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

Comparing Coro/Changes (file contents):
Revision 1.402 by root, Tue Nov 11 03:26:58 2008 UTC vs.
Revision 1.571 by root, Tue Jun 28 22:18:57 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 - implement a JIT compiler for part of the thread switch code,
12 which gives a 50% speed improvement on threaded perls, and
13 about 4% on non-thraeded perls (so threaded perls now finally
14 reach about half the speed of non-threaded perls).
15 - avoid DEFSV and ERRSV, giving another 10% improvement
16 in thread switching.
17 - Coro::State->is_destroyed is now called is_zombie.
18 - implement a Coro->safe_cancel method that might fail, but
19 cancels in a "safer" way if it succeeds.
20 - get rid of two hash-accesses when initialising a new Coro - this
21 speeds up coro creation by almost a factor of two.
22 - croak when a coro that is being cancelled tries to block
23 (e.g. while executing a guard block), instead of crashing or
24 deadlocking.
25 - use a more robust and also faster method to identify Coro::State
26 objects - speeds up everything a bit.
27 - implement Coro->cancel in XS for a 20% speed improvement, and to
28 be able to implement mutual cancellation.
29 - speed up context switches by a percent or two by more efficiently
30 allocating context stack entries.
31 - implement Coro->join and Coro->on_destroy in XS for a speedup and
32 a reduction in memory use.
33 - cancelling a coro while it itself is cancelling another coro is
34 now supported and working, instead of triggering an assertion.
35 - be a bit more crash-resistant when calling (buggy) on_destroy
36 callbacks (best effort).
37 - move on_destroy into the slf_frame, to allow extension slf
38 functions to have destructors.
39 - get rid if coro refcounting - simply crash in other interpreter
40 threads by nulling the pointers on clone.
41 - simplify warn/die hook handling when loading Coro - the convoluted
42 logic seems to be no longer neccessary.
43 - use libecb instead of our own home-grown gcc hacks.
44 - document alternatives to Coro::LWP. Please use them :)
45 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
46 sigaltstack.
47
485.372 Wed Feb 23 06:14:30 CET 2011
49 - apparently mingw doesn't provide a working gettimeofday, try to
50 work around that by relying on Time::HiRes (indirectly brought to
51 my attention by Max Maischein).
52 - fix some portability issues when Time::HiRes was used.
53
545.371 Mon Feb 21 14:36:08 CET 2011
55 - backport to windows process emulation code again.
56
575.37 Sat Feb 19 07:49:44 CET 2011
58 - add a big "Coro thread life cycle" section to "man Coro".
59 - try a tentative workaround against the breakage that 5.13 has
60 introduced without depreciation period. sigh.
61 - no longer use Time::HiRes if gettimeofday is available, which
62 saves quite a lot of memory.
63
645.36 Sun Feb 13 05:33:41 CET 2011
65 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
66 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
67 method call.
68 - undocument Coro::Timer::sleep and obsolete whole module.
69 - optimise Coro::Timer::timeout memory and cpu usage.
70 - slightly updated Coro::Intro for recent changes.
71 - do not initialise PL_dirty anymore.
72
735.25 Thu Nov 11 01:08:39 CET 2010
74 - try a different approach on netbsd - netbsd 5 finally has marginally
75 working pthreads, but still broken ucontext/sigaltstack.
76 - openbsd 4.8 finally got their act together, Coro works out of the box
77 with asm, setjmp and pthreads (no change, just informational).
78
795.24 Sat Oct 23 11:27:12 CEST 2010
80 - port to the EV 4.0 API.
81 - work around bugs in mingw32, making strawberry perl work
82 out of the box.
83 - correctly modify Coro::AIO function prototypes
84 so that they reflect the "no optional parameters" rule.
85 - "ported" libcoro to C++.
86
875.23 Mon May 17 18:50:42 CEST 2010
88 - be more resistant to ordering changes when initialising
89 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
90 Waldorf).
91 - document that perl 5.12 deliberately removed support for cloning.
92
935.22 Wed Apr 14 03:55:35 CEST 2010
94 - correctly return udnef on errors in Coro::Handle::read/write
95 (testcase by Marc Mims).
96 - convert Coro::Util into a "perl compatibility wrapper" - the functions
97 are less useful now, but are drop-in replacements for existing
98 functions, listing better alternatives in the documentation. This also
99 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
100 with Coro::Util::inet_aton.
101 - do not override $Coro::idle unconditionally in Coro.pm, as other
102 modules could have provided their own idle coro already
103 (for exmaple, Coro::AnyEvent).
104 - fix Coro::Util::gethost* functions.
105 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
106
1075.21 Wed Dec 16 07:19:51 CET 2009
108 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
109 together.
110 - add some examples on how to combine other event loops with Coro in
111 Coro::AnyEvent, and how to run it (and not to block). Seems to be
112 the most common source of confusion.
113 - try to catch people naively blocking in an event callback.
114 - work around the perl filehandle bug issue in conjunction with
115 older common::sense (as indirectly pointed out by ZSystem).
116 - clarify the "not from signal handlers" section.
117
1185.2 Sun Oct 4 14:54:24 CEST 2009
119 - Coro::Storable destroyed the prototypes of the functions it wrapped.
120 - export rouse_cb and rouse_wait by default now.
121 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
122 - new method $state->swap_sv.
123 - added section on "windows process emulation" to the manpage,
124 after a not-so-fruitful (nor-friendly) "discussion" with chip
125 salzenberg (discussion implies arguments, but his only arguments
126 were ad-hominems, one wonders why he started it in the first
127 place). I hope this explains it well enough for him to understand,
128 and maybe well enough for others to understand.
129 - use common::sense everywhere now.
130 - idle callbacks are no longer supported, use idle coros instead.
131 - print a thread listing when a deadlock is detected.
132
1335.17 Sat Aug 22 23:09:31 CEST 2009
134 - work around a bug in the perl debugger causing crashes
135 when running under the debugger by marking _pool_handler as nodebug.
136 - speed up Coro::async considerably.
137 - try some hacks to get netbsd to work "more often" - their broken
138 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
139 - convert Coro to AE 5.0 API.
140
1415.162 Tue Jul 28 04:04:03 CEST 2009
142 - perl 5.8.2 is now minimum requirement.
143 - skip t/19_handle.t on broken windows perls.
144
1455.161 Wed Jul 22 04:47:38 CEST 2009
146 - Coro::AnyEvent::poll could have a different prototype when EV was
147 used as backend (analyzed by Tatsuhiko Miyagawa).
148 - Coro::AnyEvent errornously initialised the event loop when loaded,
149 not on demand.
150 - try to workaround rare */t/01_unblock.t failures.
151
1525.16 Tue Jul 21 01:44:37 CEST 2009
153 - Coro::AnyEvent failed to hook into the event loop
154 when no threads had been readied between detecting
155 the event loop and actually running it.
156 - considerably speed up Coro::Select by taking avdantage
157 of AnyEvent > 4.8 and some other optimisations.
158 - implement paragraph readline mode in Coro::Handle
159 (based on patches by Zsbán Ambrus).
160 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
161 and analyzed by Yasuhiro MATSUMOTO).
162 - clarified libcoro license and copyright.
163 - someone stole my EXTRA_META!!!
164 - implement Coro::Select::patch_pp_sselect and it's brother,
165 for hardcode select overriding.
166
1675.151 Mon Jul 6 05:41:57 CEST 2009
168 - backport to windows process emulation code again (patch by
169 Yasuhiro MATSUMOTO).
170 - slightly update Coro::MakeMaker.
171
1725.15 Tue Jun 30 10:28:06 CEST 2009
173 - deprecate Coro::Socket, document how to get ipv6 support via
174 AnyEvent::Socket instead.
175 - implement signal->wait ($cb) interface, similar to semaphores.
176 - work around SvOK not supporting getmagic, so we have to getmagic
177 to test for undef :/ (reported by Matthias Waldorf).
178 - load Coro::AnyEvent in all modules using AnyEvent.
179 - work around perl corrupting our internal data structures,
180 reported by Tokuhiro Matsuno.
181 - enable per-coroutine real and cpu time gathering
182 (Coro::State::enable_times).
183
1845.14 Wed Jun 24 01:37:48 CEST 2009
185 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
186 as this seems to have been a difficult concept (poll, sleep, idle,
187 idle_upto).
188 - add Coro::AnyEvent::readable/writable functions.
189 - clarify Coro::EV/Event/AnyEvent manpages.
190 - free per-thread global scalars in the thread calling ->cancel, to
191 avoid crashes when $_, $@ etc., are magical but some of those
192 had already been freed.
193 - "unexperimentalise" the callback interface for Coro::Semaphore.
194 - speed up ready queue management/context switching by using a linked
195 list instead of an array (~5-10%).
196 - implement "watch" command in Coro::Debug shells.
197 - for fun, implement time-slicing as an example in the manpage.
198 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
199 Coro::Event, use the normal AnyEvent handling instead of dieing
200 (the same is true for Coro::Handle).
201 - properly document Coro::EV::timed_io_once.
202 - avoid unneccessary ->cancel calls in Coro::Handle.
203 - maybe make it work on mingw32 with win32 backend
204 (based on patch by Yasuhiro Matsumoto).
205
2065.132 Fri May 29 09:00:39 CEST 2009
207 - do not keep a reference to the argument itself in
208 Coro::Semaphore::guard, as it could change later.
209 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
210 should support a prepare callback.
211
2125.131 Mon Mar 16 23:20:37 CET 2009
213 - implement and document Coro->suspend, Coro->resume.
214 - fix Coro::Select implementation to not (often) close
215 the passed file descriptors (testcase provided by pippijn).
216
2175.13 Mon Dec 15 21:51:42 CET 2008
218 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
219 - don't set diehook to C<undef> but instead to NULL, to avoid
220 spurious warnings.
221 - fix a lot of bugs in Coro::SemaphoreSet.
222 - Coro::SemaphoreSet will less often create a semaphore needlessly.
223 - add Coro::SemaphoreSet::count and wait methods.
224 - take advantage of the new Guard module.
225 - deprecate Coro::guard.
226 - try to fix the dreaded 01_unblock tests once more. I hate it when
227 testsuites need more fixing than the code they are supposed to test.
228 - croak in more cases when a required callback isn't resolvable.
229 - fix some minor issues in Coro::State->call/eval.
230 - use current coroutine context instead of a temporary one
231 when temporarily switching to another coroutine.
232 - do not call C-level on_destroy handlers during global destruction,
233 to avoid needless segfaults.
234
2355.12 Sun Dec 7 13:30:38 CET 2008
236 - add default config for MirOS, which seems to be a bug-to-bug
237 compatible fork of openbsd ("world domination by releasing
238 openbsd cvs before the openbsd folks do it" or so :).
239 - free_padlist did destroy the names pad, not good, but didn't
240 seem to bother perl - this could fix issues such as eval ""
241 inside a function called from multiple coroutines.
242 - use a different method to detect destruction time.
243 - be more careful when freeing padlists just before global
244 destruction.
245 - fixed and expanded the call/cc example.
246 - renamed _terminate to _coro_run.
247 - new method Coro::Channel->shutdown.
248 - try pthreads on openbsd <4.4 (broken sigaltstack, will
249 pthreads fare better?).
250 - be less picky about destroying "a" running coroutine.
251
2525.11 Tue Nov 25 21:49:05 CET 2008
253 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
254 working code with failed assertions, introducing more bugs than
255 it fixes, requiring elaborate workarounds :(
256
2575.1 Mon Nov 24 08:54:59 CET 2008
258 - wrote a small introductory tutorial - Coro::Intro.
259 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
260 - Coro::Select did errornously dup the file descriptors
261 and didn't work with all AnyEvent backends.
262 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
263 LWP data transfers.
264 - disassociate c contexts from coro objects - this is agruably more
265 correct, but mostly allows sharing of cctxs between coro and state
266 objects, for added memory savings and speed increases.
267 - bumped $Coro::POOL_RSS up to 32kb by default.
268 - no longer set the optype to OP_CUSTOM, as B::* understandably
269 doesn't like this very much (and we *are* a type of entersub).
270 - implement state cloning, just to prove that call/cc can be done.
271 - automatically load Coro::AnyEvent in Coro::Handle.
272 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
273 do not have this method (and don't need it either).
274 - speed up generic anyevent methods in Coro::Handle by using rouse
275 callbacks.
276 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
277 others. It also makes the debugger happier, as you can trace
278 through the idle threads now.
279 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
280 - eg/event was pretty broken.
281 - better 5.8.6 compatibility.
282
2835.0 Thu Nov 20 10:35:05 CET 2008
284 - NEW ARCHITECTURE: use the latest 4.x version if you experience
285 stability issues.
286 - bump API version to 7 - all dependents must be recompiled.
287 - removed timed_* functions - they were not being used anyways
288 and should be replaced by a more generic mechanism -
289 and were annoying to support anyways :)
290 - removed SemaphoreSet's waiter method - use sem method instead.
291 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
292 - async_pool did free a scalar value twice
293 ("Attempt to unreference...").
294 - fix a longstanding bug where calling terminate on a coro that
295 was waiting for a semaphore that was just becoming available
296 would cause a deadlock (semaphore would get into a state where
297 it was available but waiters were still blocked).
298 - calling throw on a coroutine that is waiting for a semaphore will
299 no longer make it acquire the semaphore (and thus leak a count).
300 - perl's process emulation is now not even theoretically supported
301 anymore.
302 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
303 conversion of callback-style to blocking-style.
304 - new methods $coro->schedule_to and ->cede_to, to specifically
305 schedule or cede to a specific coroutine.
306 - new function Coro::Semaphore::wait.
307 - use named constants in Coro::Channel (Richard Hundt).
308 - directly patch the entersub opcode calling SLF functions (cede,
309 transfer and so on). this does speed up context switching, but
310 more importanly, it frees us from the hardcoded behaviour of
311 entersub, so we might actually be able to return something from
312 those functions and atcually create new ones.
313 - take advantage of __builtin_frame_address on gcc.
314 - expose THX in coroapi (not sure whether this was a wise decision,
315 as "threaded" perls are running at half speed anyways).
316 - implement execute_slf (schedule-like-function) interface that makes
317 it possible to implement schedule-like-functions in XS.
318 - use new SLF interface to massively speed up Coro::EV by roughly a
319 factor of two.
320 - used new SLF interface to massively speed up Coro::Semaphore by a
321 factor of three.
322 - used new SLF interface to speed up Coro::AIO by roughly a factor of
323 four and reduce its memory usage considerably.
324 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
325 for a nice speedup and vastly more correct behaviour. Also implement
326 a new method "sem" to get at the underlying semaphore object.
327 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
328 (in comparison) 20-40% speedup.
329 - used new SLF interface to reimplement Coro::Signal gaining
330 some unknown (because I was too lazy), but certain, speedup, and also
331 making signals reliable for the first time.
332 - used new SLF interface and other optimisations to speed up async_pool
333 by a factor of two. It also doesn't rely on perl's exception mechanism
334 to exit anymore. The overhead for terminating an async_pool, coro over
335 a normal async is now very small.
336 - sped up coroutine creation/destruction by 40%.
337 - forgot to include Coro/libcoro/README in the dist for all these years.
338 - work around a freebsd pthreads bug (manual testcancel is required as
339 pthread_cond_wait isn't a cancellation point on freebsd).
340 - use new rouse functions to speed up and simplify Coro::BDB.
341 - make "prefer perl native functions" work with threaded perls.
342 - condense Coro::Debug ps output, hint at v and w flags.
343 - (libcoro) lots of minor cleanups and portability improvements.
344
3454.914 Wed Nov 19 12:54:18 CET 2008
346 - fix a disastrous bug in the readline optimisation
347 introduced in 4.801.
348
3494.913 Sat Nov 15 07:58:28 CET 2008
350 - async_pool did free a scalar value twice
351 ("Attempt to unreference...").
352
3534.912 Thu Nov 13 18:31:23 CET 2008
354 - minor cleanups.
355 - use much larger stacks on linux and perl < 5.8.8.
356 - Coro::Debug::new_unix_server did not unlink the socket
357 when destroyed.
358
84.901 Tue Nov 11 04:26:17 CET 2008 3594.911 Tue Nov 11 04:26:17 CET 2008
9 - "port" to threaded perls. 360 - "port" to threaded perls.
10 361
114.91 Mon Nov 10 05:36:38 CET 2008 3624.91 Mon Nov 10 05:36:38 CET 2008
12 - the ->throw exception object no longer leaks. 363 - the ->throw exception object no longer leaks.
13 - creating a new cctx leaked a scopestack entry (memleak). 364 - creating a new cctx leaked a scopestack entry (memleak).

Diff Legend

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