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

Comparing Coro/Changes (file contents):
Revision 1.395 by root, Sat Nov 8 13:32:18 2008 UTC vs.
Revision 1.540 by root, Tue Jul 27 14:42:34 2010 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: unready_all
5 actor model (which is cool in a lot of important aspects (failures!), 5TODO: myhttpd header parsing
6 but very lacking in others (higher level ipc)). 6TODO: channel->maxsize(newsize)?
7TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
7 8
85.0 9 - "ported" libcoro to C++.
10
115.23 Mon May 17 18:50:42 CEST 2010
12 - be more resistant to ordering changes when initialising
13 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
14 Waldorf).
15 - document that perl 5.12 deliberately removed support for cloning.
16
175.22 Wed Apr 14 03:55:35 CEST 2010
18 - correctly return udnef on errors in Coro::Handle::read/write
19 (testcase by Marc Mims).
20 - convert Coro::Util into a "perl compatibility wrapper" - the functions
21 are less useful now, but are drop-in replacements for existing
22 functions, listing better alternatives in the documentation. This also
23 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
24 with Coro::Util::inet_aton.
25 - do not override $Coro::idle unconditionally in Coro.pm, as other
26 modules could have provided their own idle coro already
27 (for exmaple, Coro::AnyEvent).
28 - fix Coro::Util::gethost* functions.
29 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
30
315.21 Wed Dec 16 07:19:51 CET 2009
32 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
33 together.
34 - add some examples on how to combine other event loops with Coro in
35 Coro::AnyEvent, and how to run it (and not to block). Seems to be
36 the most common source of confusion.
37 - try to catch people naively blocking in an event callback.
38 - work around the perl filehandle bug issue in conjunction with
39 older common::sense (as indirectly pointed out by ZSystem).
40 - clarify the "not from signal handlers" section.
41
425.2 Sun Oct 4 14:54:24 CEST 2009
43 - Coro::Storable destroyed the prototypes of the functions it wrapped.
44 - export rouse_cb and rouse_wait by default now.
45 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
46 - new method $state->swap_sv.
47 - added section on "windows process emulation" to the manpage,
48 after a not-so-fruitful (nor-friendly) "discussion" with chip
49 salzenberg (discussion implies arguments, but his only arguments
50 were ad-hominems, one wonders why he started it in the first
51 place). I hope this explains it well enough for him to understand,
52 and maybe well enough for others to understand.
53 - use common::sense everywhere now.
54 - idle callbacks are no longer supported, use idle coros instead.
55 - print a thread listing when a deadlock is detected.
56
575.17 Sat Aug 22 23:09:31 CEST 2009
58 - work around a bug in the perl debugger causing crashes
59 when running under the debugger by marking _pool_handler as nodebug.
60 - speed up Coro::async considerably.
61 - try some hacks to get netbsd to work "more often" - their broken
62 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
63 - convert Coro to AE 5.0 API.
64
655.162 Tue Jul 28 04:04:03 CEST 2009
66 - perl 5.8.2 is now minimum requirement.
67 - skip t/19_handle.t on broken windows perls.
68
695.161 Wed Jul 22 04:47:38 CEST 2009
70 - Coro::AnyEvent::poll could have a different prototype when EV was
71 used as backend (analyzed by Tatsuhiko Miyagawa).
72 - Coro::AnyEvent errornously initialised the event loop when loaded,
73 not on demand.
74 - try to workaround rare */t/01_unblock.t failures.
75
765.16 Tue Jul 21 01:44:37 CEST 2009
77 - Coro::AnyEvent failed to hook into the event loop
78 when no threads had been readied between detecting
79 the event loop and actually running it.
80 - considerably speed up Coro::Select by taking avdantage
81 of AnyEvent > 4.8 and some other optimisations.
82 - implement paragraph readline mode in Coro::Handle
83 (based on patches by Zsbán Ambrus).
84 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
85 and analyzed by Yasuhiro MATSUMOTO).
86 - clarified libcoro license and copyright.
87 - someone stole my EXTRA_META!!!
88 - implement Coro::Select::patch_pp_sselect and it's brother,
89 for hardcode select overriding.
90
915.151 Mon Jul 6 05:41:57 CEST 2009
92 - backport to windows process emulation code again (patch by
93 Yasuhiro MATSUMOTO).
94 - slightly update Coro::MakeMaker.
95
965.15 Tue Jun 30 10:28:06 CEST 2009
97 - deprecate Coro::Socket, document how to get ipv6 support via
98 AnyEvent::Socket instead.
99 - implement signal->wait ($cb) interface, similar to semaphores.
100 - work around SvOK not supporting getmagic, so we have to getmagic
101 to test for undef :/ (reported by Matthias Waldorf).
102 - load Coro::AnyEvent in all modules using AnyEvent.
103 - work around perl corrupting our internal data structures,
104 reported by Tokuhiro Matsuno.
105 - enable per-coroutine real and cpu time gathering
106 (Coro::State::enable_times).
107
1085.14 Wed Jun 24 01:37:48 CEST 2009
109 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
110 as this seems to have been a difficult concept (poll, sleep, idle,
111 idle_upto).
112 - add Coro::AnyEvent::readable/writable functions.
113 - clarify Coro::EV/Event/AnyEvent manpages.
114 - free per-thread global scalars in the thread calling ->cancel, to
115 avoid crashes when $_, $@ etc., are magical but some of those
116 had already been freed.
117 - "unexperimentalise" the callback interface for Coro::Semaphore.
118 - speed up ready queue management/context switching by using a linked
119 list instead of an array (~5-10%).
120 - implement "watch" command in Coro::Debug shells.
121 - for fun, implement time-slicing as an example in the manpage.
122 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
123 Coro::Event, use the normal AnyEvent handling instead of dieing
124 (the same is true for Coro::Handle).
125 - properly document Coro::EV::timed_io_once.
126 - avoid unneccessary ->cancel calls in Coro::Handle.
127 - maybe make it work on mingw32 with win32 backend
128 (based on patch by Yasuhiro Matsumoto).
129
1305.132 Fri May 29 09:00:39 CEST 2009
131 - do not keep a reference to the argument itself in
132 Coro::Semaphore::guard, as it could change later.
133 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
134 should support a prepare callback.
135
1365.131 Mon Mar 16 23:20:37 CET 2009
137 - implement and document Coro->suspend, Coro->resume.
138 - fix Coro::Select implementation to not (often) close
139 the passed file descriptors (testcase provided by pippijn).
140
1415.13 Mon Dec 15 21:51:42 CET 2008
142 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
143 - don't set diehook to C<undef> but instead to NULL, to avoid
144 spurious warnings.
145 - fix a lot of bugs in Coro::SemaphoreSet.
146 - Coro::SemaphoreSet will less often create a semaphore needlessly.
147 - add Coro::SemaphoreSet::count and wait methods.
148 - take advantage of the new Guard module.
149 - deprecate Coro::guard.
150 - try to fix the dreaded 01_unblock tests once more. I hate it when
151 testsuites need more fixing than the code they are supposed to test.
152 - croak in more cases when a required callback isn't resolvable.
153 - fix some minor issues in Coro::State->call/eval.
154 - use current coroutine context instead of a temporary one
155 when temporarily switching to another coroutine.
156 - do not call C-level on_destroy handlers during global destruction,
157 to avoid needless segfaults.
158
1595.12 Sun Dec 7 13:30:38 CET 2008
160 - add default config for MirOS, which seems to be a bug-to-bug
161 compatible fork of openbsd ("world domination by releasing
162 openbsd cvs before the openbsd folks do it" or so :).
163 - free_padlist did destroy the names pad, not good, but didn't
164 seem to bother perl - this could fix issues such as eval ""
165 inside a function called from multiple coroutines.
166 - use a different method to detect destruction time.
167 - be more careful when freeing padlists just before global
168 destruction.
169 - fixed and expanded the call/cc example.
170 - renamed _terminate to _coro_run.
171 - new method Coro::Channel->shutdown.
172 - try pthreads on openbsd <4.4 (broken sigaltstack, will
173 pthreads fare better?).
174 - be less picky about destroying "a" running coroutine.
175
1765.11 Tue Nov 25 21:49:05 CET 2008
177 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
178 working code with failed assertions, introducing more bugs than
179 it fixes, requiring elaborate workarounds :(
180
1815.1 Mon Nov 24 08:54:59 CET 2008
182 - wrote a small introductory tutorial - Coro::Intro.
183 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
184 - Coro::Select did errornously dup the file descriptors
185 and didn't work with all AnyEvent backends.
186 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
187 LWP data transfers.
188 - disassociate c contexts from coro objects - this is agruably more
189 correct, but mostly allows sharing of cctxs between coro and state
190 objects, for added memory savings and speed increases.
191 - bumped $Coro::POOL_RSS up to 32kb by default.
192 - no longer set the optype to OP_CUSTOM, as B::* understandably
193 doesn't like this very much (and we *are* a type of entersub).
194 - implement state cloning, just to prove that call/cc can be done.
195 - automatically load Coro::AnyEvent in Coro::Handle.
196 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
197 do not have this method (and don't need it either).
198 - speed up generic anyevent methods in Coro::Handle by using rouse
199 callbacks.
200 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
201 others. It also makes the debugger happier, as you can trace
202 through the idle threads now.
203 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
204 - eg/event was pretty broken.
205 - better 5.8.6 compatibility.
206
2075.0 Thu Nov 20 10:35:05 CET 2008
208 - NEW ARCHITECTURE: use the latest 4.x version if you experience
209 stability issues.
210 - bump API version to 7 - all dependents must be recompiled.
211 - removed timed_* functions - they were not being used anyways
212 and should be replaced by a more generic mechanism -
213 and were annoying to support anyways :)
214 - removed SemaphoreSet's waiter method - use sem method instead.
215 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
216 - async_pool did free a scalar value twice
217 ("Attempt to unreference...").
218 - fix a longstanding bug where calling terminate on a coro that
219 was waiting for a semaphore that was just becoming available
220 would cause a deadlock (semaphore would get into a state where
221 it was available but waiters were still blocked).
222 - calling throw on a coroutine that is waiting for a semaphore will
223 no longer make it acquire the semaphore (and thus leak a count).
224 - perl's process emulation is now not even theoretically supported
225 anymore.
226 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
227 conversion of callback-style to blocking-style.
228 - new methods $coro->schedule_to and ->cede_to, to specifically
229 schedule or cede to a specific coroutine.
230 - new function Coro::Semaphore::wait.
231 - use named constants in Coro::Channel (Richard Hundt).
232 - directly patch the entersub opcode calling SLF functions (cede,
233 transfer and so on). this does speed up context switching, but
234 more importanly, it frees us from the hardcoded behaviour of
235 entersub, so we might actually be able to return something from
236 those functions and atcually create new ones.
237 - take advantage of __builtin_frame_address on gcc.
238 - expose THX in coroapi (not sure whether this was a wise decision,
239 as "threaded" perls are running at half speed anyways).
240 - implement execute_slf (schedule-like-function) interface that makes
241 it possible to implement schedule-like-functions in XS.
242 - use new SLF interface to massively speed up Coro::EV by roughly a
243 factor of two.
244 - used new SLF interface to massively speed up Coro::Semaphore by a
245 factor of three.
246 - used new SLF interface to speed up Coro::AIO by roughly a factor of
247 four and reduce its memory usage considerably.
248 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
249 for a nice speedup and vastly more correct behaviour. Also implement
250 a new method "sem" to get at the underlying semaphore object.
251 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
252 (in comparison) 20-40% speedup.
253 - used new SLF interface to reimplement Coro::Signal gaining
254 some unknown (because I was too lazy), but certain, speedup, and also
255 making signals reliable for the first time.
256 - used new SLF interface and other optimisations to speed up async_pool
257 by a factor of two. It also doesn't rely on perl's exception mechanism
258 to exit anymore. The overhead for terminating an async_pool, coro over
259 a normal async is now very small.
260 - sped up coroutine creation/destruction by 40%.
261 - forgot to include Coro/libcoro/README in the dist for all these years.
262 - work around a freebsd pthreads bug (manual testcancel is required as
263 pthread_cond_wait isn't a cancellation point on freebsd).
264 - use new rouse functions to speed up and simplify Coro::BDB.
265 - make "prefer perl native functions" work with threaded perls.
266 - condense Coro::Debug ps output, hint at v and w flags.
267 - (libcoro) lots of minor cleanups and portability improvements.
268
2694.914 Wed Nov 19 12:54:18 CET 2008
270 - fix a disastrous bug in the readline optimisation
271 introduced in 4.801.
272
2734.913 Sat Nov 15 07:58:28 CET 2008
274 - async_pool did free a scalar value twice
275 ("Attempt to unreference...").
276
2774.912 Thu Nov 13 18:31:23 CET 2008
278 - minor cleanups.
279 - use much larger stacks on linux and perl < 5.8.8.
280 - Coro::Debug::new_unix_server did not unlink the socket
281 when destroyed.
282
2834.911 Tue Nov 11 04:26:17 CET 2008
284 - "port" to threaded perls.
285
2864.91 Mon Nov 10 05:36:38 CET 2008
287 - the ->throw exception object no longer leaks.
288 - creating a new cctx leaked a scopestack entry (memleak).
289 - new coroutines didn't get created with a zero flags field
290 (unknown impact).
291 - calling ->throw on a not-yet-started coroutine should now work
292 instead of being ignored.
293 - ->throw is now supported on Coro::State objects.
294 - clean up cctx creation code a bit.
295 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
296
2974.9 Sat Nov 8 17:45:27 CET 2008
9 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 298 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
10 - (libcoro) no longer leak threads in the experimental pthread backend, 299 - (libcoro) no longer leak threads in the experimental pthread backend,
11 also speed it up considerably. 300 also speed it up considerably.
12 - (libcoro) do not rely on makecontext passing void *'s unscathed. 301 - (libcoro) do not rely on makecontext passing void *'s unscathed.
13 - fix compiletime dependencies on libcoro in the Makefile. 302 - fix compiletime dependencies on libcoro in the Makefile.
… …
17 - the default max number of idle C contexts is now 4. 306 - the default max number of idle C contexts is now 4.
18 - (libcoro) try harder to get _setjmp/_longjmp. 307 - (libcoro) try harder to get _setjmp/_longjmp.
19 - (libcoro) cleanup and extend the libcoro API to officially 308 - (libcoro) cleanup and extend the libcoro API to officially
20 allow the creation of empty source contexts. 309 allow the creation of empty source contexts.
21 - very experimental workaround for the totally broken netbsd platform. 310 - very experimental workaround for the totally broken netbsd platform.
311 - tried to hack around openbsd bugs.
22 312
234.804 Wed Nov 5 16:36:00 CET 2008 3134.804 Wed Nov 5 16:36:00 CET 2008
24 - Coro::Debug::new_unix_server would not create a non-blocking listening 314 - Coro::Debug::new_unix_server would not create a non-blocking listening
25 socket, sometimes causing freezes. 315 socket, sometimes causing freezes.
26 - (libcoro) fix misaligned stack points for setjmp and assembly 316 - (libcoro) fix misaligned stack points for setjmp and assembly

Diff Legend

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