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

Comparing Coro/Changes (file contents):
Revision 1.406 by root, Thu Nov 13 06:15:37 2008 UTC vs.
Revision 1.567 by root, Sun May 22 20:13:33 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
7TODO: tutorial
8TODO: in_eval, PL_tained
9 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
10 destroyed and cleaned up automatically (e.g. async { schedule }).
11 - Coro::State->is_destroyed is now called is_zombie.
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
385.372 Wed Feb 23 06:14:30 CET 2011
39 - apparently mingw doesn't provide a working gettimeofday, try to
40 work around that by relying on Time::HiRes (indirectly brought to
41 my attention by Max Maischein).
42 - fix some portability issues when Time::HiRes was used.
43
445.371 Mon Feb 21 14:36:08 CET 2011
45 - backport to windows process emulation code again.
46
475.37 Sat Feb 19 07:49:44 CET 2011
48 - add a big "Coro thread life cycle" section to "man Coro".
49 - try a tentative workaround against the breakage that 5.13 has
50 introduced without depreciation period. sigh.
51 - no longer use Time::HiRes if gettimeofday is available, which
52 saves quite a lot of memory.
53
545.36 Sun Feb 13 05:33:41 CET 2011
55 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
56 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
57 method call.
58 - undocument Coro::Timer::sleep and obsolete whole module.
59 - optimise Coro::Timer::timeout memory and cpu usage.
60 - slightly updated Coro::Intro for recent changes.
61 - do not initialise PL_dirty anymore.
62
635.25 Thu Nov 11 01:08:39 CET 2010
64 - try a different approach on netbsd - netbsd 5 finally has marginally
65 working pthreads, but still broken ucontext/sigaltstack.
66 - openbsd 4.8 finally got their act together, Coro works out of the box
67 with asm, setjmp and pthreads (no change, just informational).
68
695.24 Sat Oct 23 11:27:12 CEST 2010
70 - port to the EV 4.0 API.
71 - work around bugs in mingw32, making strawberry perl work
72 out of the box.
73 - correctly modify Coro::AIO function prototypes
74 so that they reflect the "no optional parameters" rule.
75 - "ported" libcoro to C++.
76
775.23 Mon May 17 18:50:42 CEST 2010
78 - be more resistant to ordering changes when initialising
79 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
80 Waldorf).
81 - document that perl 5.12 deliberately removed support for cloning.
82
835.22 Wed Apr 14 03:55:35 CEST 2010
84 - correctly return udnef on errors in Coro::Handle::read/write
85 (testcase by Marc Mims).
86 - convert Coro::Util into a "perl compatibility wrapper" - the functions
87 are less useful now, but are drop-in replacements for existing
88 functions, listing better alternatives in the documentation. This also
89 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
90 with Coro::Util::inet_aton.
91 - do not override $Coro::idle unconditionally in Coro.pm, as other
92 modules could have provided their own idle coro already
93 (for exmaple, Coro::AnyEvent).
94 - fix Coro::Util::gethost* functions.
95 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
96
975.21 Wed Dec 16 07:19:51 CET 2009
98 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
99 together.
100 - add some examples on how to combine other event loops with Coro in
101 Coro::AnyEvent, and how to run it (and not to block). Seems to be
102 the most common source of confusion.
103 - try to catch people naively blocking in an event callback.
104 - work around the perl filehandle bug issue in conjunction with
105 older common::sense (as indirectly pointed out by ZSystem).
106 - clarify the "not from signal handlers" section.
107
1085.2 Sun Oct 4 14:54:24 CEST 2009
109 - Coro::Storable destroyed the prototypes of the functions it wrapped.
110 - export rouse_cb and rouse_wait by default now.
111 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
112 - new method $state->swap_sv.
113 - added section on "windows process emulation" to the manpage,
114 after a not-so-fruitful (nor-friendly) "discussion" with chip
115 salzenberg (discussion implies arguments, but his only arguments
116 were ad-hominems, one wonders why he started it in the first
117 place). I hope this explains it well enough for him to understand,
118 and maybe well enough for others to understand.
119 - use common::sense everywhere now.
120 - idle callbacks are no longer supported, use idle coros instead.
121 - print a thread listing when a deadlock is detected.
122
1235.17 Sat Aug 22 23:09:31 CEST 2009
124 - work around a bug in the perl debugger causing crashes
125 when running under the debugger by marking _pool_handler as nodebug.
126 - speed up Coro::async considerably.
127 - try some hacks to get netbsd to work "more often" - their broken
128 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
129 - convert Coro to AE 5.0 API.
130
1315.162 Tue Jul 28 04:04:03 CEST 2009
132 - perl 5.8.2 is now minimum requirement.
133 - skip t/19_handle.t on broken windows perls.
134
1355.161 Wed Jul 22 04:47:38 CEST 2009
136 - Coro::AnyEvent::poll could have a different prototype when EV was
137 used as backend (analyzed by Tatsuhiko Miyagawa).
138 - Coro::AnyEvent errornously initialised the event loop when loaded,
139 not on demand.
140 - try to workaround rare */t/01_unblock.t failures.
141
1425.16 Tue Jul 21 01:44:37 CEST 2009
143 - Coro::AnyEvent failed to hook into the event loop
144 when no threads had been readied between detecting
145 the event loop and actually running it.
146 - considerably speed up Coro::Select by taking avdantage
147 of AnyEvent > 4.8 and some other optimisations.
148 - implement paragraph readline mode in Coro::Handle
149 (based on patches by Zsbán Ambrus).
150 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
151 and analyzed by Yasuhiro MATSUMOTO).
152 - clarified libcoro license and copyright.
153 - someone stole my EXTRA_META!!!
154 - implement Coro::Select::patch_pp_sselect and it's brother,
155 for hardcode select overriding.
156
1575.151 Mon Jul 6 05:41:57 CEST 2009
158 - backport to windows process emulation code again (patch by
159 Yasuhiro MATSUMOTO).
160 - slightly update Coro::MakeMaker.
161
1625.15 Tue Jun 30 10:28:06 CEST 2009
163 - deprecate Coro::Socket, document how to get ipv6 support via
164 AnyEvent::Socket instead.
165 - implement signal->wait ($cb) interface, similar to semaphores.
166 - work around SvOK not supporting getmagic, so we have to getmagic
167 to test for undef :/ (reported by Matthias Waldorf).
168 - load Coro::AnyEvent in all modules using AnyEvent.
169 - work around perl corrupting our internal data structures,
170 reported by Tokuhiro Matsuno.
171 - enable per-coroutine real and cpu time gathering
172 (Coro::State::enable_times).
173
1745.14 Wed Jun 24 01:37:48 CEST 2009
175 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
176 as this seems to have been a difficult concept (poll, sleep, idle,
177 idle_upto).
178 - add Coro::AnyEvent::readable/writable functions.
179 - clarify Coro::EV/Event/AnyEvent manpages.
180 - free per-thread global scalars in the thread calling ->cancel, to
181 avoid crashes when $_, $@ etc., are magical but some of those
182 had already been freed.
183 - "unexperimentalise" the callback interface for Coro::Semaphore.
184 - speed up ready queue management/context switching by using a linked
185 list instead of an array (~5-10%).
186 - implement "watch" command in Coro::Debug shells.
187 - for fun, implement time-slicing as an example in the manpage.
188 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
189 Coro::Event, use the normal AnyEvent handling instead of dieing
190 (the same is true for Coro::Handle).
191 - properly document Coro::EV::timed_io_once.
192 - avoid unneccessary ->cancel calls in Coro::Handle.
193 - maybe make it work on mingw32 with win32 backend
194 (based on patch by Yasuhiro Matsumoto).
195
1965.132 Fri May 29 09:00:39 CEST 2009
197 - do not keep a reference to the argument itself in
198 Coro::Semaphore::guard, as it could change later.
199 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
200 should support a prepare callback.
201
2025.131 Mon Mar 16 23:20:37 CET 2009
203 - implement and document Coro->suspend, Coro->resume.
204 - fix Coro::Select implementation to not (often) close
205 the passed file descriptors (testcase provided by pippijn).
206
2075.13 Mon Dec 15 21:51:42 CET 2008
208 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
209 - don't set diehook to C<undef> but instead to NULL, to avoid
210 spurious warnings.
211 - fix a lot of bugs in Coro::SemaphoreSet.
212 - Coro::SemaphoreSet will less often create a semaphore needlessly.
213 - add Coro::SemaphoreSet::count and wait methods.
214 - take advantage of the new Guard module.
215 - deprecate Coro::guard.
216 - try to fix the dreaded 01_unblock tests once more. I hate it when
217 testsuites need more fixing than the code they are supposed to test.
218 - croak in more cases when a required callback isn't resolvable.
219 - fix some minor issues in Coro::State->call/eval.
220 - use current coroutine context instead of a temporary one
221 when temporarily switching to another coroutine.
222 - do not call C-level on_destroy handlers during global destruction,
223 to avoid needless segfaults.
224
2255.12 Sun Dec 7 13:30:38 CET 2008
226 - add default config for MirOS, which seems to be a bug-to-bug
227 compatible fork of openbsd ("world domination by releasing
228 openbsd cvs before the openbsd folks do it" or so :).
229 - free_padlist did destroy the names pad, not good, but didn't
230 seem to bother perl - this could fix issues such as eval ""
231 inside a function called from multiple coroutines.
232 - use a different method to detect destruction time.
233 - be more careful when freeing padlists just before global
234 destruction.
235 - fixed and expanded the call/cc example.
236 - renamed _terminate to _coro_run.
237 - new method Coro::Channel->shutdown.
238 - try pthreads on openbsd <4.4 (broken sigaltstack, will
239 pthreads fare better?).
240 - be less picky about destroying "a" running coroutine.
241
2425.11 Tue Nov 25 21:49:05 CET 2008
243 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
244 working code with failed assertions, introducing more bugs than
245 it fixes, requiring elaborate workarounds :(
246
2475.1 Mon Nov 24 08:54:59 CET 2008
248 - wrote a small introductory tutorial - Coro::Intro.
249 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
250 - Coro::Select did errornously dup the file descriptors
251 and didn't work with all AnyEvent backends.
252 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
253 LWP data transfers.
254 - disassociate c contexts from coro objects - this is agruably more
255 correct, but mostly allows sharing of cctxs between coro and state
256 objects, for added memory savings and speed increases.
257 - bumped $Coro::POOL_RSS up to 32kb by default.
258 - no longer set the optype to OP_CUSTOM, as B::* understandably
259 doesn't like this very much (and we *are* a type of entersub).
260 - implement state cloning, just to prove that call/cc can be done.
261 - automatically load Coro::AnyEvent in Coro::Handle.
262 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
263 do not have this method (and don't need it either).
264 - speed up generic anyevent methods in Coro::Handle by using rouse
265 callbacks.
266 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
267 others. It also makes the debugger happier, as you can trace
268 through the idle threads now.
269 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
270 - eg/event was pretty broken.
271 - better 5.8.6 compatibility.
272
2735.0 Thu Nov 20 10:35:05 CET 2008
274 - NEW ARCHITECTURE: use the latest 4.x version if you experience
275 stability issues.
276 - bump API version to 7 - all dependents must be recompiled.
277 - removed timed_* functions - they were not being used anyways
278 and should be replaced by a more generic mechanism -
279 and were annoying to support anyways :)
280 - removed SemaphoreSet's waiter method - use sem method instead.
281 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
282 - async_pool did free a scalar value twice
283 ("Attempt to unreference...").
284 - fix a longstanding bug where calling terminate on a coro that
285 was waiting for a semaphore that was just becoming available
286 would cause a deadlock (semaphore would get into a state where
287 it was available but waiters were still blocked).
288 - calling throw on a coroutine that is waiting for a semaphore will
289 no longer make it acquire the semaphore (and thus leak a count).
290 - perl's process emulation is now not even theoretically supported
291 anymore.
292 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
293 conversion of callback-style to blocking-style.
294 - new methods $coro->schedule_to and ->cede_to, to specifically
295 schedule or cede to a specific coroutine.
296 - new function Coro::Semaphore::wait.
297 - use named constants in Coro::Channel (Richard Hundt).
298 - directly patch the entersub opcode calling SLF functions (cede,
299 transfer and so on). this does speed up context switching, but
300 more importanly, it frees us from the hardcoded behaviour of
301 entersub, so we might actually be able to return something from
302 those functions and atcually create new ones.
303 - take advantage of __builtin_frame_address on gcc.
304 - expose THX in coroapi (not sure whether this was a wise decision,
305 as "threaded" perls are running at half speed anyways).
306 - implement execute_slf (schedule-like-function) interface that makes
307 it possible to implement schedule-like-functions in XS.
308 - use new SLF interface to massively speed up Coro::EV by roughly a
309 factor of two.
310 - used new SLF interface to massively speed up Coro::Semaphore by a
311 factor of three.
312 - used new SLF interface to speed up Coro::AIO by roughly a factor of
313 four and reduce its memory usage considerably.
314 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
315 for a nice speedup and vastly more correct behaviour. Also implement
316 a new method "sem" to get at the underlying semaphore object.
317 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
318 (in comparison) 20-40% speedup.
319 - used new SLF interface to reimplement Coro::Signal gaining
320 some unknown (because I was too lazy), but certain, speedup, and also
321 making signals reliable for the first time.
322 - used new SLF interface and other optimisations to speed up async_pool
323 by a factor of two. It also doesn't rely on perl's exception mechanism
324 to exit anymore. The overhead for terminating an async_pool, coro over
325 a normal async is now very small.
326 - sped up coroutine creation/destruction by 40%.
327 - forgot to include Coro/libcoro/README in the dist for all these years.
328 - work around a freebsd pthreads bug (manual testcancel is required as
329 pthread_cond_wait isn't a cancellation point on freebsd).
330 - use new rouse functions to speed up and simplify Coro::BDB.
331 - make "prefer perl native functions" work with threaded perls.
332 - condense Coro::Debug ps output, hint at v and w flags.
333 - (libcoro) lots of minor cleanups and portability improvements.
334
3354.914 Wed Nov 19 12:54:18 CET 2008
336 - fix a disastrous bug in the readline optimisation
337 introduced in 4.801.
338
3394.913 Sat Nov 15 07:58:28 CET 2008
340 - async_pool did free a scalar value twice
341 ("Attempt to unreference...").
342
3434.912 Thu Nov 13 18:31:23 CET 2008
8 - minor cleanups. 344 - minor cleanups.
9 - use much larger stacks on linux and perl < 5.8.8. 345 - use much larger stacks on linux and perl < 5.8.8.
10 - Coro::Debug::new_unix_server did not unlink the socket 346 - Coro::Debug::new_unix_server did not unlink the socket
11 when destroyed. 347 when destroyed.
12 348

Diff Legend

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