ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/Coro/Changes
(Generate patch)

Comparing Coro/Changes (file contents):
Revision 1.426 by root, Mon Nov 17 07:19:31 2008 UTC vs.
Revision 1.585 by root, Thu Aug 4 19:37:58 2011 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::coroaio (perl leaks like hell). 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: maybe implement a default message channel, very much like Erlang's 4TODO: channel->maxsize(newsize)?
5 actor model (which is cool in a lot of important aspects (failures!), 5TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
6 but very lacking in others (higher level ipc)).
7 6
8TODO: testcancel 76.05 Thu Aug 4 21:36:36 CEST 2011
9TODO: aio 8 - blush, condvar values would not be propagated from send to recv anymore
10TODO: Coro::Semaphore 9 (reported by Chip Salzenberg).
11TODO: trow disturbs cleanup path inside pp_slf? 10 - use exponential increase for the readline buffer length in
12TODO: throw special-case in transfer? 11 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
135.0 12
136.04 Wed Aug 3 17:15:45 CEST 2011
14 - use even more efficient and more compatible condvars for
15 compatibility to AnyEvent 6.x :)
16 - more inconsequential ecb.h updates.
17
186.03 Wed Aug 3 11:41:30 CEST 2011
19 - change how Coro patches AnyEvent condvars for compatibility to
20 AnyEvent 6.x.
21 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
22 hasn't been included.
23
246.02 Wed Jul 13 04:35:19 CEST 2011
25 - "improve portability to Gentoo" - gentoo manages to put perl variables
26 in memory areas that are farther than 2gb apart, which the jit couldn't
27 handle and barfed. now it's just a bit slower on gentoo and similar
28 systems.
29
306.01 Sun Jul 3 12:31:14 CEST 2011
31 - workarounds are good, but the test for whether pthreads are used
32 was not good. thisone should be better.
33 - check differently whether gcc generates cfi instructions itself.
34
356.0 Wed Jun 29 19:43:35 CEST 2011
36 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
37 destroyed and cleaned up automatically (e.g. async { schedule }).
38 - implement a JIT compiler for part of the thread switch code,
39 which gives a 50% speed improvement on threaded perls, and
40 about 4% on non-thraeded perls (so threaded perls now finally
41 reach about half the speed of non-threaded perls).
42 - slightly modernise Coro::Intro, add section about rouse functions.
43 - avoid DEFSV and ERRSV, giving another 10% improvement
44 in thread switching.
45 - Coro::State->is_destroyed is now called is_zombie.
46 - implement a Coro->safe_cancel method that might fail, but
47 cancels in a "safer" way if it succeeds.
48 - add preliminary support for DEBUGGING perls.
49 - get rid of two hash-accesses when initialising a new Coro - this
50 speeds up coro creation by almost a factor of two.
51 - croak when a coro that is being cancelled tries to block
52 (e.g. while executing a guard block), instead of crashing or
53 deadlocking.
54 - use a more robust and also faster method to identify Coro::State
55 objects - speeds up everything a bit.
56 - implement Coro->cancel in XS for a 20% speed improvement, and to
57 be able to implement mutual cancellation.
58 - speed up context switches by a percent or two by more efficiently
59 allocating context stack entries.
60 - implement Coro->join and Coro->on_destroy in XS for a speedup and
61 a reduction in memory use.
62 - cancelling a coro while it itself is cancelling another coro is
63 now supported and working, instead of triggering an assertion.
64 - be a bit more crash-resistant when calling (buggy) on_destroy
65 callbacks (best effort).
66 - move on_destroy into the slf_frame, to allow extension slf
67 functions to have destructors.
68 - get rid if coro refcounting - simply crash in other interpreter
69 threads by nulling the pointers on clone.
70 - simplify warn/die hook handling when loading Coro - the convoluted
71 logic seems to be no longer neccessary.
72 - use libecb instead of our own home-grown gcc hacks.
73 - document alternatives to Coro::LWP. Please use them :)
74 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
75 sigaltstack. Really. wine suffers from it, erlang suffers from it,
76 and it's known since at least 2006.
77
785.372 Wed Feb 23 06:14:30 CET 2011
79 - apparently mingw doesn't provide a working gettimeofday, try to
80 work around that by relying on Time::HiRes (indirectly brought to
81 my attention by Max Maischein).
82 - fix some portability issues when Time::HiRes was used.
83
845.371 Mon Feb 21 14:36:08 CET 2011
85 - backport to windows process emulation code again.
86
875.37 Sat Feb 19 07:49:44 CET 2011
88 - add a big "Coro thread life cycle" section to "man Coro".
89 - try a tentative workaround against the breakage that 5.13 has
90 introduced without depreciation period. sigh.
91 - no longer use Time::HiRes if gettimeofday is available, which
92 saves quite a lot of memory.
93
945.36 Sun Feb 13 05:33:41 CET 2011
95 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
96 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
97 method call.
98 - undocument Coro::Timer::sleep and obsolete whole module.
99 - optimise Coro::Timer::timeout memory and cpu usage.
100 - slightly updated Coro::Intro for recent changes.
101 - do not initialise PL_dirty anymore.
102
1035.25 Thu Nov 11 01:08:39 CET 2010
104 - try a different approach on netbsd - netbsd 5 finally has marginally
105 working pthreads, but still broken ucontext/sigaltstack.
106 - openbsd 4.8 finally got their act together, Coro works out of the box
107 with asm, setjmp and pthreads (no change, just informational).
108
1095.24 Sat Oct 23 11:27:12 CEST 2010
110 - port to the EV 4.0 API.
111 - work around bugs in mingw32, making strawberry perl work
112 out of the box.
113 - correctly modify Coro::AIO function prototypes
114 so that they reflect the "no optional parameters" rule.
115 - "ported" libcoro to C++.
116
1175.23 Mon May 17 18:50:42 CEST 2010
118 - be more resistant to ordering changes when initialising
119 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
120 Waldorf).
121 - document that perl 5.12 deliberately removed support for cloning.
122
1235.22 Wed Apr 14 03:55:35 CEST 2010
124 - correctly return udnef on errors in Coro::Handle::read/write
125 (testcase by Marc Mims).
126 - convert Coro::Util into a "perl compatibility wrapper" - the functions
127 are less useful now, but are drop-in replacements for existing
128 functions, listing better alternatives in the documentation. This also
129 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
130 with Coro::Util::inet_aton.
131 - do not override $Coro::idle unconditionally in Coro.pm, as other
132 modules could have provided their own idle coro already
133 (for exmaple, Coro::AnyEvent).
134 - fix Coro::Util::gethost* functions.
135 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
136
1375.21 Wed Dec 16 07:19:51 CET 2009
138 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
139 together.
140 - add some examples on how to combine other event loops with Coro in
141 Coro::AnyEvent, and how to run it (and not to block). Seems to be
142 the most common source of confusion.
143 - try to catch people naively blocking in an event callback.
144 - work around the perl filehandle bug issue in conjunction with
145 older common::sense (as indirectly pointed out by ZSystem).
146 - clarify the "not from signal handlers" section.
147
1485.2 Sun Oct 4 14:54:24 CEST 2009
149 - Coro::Storable destroyed the prototypes of the functions it wrapped.
150 - export rouse_cb and rouse_wait by default now.
151 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
152 - new method $state->swap_sv.
153 - added section on "windows process emulation" to the manpage,
154 after a not-so-fruitful (nor-friendly) "discussion" with chip
155 salzenberg (discussion implies arguments, but his only arguments
156 were ad-hominems, one wonders why he started it in the first
157 place). I hope this explains it well enough for him to understand,
158 and maybe well enough for others to understand.
159 - use common::sense everywhere now.
160 - idle callbacks are no longer supported, use idle coros instead.
161 - print a thread listing when a deadlock is detected.
162
1635.17 Sat Aug 22 23:09:31 CEST 2009
164 - work around a bug in the perl debugger causing crashes
165 when running under the debugger by marking _pool_handler as nodebug.
166 - speed up Coro::async considerably.
167 - try some hacks to get netbsd to work "more often" - their broken
168 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
169 - convert Coro to AE 5.0 API.
170
1715.162 Tue Jul 28 04:04:03 CEST 2009
172 - perl 5.8.2 is now minimum requirement.
173 - skip t/19_handle.t on broken windows perls.
174
1755.161 Wed Jul 22 04:47:38 CEST 2009
176 - Coro::AnyEvent::poll could have a different prototype when EV was
177 used as backend (analyzed by Tatsuhiko Miyagawa).
178 - Coro::AnyEvent errornously initialised the event loop when loaded,
179 not on demand.
180 - try to workaround rare */t/01_unblock.t failures.
181
1825.16 Tue Jul 21 01:44:37 CEST 2009
183 - Coro::AnyEvent failed to hook into the event loop
184 when no threads had been readied between detecting
185 the event loop and actually running it.
186 - considerably speed up Coro::Select by taking avdantage
187 of AnyEvent > 4.8 and some other optimisations.
188 - implement paragraph readline mode in Coro::Handle
189 (based on patches by Zsbán Ambrus).
190 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
191 and analyzed by Yasuhiro MATSUMOTO).
192 - clarified libcoro license and copyright.
193 - someone stole my EXTRA_META!!!
194 - implement Coro::Select::patch_pp_sselect and it's brother,
195 for hardcode select overriding.
196
1975.151 Mon Jul 6 05:41:57 CEST 2009
198 - backport to windows process emulation code again (patch by
199 Yasuhiro MATSUMOTO).
200 - slightly update Coro::MakeMaker.
201
2025.15 Tue Jun 30 10:28:06 CEST 2009
203 - deprecate Coro::Socket, document how to get ipv6 support via
204 AnyEvent::Socket instead.
205 - implement signal->wait ($cb) interface, similar to semaphores.
206 - work around SvOK not supporting getmagic, so we have to getmagic
207 to test for undef :/ (reported by Matthias Waldorf).
208 - load Coro::AnyEvent in all modules using AnyEvent.
209 - work around perl corrupting our internal data structures,
210 reported by Tokuhiro Matsuno.
211 - enable per-coroutine real and cpu time gathering
212 (Coro::State::enable_times).
213
2145.14 Wed Jun 24 01:37:48 CEST 2009
215 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
216 as this seems to have been a difficult concept (poll, sleep, idle,
217 idle_upto).
218 - add Coro::AnyEvent::readable/writable functions.
219 - clarify Coro::EV/Event/AnyEvent manpages.
220 - free per-thread global scalars in the thread calling ->cancel, to
221 avoid crashes when $_, $@ etc., are magical but some of those
222 had already been freed.
223 - "unexperimentalise" the callback interface for Coro::Semaphore.
224 - speed up ready queue management/context switching by using a linked
225 list instead of an array (~5-10%).
226 - implement "watch" command in Coro::Debug shells.
227 - for fun, implement time-slicing as an example in the manpage.
228 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
229 Coro::Event, use the normal AnyEvent handling instead of dieing
230 (the same is true for Coro::Handle).
231 - properly document Coro::EV::timed_io_once.
232 - avoid unneccessary ->cancel calls in Coro::Handle.
233 - maybe make it work on mingw32 with win32 backend
234 (based on patch by Yasuhiro Matsumoto).
235
2365.132 Fri May 29 09:00:39 CEST 2009
237 - do not keep a reference to the argument itself in
238 Coro::Semaphore::guard, as it could change later.
239 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
240 should support a prepare callback.
241
2425.131 Mon Mar 16 23:20:37 CET 2009
243 - implement and document Coro->suspend, Coro->resume.
244 - fix Coro::Select implementation to not (often) close
245 the passed file descriptors (testcase provided by pippijn).
246
2475.13 Mon Dec 15 21:51:42 CET 2008
248 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
249 - don't set diehook to C<undef> but instead to NULL, to avoid
250 spurious warnings.
251 - fix a lot of bugs in Coro::SemaphoreSet.
252 - Coro::SemaphoreSet will less often create a semaphore needlessly.
253 - add Coro::SemaphoreSet::count and wait methods.
254 - take advantage of the new Guard module.
255 - deprecate Coro::guard.
256 - try to fix the dreaded 01_unblock tests once more. I hate it when
257 testsuites need more fixing than the code they are supposed to test.
258 - croak in more cases when a required callback isn't resolvable.
259 - fix some minor issues in Coro::State->call/eval.
260 - use current coroutine context instead of a temporary one
261 when temporarily switching to another coroutine.
262 - do not call C-level on_destroy handlers during global destruction,
263 to avoid needless segfaults.
264
2655.12 Sun Dec 7 13:30:38 CET 2008
266 - add default config for MirOS, which seems to be a bug-to-bug
267 compatible fork of openbsd ("world domination by releasing
268 openbsd cvs before the openbsd folks do it" or so :).
269 - free_padlist did destroy the names pad, not good, but didn't
270 seem to bother perl - this could fix issues such as eval ""
271 inside a function called from multiple coroutines.
272 - use a different method to detect destruction time.
273 - be more careful when freeing padlists just before global
274 destruction.
275 - fixed and expanded the call/cc example.
276 - renamed _terminate to _coro_run.
277 - new method Coro::Channel->shutdown.
278 - try pthreads on openbsd <4.4 (broken sigaltstack, will
279 pthreads fare better?).
280 - be less picky about destroying "a" running coroutine.
281
2825.11 Tue Nov 25 21:49:05 CET 2008
283 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
284 working code with failed assertions, introducing more bugs than
285 it fixes, requiring elaborate workarounds :(
286
2875.1 Mon Nov 24 08:54:59 CET 2008
288 - wrote a small introductory tutorial - Coro::Intro.
289 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
290 - Coro::Select did errornously dup the file descriptors
291 and didn't work with all AnyEvent backends.
292 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
293 LWP data transfers.
294 - disassociate c contexts from coro objects - this is agruably more
295 correct, but mostly allows sharing of cctxs between coro and state
296 objects, for added memory savings and speed increases.
297 - bumped $Coro::POOL_RSS up to 32kb by default.
298 - no longer set the optype to OP_CUSTOM, as B::* understandably
299 doesn't like this very much (and we *are* a type of entersub).
300 - implement state cloning, just to prove that call/cc can be done.
301 - automatically load Coro::AnyEvent in Coro::Handle.
302 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
303 do not have this method (and don't need it either).
304 - speed up generic anyevent methods in Coro::Handle by using rouse
305 callbacks.
306 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
307 others. It also makes the debugger happier, as you can trace
308 through the idle threads now.
309 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
310 - eg/event was pretty broken.
311 - better 5.8.6 compatibility.
312
3135.0 Thu Nov 20 10:35:05 CET 2008
14 - NEW ARCHITECTURE: use the latest 4.x version if you experience 314 - NEW ARCHITECTURE: use the latest 4.x version if you experience
15 stability issues. 315 stability issues.
16 - bump API version to 7 - all dependents must be recompiled. 316 - bump API version to 7 - all dependents must be recompiled.
317 - removed timed_* functions - they were not being used anyways
318 and should be replaced by a more generic mechanism -
319 and were annoying to support anyways :)
320 - removed SemaphoreSet's waiter method - use sem method instead.
17 - Coro::Semaphore->adjust didn't correctly wake up enough waiters. 321 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
18 - async_pool did free a scalar value twice 322 - async_pool did free a scalar value twice
19 ("Attempt to unreference..."). 323 ("Attempt to unreference...").
20 - fix a longstanding bug where calling terminate on a coro that 324 - fix a longstanding bug where calling terminate on a coro that
21 was waiting for a semaphore that was just becoming available 325 was waiting for a semaphore that was just becoming available
22 would cause a deadlock (semaphore would get into a state where 326 would cause a deadlock (semaphore would get into a state where
23 it was available but waiters were waiting(. 327 it was available but waiters were still blocked).
24 - avoid deadlock in Coro::Channel if maxsize == 1 (Richard Hundt). 328 - calling throw on a coroutine that is waiting for a semaphore will
329 no longer make it acquire the semaphore (and thus leak a count).
25 - perl's process emulation is now not even theoretically supported 330 - perl's process emulation is now not even theoretically supported
26 anymore. 331 anymore.
332 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
333 conversion of callback-style to blocking-style.
334 - new methods $coro->schedule_to and ->cede_to, to specifically
335 schedule or cede to a specific coroutine.
336 - new function Coro::Semaphore::wait.
337 - use named constants in Coro::Channel (Richard Hundt).
27 - directly patch the entersub opcode calling SLF functions (cede, 338 - directly patch the entersub opcode calling SLF functions (cede,
28 transfer and so on). this does speed up context switching, but 339 transfer and so on). this does speed up context switching, but
29 more importanly, it frees us from the hardcoded behaviour of 340 more importanly, it frees us from the hardcoded behaviour of
30 entersub, so we might actually be able to return something from 341 entersub, so we might actually be able to return something from
31 those functions and atcually create new ones. 342 those functions and atcually create new ones.
… …
34 as "threaded" perls are running at half speed anyways). 345 as "threaded" perls are running at half speed anyways).
35 - implement execute_slf (schedule-like-function) interface that makes 346 - implement execute_slf (schedule-like-function) interface that makes
36 it possible to implement schedule-like-functions in XS. 347 it possible to implement schedule-like-functions in XS.
37 - use new SLF interface to massively speed up Coro::EV by roughly a 348 - use new SLF interface to massively speed up Coro::EV by roughly a
38 factor of two. 349 factor of two.
39 - removed timed_down and timed_guard functions - they were not being
40 used anyways and should be replaced by a more generic mechanism -
41 and were annoying to support anyways :)
42 - used new SLF interface to massively speed up Coro::Semaphore by a 350 - used new SLF interface to massively speed up Coro::Semaphore by a
43 factor of three. 351 factor of three.
44 - used new SLF interface to speed up Coro::AIO by roughly a factor of 352 - used new SLF interface to speed up Coro::AIO by roughly a factor of
45 four and reduce its memory usage considerably. 353 four and reduce its memory usage considerably.
354 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
355 for a nice speedup and vastly more correct behaviour. Also implement
356 a new method "sem" to get at the underlying semaphore object.
357 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
358 (in comparison) 20-40% speedup.
359 - used new SLF interface to reimplement Coro::Signal gaining
360 some unknown (because I was too lazy), but certain, speedup, and also
361 making signals reliable for the first time.
362 - used new SLF interface and other optimisations to speed up async_pool
363 by a factor of two. It also doesn't rely on perl's exception mechanism
364 to exit anymore. The overhead for terminating an async_pool, coro over
365 a normal async is now very small.
366 - sped up coroutine creation/destruction by 40%.
46 - forgot to include Coro/libcoro/README in the dist for all these years. 367 - forgot to include Coro/libcoro/README in the dist for all these years.
47 - work around a freebsd pthreads bug (manual testcancel is required as 368 - work around a freebsd pthreads bug (manual testcancel is required as
48 pthread_cond_wait isn't a cancellation point on freebsd). 369 pthread_cond_wait isn't a cancellation point on freebsd).
370 - use new rouse functions to speed up and simplify Coro::BDB.
371 - make "prefer perl native functions" work with threaded perls.
372 - condense Coro::Debug ps output, hint at v and w flags.
373 - (libcoro) lots of minor cleanups and portability improvements.
374
3754.914 Wed Nov 19 12:54:18 CET 2008
376 - fix a disastrous bug in the readline optimisation
377 introduced in 4.801.
49 378
504.913 Sat Nov 15 07:58:28 CET 2008 3794.913 Sat Nov 15 07:58:28 CET 2008
51 - async_pool did free a scalar value twice 380 - async_pool did free a scalar value twice
52 ("Attempt to unreference..."). 381 ("Attempt to unreference...").
53 382

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines