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

Comparing Coro/Changes (file contents):
Revision 1.408 by root, Fri Nov 14 02:29:09 2008 UTC vs.
Revision 1.561 by root, Sat May 7 14:11:10 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
85.0 7 - croak when a coro being cancelled tries to block (e.g. while
9 - ARCHITECTURAL REDESIGN, use 4.912 for the stable version. 8 executing a guard block), instead of crashing or deadlocking.
9 - implement a Coro->safe_cancel method that might fail, but
10 cancels in a "safer" way if it succeeds.
11 - implement Coro->cancel in XS for a 20% speed improvement, and to
12 be able to implement mutual cancellation.
13 - speed up context switches by a percent or two by more efficiently
14 allocating context stack entries.
15 - cancelling a coro while it itself is cancelling another coro is
16 now supported and working, instead of triggering an assertion.
17 - be a bit more crash-resistant when calling (buggy) on_destroy
18 callbacks (best effort).
19 - move on_destroy into the slf_frame, to allow extension slf
20 functions to have destructors.
21
225.372 Wed Feb 23 06:14:30 CET 2011
23 - apparently mingw doesn't provide a working gettimeofday, try to
24 work around that by relying on Time::HiRes (indirectly brought to
25 my attention by Max Maischein).
26 - fix some portability issues when Time::HiRes was used.
27
285.371 Mon Feb 21 14:36:08 CET 2011
29 - backport to windows process emulation code again.
30
315.37 Sat Feb 19 07:49:44 CET 2011
32 - add a big "Coro thread life cycle" section to "man Coro".
33 - try a tentative workaround against the breakage that 5.13 has
34 introduced without depreciation period. sigh.
35 - no longer use Time::HiRes if gettimeofday is available, which
36 saves quite a lot of memory.
37
385.36 Sun Feb 13 05:33:41 CET 2011
39 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
40 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
41 method call.
42 - undocument Coro::Timer::sleep and obsolete whole module.
43 - optimise Coro::Timer::timeout memory and cpu usage.
44 - slightly updated Coro::Intro for recent changes.
45 - do not initialise PL_dirty anymore.
46
475.25 Thu Nov 11 01:08:39 CET 2010
48 - try a different approach on netbsd - netbsd 5 finally has marginally
49 working pthreads, but still broken ucontext/sigaltstack.
50 - openbsd 4.8 finally got their act together, Coro works out of the box
51 with asm, setjmp and pthreads (no change, just informational).
52
535.24 Sat Oct 23 11:27:12 CEST 2010
54 - port to the EV 4.0 API.
55 - work around bugs in mingw32, making strawberry perl work
56 out of the box.
57 - correctly modify Coro::AIO function prototypes
58 so that they reflect the "no optional parameters" rule.
59 - "ported" libcoro to C++.
60
615.23 Mon May 17 18:50:42 CEST 2010
62 - be more resistant to ordering changes when initialising
63 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
64 Waldorf).
65 - document that perl 5.12 deliberately removed support for cloning.
66
675.22 Wed Apr 14 03:55:35 CEST 2010
68 - correctly return udnef on errors in Coro::Handle::read/write
69 (testcase by Marc Mims).
70 - convert Coro::Util into a "perl compatibility wrapper" - the functions
71 are less useful now, but are drop-in replacements for existing
72 functions, listing better alternatives in the documentation. This also
73 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
74 with Coro::Util::inet_aton.
75 - do not override $Coro::idle unconditionally in Coro.pm, as other
76 modules could have provided their own idle coro already
77 (for exmaple, Coro::AnyEvent).
78 - fix Coro::Util::gethost* functions.
79 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
80
815.21 Wed Dec 16 07:19:51 CET 2009
82 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
83 together.
84 - add some examples on how to combine other event loops with Coro in
85 Coro::AnyEvent, and how to run it (and not to block). Seems to be
86 the most common source of confusion.
87 - try to catch people naively blocking in an event callback.
88 - work around the perl filehandle bug issue in conjunction with
89 older common::sense (as indirectly pointed out by ZSystem).
90 - clarify the "not from signal handlers" section.
91
925.2 Sun Oct 4 14:54:24 CEST 2009
93 - Coro::Storable destroyed the prototypes of the functions it wrapped.
94 - export rouse_cb and rouse_wait by default now.
95 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
96 - new method $state->swap_sv.
97 - added section on "windows process emulation" to the manpage,
98 after a not-so-fruitful (nor-friendly) "discussion" with chip
99 salzenberg (discussion implies arguments, but his only arguments
100 were ad-hominems, one wonders why he started it in the first
101 place). I hope this explains it well enough for him to understand,
102 and maybe well enough for others to understand.
103 - use common::sense everywhere now.
104 - idle callbacks are no longer supported, use idle coros instead.
105 - print a thread listing when a deadlock is detected.
106
1075.17 Sat Aug 22 23:09:31 CEST 2009
108 - work around a bug in the perl debugger causing crashes
109 when running under the debugger by marking _pool_handler as nodebug.
110 - speed up Coro::async considerably.
111 - try some hacks to get netbsd to work "more often" - their broken
112 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
113 - convert Coro to AE 5.0 API.
114
1155.162 Tue Jul 28 04:04:03 CEST 2009
116 - perl 5.8.2 is now minimum requirement.
117 - skip t/19_handle.t on broken windows perls.
118
1195.161 Wed Jul 22 04:47:38 CEST 2009
120 - Coro::AnyEvent::poll could have a different prototype when EV was
121 used as backend (analyzed by Tatsuhiko Miyagawa).
122 - Coro::AnyEvent errornously initialised the event loop when loaded,
123 not on demand.
124 - try to workaround rare */t/01_unblock.t failures.
125
1265.16 Tue Jul 21 01:44:37 CEST 2009
127 - Coro::AnyEvent failed to hook into the event loop
128 when no threads had been readied between detecting
129 the event loop and actually running it.
130 - considerably speed up Coro::Select by taking avdantage
131 of AnyEvent > 4.8 and some other optimisations.
132 - implement paragraph readline mode in Coro::Handle
133 (based on patches by Zsbán Ambrus).
134 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
135 and analyzed by Yasuhiro MATSUMOTO).
136 - clarified libcoro license and copyright.
137 - someone stole my EXTRA_META!!!
138 - implement Coro::Select::patch_pp_sselect and it's brother,
139 for hardcode select overriding.
140
1415.151 Mon Jul 6 05:41:57 CEST 2009
142 - backport to windows process emulation code again (patch by
143 Yasuhiro MATSUMOTO).
144 - slightly update Coro::MakeMaker.
145
1465.15 Tue Jun 30 10:28:06 CEST 2009
147 - deprecate Coro::Socket, document how to get ipv6 support via
148 AnyEvent::Socket instead.
149 - implement signal->wait ($cb) interface, similar to semaphores.
150 - work around SvOK not supporting getmagic, so we have to getmagic
151 to test for undef :/ (reported by Matthias Waldorf).
152 - load Coro::AnyEvent in all modules using AnyEvent.
153 - work around perl corrupting our internal data structures,
154 reported by Tokuhiro Matsuno.
155 - enable per-coroutine real and cpu time gathering
156 (Coro::State::enable_times).
157
1585.14 Wed Jun 24 01:37:48 CEST 2009
159 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
160 as this seems to have been a difficult concept (poll, sleep, idle,
161 idle_upto).
162 - add Coro::AnyEvent::readable/writable functions.
163 - clarify Coro::EV/Event/AnyEvent manpages.
164 - free per-thread global scalars in the thread calling ->cancel, to
165 avoid crashes when $_, $@ etc., are magical but some of those
166 had already been freed.
167 - "unexperimentalise" the callback interface for Coro::Semaphore.
168 - speed up ready queue management/context switching by using a linked
169 list instead of an array (~5-10%).
170 - implement "watch" command in Coro::Debug shells.
171 - for fun, implement time-slicing as an example in the manpage.
172 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
173 Coro::Event, use the normal AnyEvent handling instead of dieing
174 (the same is true for Coro::Handle).
175 - properly document Coro::EV::timed_io_once.
176 - avoid unneccessary ->cancel calls in Coro::Handle.
177 - maybe make it work on mingw32 with win32 backend
178 (based on patch by Yasuhiro Matsumoto).
179
1805.132 Fri May 29 09:00:39 CEST 2009
181 - do not keep a reference to the argument itself in
182 Coro::Semaphore::guard, as it could change later.
183 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
184 should support a prepare callback.
185
1865.131 Mon Mar 16 23:20:37 CET 2009
187 - implement and document Coro->suspend, Coro->resume.
188 - fix Coro::Select implementation to not (often) close
189 the passed file descriptors (testcase provided by pippijn).
190
1915.13 Mon Dec 15 21:51:42 CET 2008
192 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
193 - don't set diehook to C<undef> but instead to NULL, to avoid
194 spurious warnings.
195 - fix a lot of bugs in Coro::SemaphoreSet.
196 - Coro::SemaphoreSet will less often create a semaphore needlessly.
197 - add Coro::SemaphoreSet::count and wait methods.
198 - take advantage of the new Guard module.
199 - deprecate Coro::guard.
200 - try to fix the dreaded 01_unblock tests once more. I hate it when
201 testsuites need more fixing than the code they are supposed to test.
202 - croak in more cases when a required callback isn't resolvable.
203 - fix some minor issues in Coro::State->call/eval.
204 - use current coroutine context instead of a temporary one
205 when temporarily switching to another coroutine.
206 - do not call C-level on_destroy handlers during global destruction,
207 to avoid needless segfaults.
208
2095.12 Sun Dec 7 13:30:38 CET 2008
210 - add default config for MirOS, which seems to be a bug-to-bug
211 compatible fork of openbsd ("world domination by releasing
212 openbsd cvs before the openbsd folks do it" or so :).
213 - free_padlist did destroy the names pad, not good, but didn't
214 seem to bother perl - this could fix issues such as eval ""
215 inside a function called from multiple coroutines.
216 - use a different method to detect destruction time.
217 - be more careful when freeing padlists just before global
218 destruction.
219 - fixed and expanded the call/cc example.
220 - renamed _terminate to _coro_run.
221 - new method Coro::Channel->shutdown.
222 - try pthreads on openbsd <4.4 (broken sigaltstack, will
223 pthreads fare better?).
224 - be less picky about destroying "a" running coroutine.
225
2265.11 Tue Nov 25 21:49:05 CET 2008
227 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
228 working code with failed assertions, introducing more bugs than
229 it fixes, requiring elaborate workarounds :(
230
2315.1 Mon Nov 24 08:54:59 CET 2008
232 - wrote a small introductory tutorial - Coro::Intro.
233 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
234 - Coro::Select did errornously dup the file descriptors
235 and didn't work with all AnyEvent backends.
236 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
237 LWP data transfers.
238 - disassociate c contexts from coro objects - this is agruably more
239 correct, but mostly allows sharing of cctxs between coro and state
240 objects, for added memory savings and speed increases.
241 - bumped $Coro::POOL_RSS up to 32kb by default.
242 - no longer set the optype to OP_CUSTOM, as B::* understandably
243 doesn't like this very much (and we *are* a type of entersub).
244 - implement state cloning, just to prove that call/cc can be done.
245 - automatically load Coro::AnyEvent in Coro::Handle.
246 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
247 do not have this method (and don't need it either).
248 - speed up generic anyevent methods in Coro::Handle by using rouse
249 callbacks.
250 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
251 others. It also makes the debugger happier, as you can trace
252 through the idle threads now.
253 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
254 - eg/event was pretty broken.
255 - better 5.8.6 compatibility.
256
2575.0 Thu Nov 20 10:35:05 CET 2008
258 - NEW ARCHITECTURE: use the latest 4.x version if you experience
259 stability issues.
260 - bump API version to 7 - all dependents must be recompiled.
261 - removed timed_* functions - they were not being used anyways
262 and should be replaced by a more generic mechanism -
263 and were annoying to support anyways :)
264 - removed SemaphoreSet's waiter method - use sem method instead.
265 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
266 - async_pool did free a scalar value twice
267 ("Attempt to unreference...").
268 - fix a longstanding bug where calling terminate on a coro that
269 was waiting for a semaphore that was just becoming available
270 would cause a deadlock (semaphore would get into a state where
271 it was available but waiters were still blocked).
272 - calling throw on a coroutine that is waiting for a semaphore will
273 no longer make it acquire the semaphore (and thus leak a count).
274 - perl's process emulation is now not even theoretically supported
275 anymore.
276 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
277 conversion of callback-style to blocking-style.
278 - new methods $coro->schedule_to and ->cede_to, to specifically
279 schedule or cede to a specific coroutine.
280 - new function Coro::Semaphore::wait.
281 - use named constants in Coro::Channel (Richard Hundt).
10 - directly patch the entersub opcode calling ssl functions (cede, 282 - directly patch the entersub opcode calling SLF functions (cede,
11 transfer and so on). this does speed up context switching by 283 transfer and so on). this does speed up context switching, but
12 up to 10%. More importanly, it frees us from the hardcoded 284 more importanly, it frees us from the hardcoded behaviour of
13 behaviour of entersub, so we might actually be able to return 285 entersub, so we might actually be able to return something from
14 something from those functions. 286 those functions and atcually create new ones.
287 - take advantage of __builtin_frame_address on gcc.
288 - expose THX in coroapi (not sure whether this was a wise decision,
289 as "threaded" perls are running at half speed anyways).
290 - implement execute_slf (schedule-like-function) interface that makes
291 it possible to implement schedule-like-functions in XS.
292 - use new SLF interface to massively speed up Coro::EV by roughly a
293 factor of two.
294 - used new SLF interface to massively speed up Coro::Semaphore by a
295 factor of three.
296 - used new SLF interface to speed up Coro::AIO by roughly a factor of
297 four and reduce its memory usage considerably.
298 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
299 for a nice speedup and vastly more correct behaviour. Also implement
300 a new method "sem" to get at the underlying semaphore object.
301 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
302 (in comparison) 20-40% speedup.
303 - used new SLF interface to reimplement Coro::Signal gaining
304 some unknown (because I was too lazy), but certain, speedup, and also
305 making signals reliable for the first time.
306 - used new SLF interface and other optimisations to speed up async_pool
307 by a factor of two. It also doesn't rely on perl's exception mechanism
308 to exit anymore. The overhead for terminating an async_pool, coro over
309 a normal async is now very small.
310 - sped up coroutine creation/destruction by 40%.
311 - forgot to include Coro/libcoro/README in the dist for all these years.
312 - work around a freebsd pthreads bug (manual testcancel is required as
313 pthread_cond_wait isn't a cancellation point on freebsd).
314 - use new rouse functions to speed up and simplify Coro::BDB.
315 - make "prefer perl native functions" work with threaded perls.
316 - condense Coro::Debug ps output, hint at v and w flags.
317 - (libcoro) lots of minor cleanups and portability improvements.
318
3194.914 Wed Nov 19 12:54:18 CET 2008
320 - fix a disastrous bug in the readline optimisation
321 introduced in 4.801.
322
3234.913 Sat Nov 15 07:58:28 CET 2008
324 - async_pool did free a scalar value twice
325 ("Attempt to unreference...").
15 326
164.912 Thu Nov 13 18:31:23 CET 2008 3274.912 Thu Nov 13 18:31:23 CET 2008
17 - minor cleanups. 328 - minor cleanups.
18 - use much larger stacks on linux and perl < 5.8.8. 329 - use much larger stacks on linux and perl < 5.8.8.
19 - Coro::Debug::new_unix_server did not unlink the socket 330 - Coro::Debug::new_unix_server did not unlink the socket

Diff Legend

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