywhere?)
759
760 0.532 Wed Jan 16 02:45:32 CET 2002
761 - removed the reference to pp_entersub, might work on
762 (some) win32 perls now (testsuite works under cygwin,
763 without Event, which freezes).
764
765 0.531 Mon Dec 10 22:18:44 CET 2001
766 - Fixed a bug in SemaphoreSet::timed_down.
767
768 0.53 Tue Nov 27 21:11:13 CET 2001
769 - some tests for Event.
770 - slightly different internal architecture to get refcnt into the object
771 and not the reference. might fix a few bugs, certainly did introduce
772 new ones.
773 - Coro::Timer (independent of Event).
774 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
775
776 0.52 Tue Nov 6 21:36:18 CET 2001
777 - ported to cygwin (trivial).
778 patch by Gerrit P. Haase <gp@familiehaase.de>.
779 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
780
781 0.51 Thu Nov 1 20:39:01 CET 2001
782 - terminate/cancel now work properly (otherwise termination
783 could cause a "next coroutine is not and contains not..."
784 error.
785 - added Coro::Socket::shutdown.
786 - Coro::Event::loop no is the same as Event::loop.
787 - implemented terminate with args + join.
788
789 0.5 Fri Sep 28 16:15:35 CEST 2001
790 - fixed "print" on a Coro::Handle. The print method worked.
791 - small tweaks (seem to reduce memory consumption a lot)
792 in various modules.
793 - splendid use of "no warnings" scattered throughout.
794 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.
795
796 0.49 Sun Sep 16 02:42:45 CEST 2001
797 - changed some method calls to function calls for speed
798 inside Coro::Handle.
799 - make Coro::Handle use an array instead of a hash for
800 speed reasons.
801 - IRIX mystery solved: it's SGI's NT, after all: "standard, huh?".
802 sjlj and ucontext should now work.
803 - IRIX-specific port for libcoro.
804 - swapped order of accept results to match IO::Socket.
805 - changed getsock/peername to sock/peername to match IO::Socket.
806 - fixed a bug that caused segfault when returning to main task
807 under some circumstances.
808 - other bugfixes.
809
810 0.45 Sun Sep 2 02:54:01 CEST 2001
811 - new method Coro::Handle::timeout.
812 - corrected speling of set...name to get...name in Coro::Socket.
813 - Coro::Socket::accept now returns a Coro::Socket, not a Coro::Handle.
814 - Coro::Handle now supports fileno and FILENO.
815 - added eg/myhttpd, a web server.
816 - Coro::Socket now accepts numeric ports as well as "port(number)"
817 syntax.
818 - moved some scheduling primitives into xs code. more to come.
819 - new simple priority system.
820 - implemented Coro::Event using XS, almost four times faster!
821 - small memory corruption problem fixed (boy that was difficult).
822
823 0.13 Wed Aug 8 16:53:07 CEST 2001
824 - new method Coro::Handle::readline.
825 - added eg/lwp and Coro::Handle::autoflush, to make LWP work.
826 (see eg/lwp on how to make LWP non-blocking).
827 - renamed Coro::Socket::new_inet to new.
828 - added Coro::Util, some utility functions.
829 - die/eval now works better (still get annoying (but true ;)
830 "Callback called exit" messages).
831 - Coro::Handle now supports timeout, and Coro::Socket Timeout.
832 - much better Coro::Socket support.
833
834 0.12 Fri Jul 27 04:19:01 CEST 2001
835 - do not use mmap if neither MAP_ANON nor MAP_ANONYMOUS is defined.
836 - very experimental stack sharing algorithm. not 100% safe but
837 should work well in practise ;)
838 - added Coro::RWLock.
839 - Coro::Specific now works.
840
841 0.11 Tue Jul 24 22:49:21 CEST 2001
842 - added specialized hack for newer and older linux versions (fast).
843 - renamed Coro::Event::IO to Coro::Handle.
844 - new module Coro::Socket.
845
846 0.1 Tue Jul 24 01:47:53 CEST 2001
847 - release candidate 3. A rather new internal structure :(
848 - the great renaming: Damian Conway gave me a suitable replacement
849 for yield.
850 - added Coro/libcoro, a portable coroutine implementation for C
851 (not even perl-dependent!!!), which can be used by Coro::State.
852 - renamed SAVE_DEFSV (on xs-level only) to avoid symbol clash in
853 perl-5.6.
854 - new function Coro::Event::idle.
855 - the idle process is now overriden
856 by default in Coro::Event.
857 - Coro::Channel now enforces the size.
858 - canceling events now works in all cases.
859 - Coro state now include $_ and $@.
860 - yet another bug workaround that I do not really understand :(
861 - new module Coro::Event::IO (very undocumented) to do
862 non-blocking i/o.
863 - performance tuning.
864
865 0.08 Thu Jul 19 06:13:25 CEST 2001
866 - release candidate 2 (new functionality).
867 - add Coro::State::flush function.
868 - transfer now supports a flags value (mostly for speed ATM).
869 - might compile and work in the presence of threads now.
870 - continuations have a saner syntax.
871 - no more memleaks.
872
873 0.07 Tue Jul 17 17:40:18 CEST 2001
874 - release candidate 1 ;)
875 - slightly nicer code.
876 - fixed a scheduling bug in Coro::Event.
877
878 0.06 Tue Jul 17 04:23:24 CEST 2001
879 - ok, I found the showstopper - the same sub must not be
880 re-entered in two different coroutines, otherwise => crash. I
881 see no easy solution to this problem, except by walking the call
882 chaing and saving/restoring the cv's, which is what I do now.
883 - memory leaks still latent, especially at thread termination.
884 - Coro::Event now works (haha).
885
886 0.05 Sun Jul 15 17:32:20 CEST 2001
887 - fixed a few issues in Coro::Event.
888 - I forgot to include Coro::Event in 0.04 :(:(:(
889
890 0.04 Sun Jul 15 05:24:59 CEST 2001
891 - @_ is now properly localized.
892 - Coro::State is now easier subclassable.
893 - Coro::Cont now coroutine-aware.
894 - Coro::Specific is a low-overhead module to create
895 coroutine-specific vars.
896 - Coro::Event provides a simple interface to Event.
897
898 0.03 Fri Jul 13 14:51:52 CEST 2001
899 - transfer() now implemented in XS (beware).
900 - new module Coro::Cont for really faked continuations.
901 - big internal architecture changes: Coro::State is now
902 really low-level and can thus be used to implement other
903 interesting things, While "Coro::" implements a process-like
904 model. Still crude and subject to change.
905 - $_ and $@ are no longer being localized.
906
907 0.02 Tue Jul 10 01:38:17 CEST 2001
908 - implemented "async" attribute.
909 - $_ and $@ are now localized.
910 - added Coro::Channel.
911 - more testcases, still no docs.
912
913 0.01 Tue Jul 3 02:18:41 CEST 2001
914 - original version; copied from Convert::Scalar.
915