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

Comparing Coro/Changes (file contents):
Revision 1.374 by root, Wed Sep 24 21:20:05 2008 UTC vs.
Revision 1.555 by root, Fri Apr 29 15:43:26 2011 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::coroaio. 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)). 6TODO: cancel calling another coro that cancels ourselves => crash (CF_DELAYCANCEL?) -> canceltest
7 7
84.748 8 - implement Coro->cancel in XS for a 20% speed improvement, and to
9 be able to implement mutual cancellation.
10 - be a bit more crash-resistant when calling (buggy) on_destroy
11 callbacks (best effort).
12
135.372 Wed Feb 23 06:14:30 CET 2011
14 - apparently mingw doesn't provide a working gettimeofday, try to
15 work around that by relying on Time::HiRes (indirectly brought to
16 my attention by Max Maischein).
17 - fix some portability issues when Time::HiRes was used.
18
195.371 Mon Feb 21 14:36:08 CET 2011
20 - backport to windows process emulation code again.
21
225.37 Sat Feb 19 07:49:44 CET 2011
23 - add a big "Coro thread life cycle" section to "man Coro".
24 - try a tentative workaround against the breakage that 5.13 has
25 introduced without depreciation period. sigh.
26 - no longer use Time::HiRes if gettimeofday is available, which
27 saves quite a lot of memory.
28
295.36 Sun Feb 13 05:33:41 CET 2011
30 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
31 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
32 method call.
33 - undocument Coro::Timer::sleep and obsolete whole module.
34 - optimise Coro::Timer::timeout memory and cpu usage.
35 - slightly updated Coro::Intro for recent changes.
36 - do not initialise PL_dirty anymore.
37
385.25 Thu Nov 11 01:08:39 CET 2010
39 - try a different approach on netbsd - netbsd 5 finally has marginally
40 working pthreads, but still broken ucontext/sigaltstack.
41 - openbsd 4.8 finally got their act together, Coro works out of the box
42 with asm, setjmp and pthreads (no change, just informational).
43
445.24 Sat Oct 23 11:27:12 CEST 2010
45 - port to the EV 4.0 API.
46 - work around bugs in mingw32, making strawberry perl work
47 out of the box.
48 - correctly modify Coro::AIO function prototypes
49 so that they reflect the "no optional parameters" rule.
50 - "ported" libcoro to C++.
51
525.23 Mon May 17 18:50:42 CEST 2010
53 - be more resistant to ordering changes when initialising
54 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
55 Waldorf).
56 - document that perl 5.12 deliberately removed support for cloning.
57
585.22 Wed Apr 14 03:55:35 CEST 2010
59 - correctly return udnef on errors in Coro::Handle::read/write
60 (testcase by Marc Mims).
61 - convert Coro::Util into a "perl compatibility wrapper" - the functions
62 are less useful now, but are drop-in replacements for existing
63 functions, listing better alternatives in the documentation. This also
64 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
65 with Coro::Util::inet_aton.
66 - do not override $Coro::idle unconditionally in Coro.pm, as other
67 modules could have provided their own idle coro already
68 (for exmaple, Coro::AnyEvent).
69 - fix Coro::Util::gethost* functions.
70 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
71
725.21 Wed Dec 16 07:19:51 CET 2009
73 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
74 together.
75 - add some examples on how to combine other event loops with Coro in
76 Coro::AnyEvent, and how to run it (and not to block). Seems to be
77 the most common source of confusion.
78 - try to catch people naively blocking in an event callback.
79 - work around the perl filehandle bug issue in conjunction with
80 older common::sense (as indirectly pointed out by ZSystem).
81 - clarify the "not from signal handlers" section.
82
835.2 Sun Oct 4 14:54:24 CEST 2009
84 - Coro::Storable destroyed the prototypes of the functions it wrapped.
85 - export rouse_cb and rouse_wait by default now.
86 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
87 - new method $state->swap_sv.
88 - added section on "windows process emulation" to the manpage,
89 after a not-so-fruitful (nor-friendly) "discussion" with chip
90 salzenberg (discussion implies arguments, but his only arguments
91 were ad-hominems, one wonders why he started it in the first
92 place). I hope this explains it well enough for him to understand,
93 and maybe well enough for others to understand.
94 - use common::sense everywhere now.
95 - idle callbacks are no longer supported, use idle coros instead.
96 - print a thread listing when a deadlock is detected.
97
985.17 Sat Aug 22 23:09:31 CEST 2009
99 - work around a bug in the perl debugger causing crashes
100 when running under the debugger by marking _pool_handler as nodebug.
101 - speed up Coro::async considerably.
102 - try some hacks to get netbsd to work "more often" - their broken
103 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
104 - convert Coro to AE 5.0 API.
105
1065.162 Tue Jul 28 04:04:03 CEST 2009
107 - perl 5.8.2 is now minimum requirement.
108 - skip t/19_handle.t on broken windows perls.
109
1105.161 Wed Jul 22 04:47:38 CEST 2009
111 - Coro::AnyEvent::poll could have a different prototype when EV was
112 used as backend (analyzed by Tatsuhiko Miyagawa).
113 - Coro::AnyEvent errornously initialised the event loop when loaded,
114 not on demand.
115 - try to workaround rare */t/01_unblock.t failures.
116
1175.16 Tue Jul 21 01:44:37 CEST 2009
118 - Coro::AnyEvent failed to hook into the event loop
119 when no threads had been readied between detecting
120 the event loop and actually running it.
121 - considerably speed up Coro::Select by taking avdantage
122 of AnyEvent > 4.8 and some other optimisations.
123 - implement paragraph readline mode in Coro::Handle
124 (based on patches by Zsbán Ambrus).
125 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
126 and analyzed by Yasuhiro MATSUMOTO).
127 - clarified libcoro license and copyright.
128 - someone stole my EXTRA_META!!!
129 - implement Coro::Select::patch_pp_sselect and it's brother,
130 for hardcode select overriding.
131
1325.151 Mon Jul 6 05:41:57 CEST 2009
133 - backport to windows process emulation code again (patch by
134 Yasuhiro MATSUMOTO).
135 - slightly update Coro::MakeMaker.
136
1375.15 Tue Jun 30 10:28:06 CEST 2009
138 - deprecate Coro::Socket, document how to get ipv6 support via
139 AnyEvent::Socket instead.
140 - implement signal->wait ($cb) interface, similar to semaphores.
141 - work around SvOK not supporting getmagic, so we have to getmagic
142 to test for undef :/ (reported by Matthias Waldorf).
143 - load Coro::AnyEvent in all modules using AnyEvent.
144 - work around perl corrupting our internal data structures,
145 reported by Tokuhiro Matsuno.
146 - enable per-coroutine real and cpu time gathering
147 (Coro::State::enable_times).
148
1495.14 Wed Jun 24 01:37:48 CEST 2009
150 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
151 as this seems to have been a difficult concept (poll, sleep, idle,
152 idle_upto).
153 - add Coro::AnyEvent::readable/writable functions.
154 - clarify Coro::EV/Event/AnyEvent manpages.
155 - free per-thread global scalars in the thread calling ->cancel, to
156 avoid crashes when $_, $@ etc., are magical but some of those
157 had already been freed.
158 - "unexperimentalise" the callback interface for Coro::Semaphore.
159 - speed up ready queue management/context switching by using a linked
160 list instead of an array (~5-10%).
161 - implement "watch" command in Coro::Debug shells.
162 - for fun, implement time-slicing as an example in the manpage.
163 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
164 Coro::Event, use the normal AnyEvent handling instead of dieing
165 (the same is true for Coro::Handle).
166 - properly document Coro::EV::timed_io_once.
167 - avoid unneccessary ->cancel calls in Coro::Handle.
168 - maybe make it work on mingw32 with win32 backend
169 (based on patch by Yasuhiro Matsumoto).
170
1715.132 Fri May 29 09:00:39 CEST 2009
172 - do not keep a reference to the argument itself in
173 Coro::Semaphore::guard, as it could change later.
174 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
175 should support a prepare callback.
176
1775.131 Mon Mar 16 23:20:37 CET 2009
178 - implement and document Coro->suspend, Coro->resume.
179 - fix Coro::Select implementation to not (often) close
180 the passed file descriptors (testcase provided by pippijn).
181
1825.13 Mon Dec 15 21:51:42 CET 2008
183 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
184 - don't set diehook to C<undef> but instead to NULL, to avoid
185 spurious warnings.
186 - fix a lot of bugs in Coro::SemaphoreSet.
187 - Coro::SemaphoreSet will less often create a semaphore needlessly.
188 - add Coro::SemaphoreSet::count and wait methods.
189 - take advantage of the new Guard module.
190 - deprecate Coro::guard.
191 - try to fix the dreaded 01_unblock tests once more. I hate it when
192 testsuites need more fixing than the code they are supposed to test.
193 - croak in more cases when a required callback isn't resolvable.
194 - fix some minor issues in Coro::State->call/eval.
195 - use current coroutine context instead of a temporary one
196 when temporarily switching to another coroutine.
197 - do not call C-level on_destroy handlers during global destruction,
198 to avoid needless segfaults.
199
2005.12 Sun Dec 7 13:30:38 CET 2008
201 - add default config for MirOS, which seems to be a bug-to-bug
202 compatible fork of openbsd ("world domination by releasing
203 openbsd cvs before the openbsd folks do it" or so :).
204 - free_padlist did destroy the names pad, not good, but didn't
205 seem to bother perl - this could fix issues such as eval ""
206 inside a function called from multiple coroutines.
207 - use a different method to detect destruction time.
208 - be more careful when freeing padlists just before global
209 destruction.
210 - fixed and expanded the call/cc example.
211 - renamed _terminate to _coro_run.
212 - new method Coro::Channel->shutdown.
213 - try pthreads on openbsd <4.4 (broken sigaltstack, will
214 pthreads fare better?).
215 - be less picky about destroying "a" running coroutine.
216
2175.11 Tue Nov 25 21:49:05 CET 2008
218 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
219 working code with failed assertions, introducing more bugs than
220 it fixes, requiring elaborate workarounds :(
221
2225.1 Mon Nov 24 08:54:59 CET 2008
223 - wrote a small introductory tutorial - Coro::Intro.
224 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
225 - Coro::Select did errornously dup the file descriptors
226 and didn't work with all AnyEvent backends.
227 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
228 LWP data transfers.
229 - disassociate c contexts from coro objects - this is agruably more
230 correct, but mostly allows sharing of cctxs between coro and state
231 objects, for added memory savings and speed increases.
232 - bumped $Coro::POOL_RSS up to 32kb by default.
233 - no longer set the optype to OP_CUSTOM, as B::* understandably
234 doesn't like this very much (and we *are* a type of entersub).
235 - implement state cloning, just to prove that call/cc can be done.
236 - automatically load Coro::AnyEvent in Coro::Handle.
237 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
238 do not have this method (and don't need it either).
239 - speed up generic anyevent methods in Coro::Handle by using rouse
240 callbacks.
241 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
242 others. It also makes the debugger happier, as you can trace
243 through the idle threads now.
244 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
245 - eg/event was pretty broken.
246 - better 5.8.6 compatibility.
247
2485.0 Thu Nov 20 10:35:05 CET 2008
249 - NEW ARCHITECTURE: use the latest 4.x version if you experience
250 stability issues.
251 - bump API version to 7 - all dependents must be recompiled.
252 - removed timed_* functions - they were not being used anyways
253 and should be replaced by a more generic mechanism -
254 and were annoying to support anyways :)
255 - removed SemaphoreSet's waiter method - use sem method instead.
256 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
257 - async_pool did free a scalar value twice
258 ("Attempt to unreference...").
259 - fix a longstanding bug where calling terminate on a coro that
260 was waiting for a semaphore that was just becoming available
261 would cause a deadlock (semaphore would get into a state where
262 it was available but waiters were still blocked).
263 - calling throw on a coroutine that is waiting for a semaphore will
264 no longer make it acquire the semaphore (and thus leak a count).
265 - perl's process emulation is now not even theoretically supported
266 anymore.
267 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
268 conversion of callback-style to blocking-style.
269 - new methods $coro->schedule_to and ->cede_to, to specifically
270 schedule or cede to a specific coroutine.
271 - new function Coro::Semaphore::wait.
272 - use named constants in Coro::Channel (Richard Hundt).
273 - directly patch the entersub opcode calling SLF functions (cede,
274 transfer and so on). this does speed up context switching, but
275 more importanly, it frees us from the hardcoded behaviour of
276 entersub, so we might actually be able to return something from
277 those functions and atcually create new ones.
278 - take advantage of __builtin_frame_address on gcc.
279 - expose THX in coroapi (not sure whether this was a wise decision,
280 as "threaded" perls are running at half speed anyways).
281 - implement execute_slf (schedule-like-function) interface that makes
282 it possible to implement schedule-like-functions in XS.
283 - use new SLF interface to massively speed up Coro::EV by roughly a
284 factor of two.
285 - used new SLF interface to massively speed up Coro::Semaphore by a
286 factor of three.
287 - used new SLF interface to speed up Coro::AIO by roughly a factor of
288 four and reduce its memory usage considerably.
289 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
290 for a nice speedup and vastly more correct behaviour. Also implement
291 a new method "sem" to get at the underlying semaphore object.
292 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
293 (in comparison) 20-40% speedup.
294 - used new SLF interface to reimplement Coro::Signal gaining
295 some unknown (because I was too lazy), but certain, speedup, and also
296 making signals reliable for the first time.
297 - used new SLF interface and other optimisations to speed up async_pool
298 by a factor of two. It also doesn't rely on perl's exception mechanism
299 to exit anymore. The overhead for terminating an async_pool, coro over
300 a normal async is now very small.
301 - sped up coroutine creation/destruction by 40%.
302 - forgot to include Coro/libcoro/README in the dist for all these years.
303 - work around a freebsd pthreads bug (manual testcancel is required as
304 pthread_cond_wait isn't a cancellation point on freebsd).
305 - use new rouse functions to speed up and simplify Coro::BDB.
306 - make "prefer perl native functions" work with threaded perls.
307 - condense Coro::Debug ps output, hint at v and w flags.
308 - (libcoro) lots of minor cleanups and portability improvements.
309
3104.914 Wed Nov 19 12:54:18 CET 2008
311 - fix a disastrous bug in the readline optimisation
312 introduced in 4.801.
313
3144.913 Sat Nov 15 07:58:28 CET 2008
315 - async_pool did free a scalar value twice
316 ("Attempt to unreference...").
317
3184.912 Thu Nov 13 18:31:23 CET 2008
319 - minor cleanups.
320 - use much larger stacks on linux and perl < 5.8.8.
321 - Coro::Debug::new_unix_server did not unlink the socket
322 when destroyed.
323
3244.911 Tue Nov 11 04:26:17 CET 2008
325 - "port" to threaded perls.
326
3274.91 Mon Nov 10 05:36:38 CET 2008
328 - the ->throw exception object no longer leaks.
329 - creating a new cctx leaked a scopestack entry (memleak).
330 - new coroutines didn't get created with a zero flags field
331 (unknown impact).
332 - calling ->throw on a not-yet-started coroutine should now work
333 instead of being ignored.
334 - ->throw is now supported on Coro::State objects.
335 - clean up cctx creation code a bit.
336 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
337
3384.9 Sat Nov 8 17:45:27 CET 2008
339 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
340 - (libcoro) no longer leak threads in the experimental pthread backend,
341 also speed it up considerably.
342 - (libcoro) do not rely on makecontext passing void *'s unscathed.
343 - fix compiletime dependencies on libcoro in the Makefile.
344 - cctx_count wasn't always updated properly.
345 - Coro::State::cctx_stacksize wasn't applied correctly.
346 - new function Coro::State::cctx_max_idle.
347 - the default max number of idle C contexts is now 4.
348 - (libcoro) try harder to get _setjmp/_longjmp.
349 - (libcoro) cleanup and extend the libcoro API to officially
350 allow the creation of empty source contexts.
351 - very experimental workaround for the totally broken netbsd platform.
352 - tried to hack around openbsd bugs.
353
3544.804 Wed Nov 5 16:36:00 CET 2008
355 - Coro::Debug::new_unix_server would not create a non-blocking listening
356 socket, sometimes causing freezes.
357 - (libcoro) fix misaligned stack points for setjmp and assembly
358 methods, which can cause crashes on x86/x86_64 with a sufficiently
359 aggressive compiler.
360 - new function: Coro::Debug::new_tcp_server.
361 - move ->throw into the Coro class because it only works on coro objects.
362
3634.803 Mon Nov 3 17:16:12 CET 2008
364 - (libcoro) use a global asm statement to become independent of gcc
365 otpimisations for CORO_ASM (thanks to pippijn for the idea).
366 - try to workaround yet another broken bsd, this time dragonfly.
367
3684.802 Thu Oct 30 10:56:12 CET 2008
369 - support -fno-omit-frame-pointer on x86 with the assembly method.
370 - tune 01_unblock.t tests a bit.
371
3724.801 Wed Oct 22 18:33:37 CEST 2008
373 - improve readline speed for very long "lines".
374 - backport to 5.8.8.
375
3764.8 Thu Oct 2 13:34:40 CEST 2008
377 - new function Coro::AIO::aio_wait.
378 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
379 - greatly speed up and reduce memory usage of Coro::AIO requests.
380 - implement some other µ-optimisations.
381
3824.749 Mon Sep 29 14:40:12 CEST 2008
383 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
384
3854.748 Sat Sep 27 14:03:19 CEST 2008
9 - implement, but do not document, PerlIO::cede(granularity). 386 - implement, but do not document, PerlIO::cede(granularity).
10 - Coro::Storable forgot to wrap Storable::pstore. 387 - Coro::Storable forgot to wrap Storable::pstore.
11 - work around the multitude of leaks and memory corruption 388 - work around the multitude of leaks and memory corruption
12 bugs in PerlIO::via by using our own C-level perliol. 389 bugs in PerlIO::via by using our own C-level perliol. As a side
390 effect, Coro::Storable is now much, much, much faster.
13 391
144.747 Tue Sep 23 01:59:41 CEST 2008 3924.747 Tue Sep 23 01:59:41 CEST 2008
15 - fix a per-cv memleak (one empty array was leaked per 393 - fix a per-cv memleak (one empty array was leaked per
16 code reference). 394 code reference).
17 - avoid a crash in coro->call|rss when the coroutine was already 395 - avoid a crash in coro->call|rss when the coroutine was already

Diff Legend

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