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

Comparing Coro/Changes (file contents):
Revision 1.425 by root, Mon Nov 17 07:19:15 2008 UTC vs.
Revision 1.568 by root, Fri Jun 10 12:27:02 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 7TODO: tutorial
9TODO: aio 8TODO: in_eval, PL_tained
10TODO: Coro::Semaphore 9 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
11TODO: trow disturbs cleanup path inside pp_slf? 10 destroyed and cleaned up automatically (e.g. async { schedule }).
12TODO: throw special-case in transfer? 11 - Coro::State->is_destroyed is now called is_zombie.
135.0 12 - implement a Coro->safe_cancel method that might fail, but
13 cancels in a "safer" way if it succeeds.
14 - get rid of two hash-accesses when initialising a new Coro - this
15 speeds up coro creation by almost a factor of two.
16 - croak when a coro that is being cancelled tries to block
17 (e.g. while executing a guard block), instead of crashing or
18 deadlocking.
19 - use a more robust and also faster method to identify Coro::State
20 objects - speeds up everything a bit.
21 - implement Coro->cancel in XS for a 20% speed improvement, and to
22 be able to implement mutual cancellation.
23 - speed up context switches by a percent or two by more efficiently
24 allocating context stack entries.
25 - implement Coro->join and Coro->on_destroy in XS for a speedup and
26 a reduction in memory use.
27 - cancelling a coro while it itself is cancelling another coro is
28 now supported and working, instead of triggering an assertion.
29 - be a bit more crash-resistant when calling (buggy) on_destroy
30 callbacks (best effort).
31 - move on_destroy into the slf_frame, to allow extension slf
32 functions to have destructors.
33 - get rid if coro refcounting - simply crash in other interpreter
34 threads by nulling the pointers on clone.
35 - simplify warn/die hook handling when loading Coro - the convoluted
36 logic seems to be no longer neccessary.
37 - use libecb instead of our own home-grown gcc hacks.
38
395.372 Wed Feb 23 06:14:30 CET 2011
40 - apparently mingw doesn't provide a working gettimeofday, try to
41 work around that by relying on Time::HiRes (indirectly brought to
42 my attention by Max Maischein).
43 - fix some portability issues when Time::HiRes was used.
44
455.371 Mon Feb 21 14:36:08 CET 2011
46 - backport to windows process emulation code again.
47
485.37 Sat Feb 19 07:49:44 CET 2011
49 - add a big "Coro thread life cycle" section to "man Coro".
50 - try a tentative workaround against the breakage that 5.13 has
51 introduced without depreciation period. sigh.
52 - no longer use Time::HiRes if gettimeofday is available, which
53 saves quite a lot of memory.
54
555.36 Sun Feb 13 05:33:41 CET 2011
56 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
57 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
58 method call.
59 - undocument Coro::Timer::sleep and obsolete whole module.
60 - optimise Coro::Timer::timeout memory and cpu usage.
61 - slightly updated Coro::Intro for recent changes.
62 - do not initialise PL_dirty anymore.
63
645.25 Thu Nov 11 01:08:39 CET 2010
65 - try a different approach on netbsd - netbsd 5 finally has marginally
66 working pthreads, but still broken ucontext/sigaltstack.
67 - openbsd 4.8 finally got their act together, Coro works out of the box
68 with asm, setjmp and pthreads (no change, just informational).
69
705.24 Sat Oct 23 11:27:12 CEST 2010
71 - port to the EV 4.0 API.
72 - work around bugs in mingw32, making strawberry perl work
73 out of the box.
74 - correctly modify Coro::AIO function prototypes
75 so that they reflect the "no optional parameters" rule.
76 - "ported" libcoro to C++.
77
785.23 Mon May 17 18:50:42 CEST 2010
79 - be more resistant to ordering changes when initialising
80 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
81 Waldorf).
82 - document that perl 5.12 deliberately removed support for cloning.
83
845.22 Wed Apr 14 03:55:35 CEST 2010
85 - correctly return udnef on errors in Coro::Handle::read/write
86 (testcase by Marc Mims).
87 - convert Coro::Util into a "perl compatibility wrapper" - the functions
88 are less useful now, but are drop-in replacements for existing
89 functions, listing better alternatives in the documentation. This also
90 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
91 with Coro::Util::inet_aton.
92 - do not override $Coro::idle unconditionally in Coro.pm, as other
93 modules could have provided their own idle coro already
94 (for exmaple, Coro::AnyEvent).
95 - fix Coro::Util::gethost* functions.
96 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
97
985.21 Wed Dec 16 07:19:51 CET 2009
99 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
100 together.
101 - add some examples on how to combine other event loops with Coro in
102 Coro::AnyEvent, and how to run it (and not to block). Seems to be
103 the most common source of confusion.
104 - try to catch people naively blocking in an event callback.
105 - work around the perl filehandle bug issue in conjunction with
106 older common::sense (as indirectly pointed out by ZSystem).
107 - clarify the "not from signal handlers" section.
108
1095.2 Sun Oct 4 14:54:24 CEST 2009
110 - Coro::Storable destroyed the prototypes of the functions it wrapped.
111 - export rouse_cb and rouse_wait by default now.
112 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
113 - new method $state->swap_sv.
114 - added section on "windows process emulation" to the manpage,
115 after a not-so-fruitful (nor-friendly) "discussion" with chip
116 salzenberg (discussion implies arguments, but his only arguments
117 were ad-hominems, one wonders why he started it in the first
118 place). I hope this explains it well enough for him to understand,
119 and maybe well enough for others to understand.
120 - use common::sense everywhere now.
121 - idle callbacks are no longer supported, use idle coros instead.
122 - print a thread listing when a deadlock is detected.
123
1245.17 Sat Aug 22 23:09:31 CEST 2009
125 - work around a bug in the perl debugger causing crashes
126 when running under the debugger by marking _pool_handler as nodebug.
127 - speed up Coro::async considerably.
128 - try some hacks to get netbsd to work "more often" - their broken
129 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
130 - convert Coro to AE 5.0 API.
131
1325.162 Tue Jul 28 04:04:03 CEST 2009
133 - perl 5.8.2 is now minimum requirement.
134 - skip t/19_handle.t on broken windows perls.
135
1365.161 Wed Jul 22 04:47:38 CEST 2009
137 - Coro::AnyEvent::poll could have a different prototype when EV was
138 used as backend (analyzed by Tatsuhiko Miyagawa).
139 - Coro::AnyEvent errornously initialised the event loop when loaded,
140 not on demand.
141 - try to workaround rare */t/01_unblock.t failures.
142
1435.16 Tue Jul 21 01:44:37 CEST 2009
144 - Coro::AnyEvent failed to hook into the event loop
145 when no threads had been readied between detecting
146 the event loop and actually running it.
147 - considerably speed up Coro::Select by taking avdantage
148 of AnyEvent > 4.8 and some other optimisations.
149 - implement paragraph readline mode in Coro::Handle
150 (based on patches by Zsbán Ambrus).
151 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
152 and analyzed by Yasuhiro MATSUMOTO).
153 - clarified libcoro license and copyright.
154 - someone stole my EXTRA_META!!!
155 - implement Coro::Select::patch_pp_sselect and it's brother,
156 for hardcode select overriding.
157
1585.151 Mon Jul 6 05:41:57 CEST 2009
159 - backport to windows process emulation code again (patch by
160 Yasuhiro MATSUMOTO).
161 - slightly update Coro::MakeMaker.
162
1635.15 Tue Jun 30 10:28:06 CEST 2009
164 - deprecate Coro::Socket, document how to get ipv6 support via
165 AnyEvent::Socket instead.
166 - implement signal->wait ($cb) interface, similar to semaphores.
167 - work around SvOK not supporting getmagic, so we have to getmagic
168 to test for undef :/ (reported by Matthias Waldorf).
169 - load Coro::AnyEvent in all modules using AnyEvent.
170 - work around perl corrupting our internal data structures,
171 reported by Tokuhiro Matsuno.
172 - enable per-coroutine real and cpu time gathering
173 (Coro::State::enable_times).
174
1755.14 Wed Jun 24 01:37:48 CEST 2009
176 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
177 as this seems to have been a difficult concept (poll, sleep, idle,
178 idle_upto).
179 - add Coro::AnyEvent::readable/writable functions.
180 - clarify Coro::EV/Event/AnyEvent manpages.
181 - free per-thread global scalars in the thread calling ->cancel, to
182 avoid crashes when $_, $@ etc., are magical but some of those
183 had already been freed.
184 - "unexperimentalise" the callback interface for Coro::Semaphore.
185 - speed up ready queue management/context switching by using a linked
186 list instead of an array (~5-10%).
187 - implement "watch" command in Coro::Debug shells.
188 - for fun, implement time-slicing as an example in the manpage.
189 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
190 Coro::Event, use the normal AnyEvent handling instead of dieing
191 (the same is true for Coro::Handle).
192 - properly document Coro::EV::timed_io_once.
193 - avoid unneccessary ->cancel calls in Coro::Handle.
194 - maybe make it work on mingw32 with win32 backend
195 (based on patch by Yasuhiro Matsumoto).
196
1975.132 Fri May 29 09:00:39 CEST 2009
198 - do not keep a reference to the argument itself in
199 Coro::Semaphore::guard, as it could change later.
200 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
201 should support a prepare callback.
202
2035.131 Mon Mar 16 23:20:37 CET 2009
204 - implement and document Coro->suspend, Coro->resume.
205 - fix Coro::Select implementation to not (often) close
206 the passed file descriptors (testcase provided by pippijn).
207
2085.13 Mon Dec 15 21:51:42 CET 2008
209 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
210 - don't set diehook to C<undef> but instead to NULL, to avoid
211 spurious warnings.
212 - fix a lot of bugs in Coro::SemaphoreSet.
213 - Coro::SemaphoreSet will less often create a semaphore needlessly.
214 - add Coro::SemaphoreSet::count and wait methods.
215 - take advantage of the new Guard module.
216 - deprecate Coro::guard.
217 - try to fix the dreaded 01_unblock tests once more. I hate it when
218 testsuites need more fixing than the code they are supposed to test.
219 - croak in more cases when a required callback isn't resolvable.
220 - fix some minor issues in Coro::State->call/eval.
221 - use current coroutine context instead of a temporary one
222 when temporarily switching to another coroutine.
223 - do not call C-level on_destroy handlers during global destruction,
224 to avoid needless segfaults.
225
2265.12 Sun Dec 7 13:30:38 CET 2008
227 - add default config for MirOS, which seems to be a bug-to-bug
228 compatible fork of openbsd ("world domination by releasing
229 openbsd cvs before the openbsd folks do it" or so :).
230 - free_padlist did destroy the names pad, not good, but didn't
231 seem to bother perl - this could fix issues such as eval ""
232 inside a function called from multiple coroutines.
233 - use a different method to detect destruction time.
234 - be more careful when freeing padlists just before global
235 destruction.
236 - fixed and expanded the call/cc example.
237 - renamed _terminate to _coro_run.
238 - new method Coro::Channel->shutdown.
239 - try pthreads on openbsd <4.4 (broken sigaltstack, will
240 pthreads fare better?).
241 - be less picky about destroying "a" running coroutine.
242
2435.11 Tue Nov 25 21:49:05 CET 2008
244 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
245 working code with failed assertions, introducing more bugs than
246 it fixes, requiring elaborate workarounds :(
247
2485.1 Mon Nov 24 08:54:59 CET 2008
249 - wrote a small introductory tutorial - Coro::Intro.
250 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
251 - Coro::Select did errornously dup the file descriptors
252 and didn't work with all AnyEvent backends.
253 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
254 LWP data transfers.
255 - disassociate c contexts from coro objects - this is agruably more
256 correct, but mostly allows sharing of cctxs between coro and state
257 objects, for added memory savings and speed increases.
258 - bumped $Coro::POOL_RSS up to 32kb by default.
259 - no longer set the optype to OP_CUSTOM, as B::* understandably
260 doesn't like this very much (and we *are* a type of entersub).
261 - implement state cloning, just to prove that call/cc can be done.
262 - automatically load Coro::AnyEvent in Coro::Handle.
263 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
264 do not have this method (and don't need it either).
265 - speed up generic anyevent methods in Coro::Handle by using rouse
266 callbacks.
267 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
268 others. It also makes the debugger happier, as you can trace
269 through the idle threads now.
270 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
271 - eg/event was pretty broken.
272 - better 5.8.6 compatibility.
273
2745.0 Thu Nov 20 10:35:05 CET 2008
14 - NEW ARCHITECTURE: use the latest 4.x version if you experience 275 - NEW ARCHITECTURE: use the latest 4.x version if you experience
15 stability issues. 276 stability issues.
16 - bump API version to 7 - all dependents must be recompiled. 277 - bump API version to 7 - all dependents must be recompiled.
278 - removed timed_* functions - they were not being used anyways
279 and should be replaced by a more generic mechanism -
280 and were annoying to support anyways :)
281 - removed SemaphoreSet's waiter method - use sem method instead.
17 - Coro::Semaphore->adjust didn't correctly wake up enough waiters. 282 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
18 - async_pool did free a scalar value twice 283 - async_pool did free a scalar value twice
19 ("Attempt to unreference..."). 284 ("Attempt to unreference...").
20 - fix a longstanding bug where calling terminate on a coro that 285 - fix a longstanding bug where calling terminate on a coro that
21 was waiting for a semaphore that was just becoming available 286 was waiting for a semaphore that was just becoming available
22 would cause a deadlock (semaphore would get into a state where 287 would cause a deadlock (semaphore would get into a state where
23 it was available but waiters were waiting(. 288 it was available but waiters were still blocked).
24 - avoid deadlock in Coro::Channel if maxsize == 1 (Richard Hundt). 289 - calling throw on a coroutine that is waiting for a semaphore will
290 no longer make it acquire the semaphore (and thus leak a count).
25 - perl's process emulation is now not even theoretically supported 291 - perl's process emulation is now not even theoretically supported
26 anymore. 292 anymore.
293 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
294 conversion of callback-style to blocking-style.
295 - new methods $coro->schedule_to and ->cede_to, to specifically
296 schedule or cede to a specific coroutine.
297 - new function Coro::Semaphore::wait.
298 - use named constants in Coro::Channel (Richard Hundt).
27 - directly patch the entersub opcode calling SLF functions (cede, 299 - directly patch the entersub opcode calling SLF functions (cede,
28 transfer and so on). this does speed up context switching, but 300 transfer and so on). this does speed up context switching, but
29 more importanly, it frees us from the hardcoded behaviour of 301 more importanly, it frees us from the hardcoded behaviour of
30 entersub, so we might actually be able to return something from 302 entersub, so we might actually be able to return something from
31 those functions and atcually create new ones. 303 those functions and atcually create new ones.
… …
34 as "threaded" perls are running at half speed anyways). 306 as "threaded" perls are running at half speed anyways).
35 - implement execute_slf (schedule-like-function) interface that makes 307 - implement execute_slf (schedule-like-function) interface that makes
36 it possible to implement schedule-like-functions in XS. 308 it possible to implement schedule-like-functions in XS.
37 - use new SLF interface to massively speed up Coro::EV by roughly a 309 - use new SLF interface to massively speed up Coro::EV by roughly a
38 factor of two. 310 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 311 - used new SLF interface to massively speed up Coro::Semaphore by a
43 factor of three. 312 factor of three.
44 - used new SLF interface to speed up Coro::AIO by roughly 400% and 313 - used new SLF interface to speed up Coro::AIO by roughly a factor of
45 reduce its memory usage considerably. 314 four and reduce its memory usage considerably.
315 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
316 for a nice speedup and vastly more correct behaviour. Also implement
317 a new method "sem" to get at the underlying semaphore object.
318 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
319 (in comparison) 20-40% speedup.
320 - used new SLF interface to reimplement Coro::Signal gaining
321 some unknown (because I was too lazy), but certain, speedup, and also
322 making signals reliable for the first time.
323 - used new SLF interface and other optimisations to speed up async_pool
324 by a factor of two. It also doesn't rely on perl's exception mechanism
325 to exit anymore. The overhead for terminating an async_pool, coro over
326 a normal async is now very small.
327 - sped up coroutine creation/destruction by 40%.
46 - forgot to include Coro/libcoro/README in the dist for all these years. 328 - 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 329 - work around a freebsd pthreads bug (manual testcancel is required as
48 pthread_cond_wait isn't a cancellation point on freebsd). 330 pthread_cond_wait isn't a cancellation point on freebsd).
331 - use new rouse functions to speed up and simplify Coro::BDB.
332 - make "prefer perl native functions" work with threaded perls.
333 - condense Coro::Debug ps output, hint at v and w flags.
334 - (libcoro) lots of minor cleanups and portability improvements.
335
3364.914 Wed Nov 19 12:54:18 CET 2008
337 - fix a disastrous bug in the readline optimisation
338 introduced in 4.801.
49 339
504.913 Sat Nov 15 07:58:28 CET 2008 3404.913 Sat Nov 15 07:58:28 CET 2008
51 - async_pool did free a scalar value twice 341 - async_pool did free a scalar value twice
52 ("Attempt to unreference..."). 342 ("Attempt to unreference...").
53 343

Diff Legend

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