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

Comparing Coro/Changes (file contents):
Revision 1.429 by root, Tue Nov 18 09:59:27 2008 UTC vs.
Revision 1.558 by root, Fri May 6 18:01:13 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
8TODO: testcancel 7 - croak when a coro being cancelled tries to block (e.g. while executing
9TODO: guard as SLF - to avoid deadlocks. 8 a guard block).
10 - avoid deadlock in Coro::Channel if maxsize == 1 (Richard Hundt). 9 - implement Coro->cancel in XS for a 20% speed improvement, and to
10 be able to implement mutual cancellation.
11 - cancelling a coro while it itself is cancelling another coro is
12 now supported and working, instead of triggering an assertion.
13 - be a bit more crash-resistant when calling (buggy) on_destroy
14 callbacks (best effort).
11 15
125.0 165.372 Wed Feb 23 06:14:30 CET 2011
17 - apparently mingw doesn't provide a working gettimeofday, try to
18 work around that by relying on Time::HiRes (indirectly brought to
19 my attention by Max Maischein).
20 - fix some portability issues when Time::HiRes was used.
21
225.371 Mon Feb 21 14:36:08 CET 2011
23 - backport to windows process emulation code again.
24
255.37 Sat Feb 19 07:49:44 CET 2011
26 - add a big "Coro thread life cycle" section to "man Coro".
27 - try a tentative workaround against the breakage that 5.13 has
28 introduced without depreciation period. sigh.
29 - no longer use Time::HiRes if gettimeofday is available, which
30 saves quite a lot of memory.
31
325.36 Sun Feb 13 05:33:41 CET 2011
33 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
34 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
35 method call.
36 - undocument Coro::Timer::sleep and obsolete whole module.
37 - optimise Coro::Timer::timeout memory and cpu usage.
38 - slightly updated Coro::Intro for recent changes.
39 - do not initialise PL_dirty anymore.
40
415.25 Thu Nov 11 01:08:39 CET 2010
42 - try a different approach on netbsd - netbsd 5 finally has marginally
43 working pthreads, but still broken ucontext/sigaltstack.
44 - openbsd 4.8 finally got their act together, Coro works out of the box
45 with asm, setjmp and pthreads (no change, just informational).
46
475.24 Sat Oct 23 11:27:12 CEST 2010
48 - port to the EV 4.0 API.
49 - work around bugs in mingw32, making strawberry perl work
50 out of the box.
51 - correctly modify Coro::AIO function prototypes
52 so that they reflect the "no optional parameters" rule.
53 - "ported" libcoro to C++.
54
555.23 Mon May 17 18:50:42 CEST 2010
56 - be more resistant to ordering changes when initialising
57 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
58 Waldorf).
59 - document that perl 5.12 deliberately removed support for cloning.
60
615.22 Wed Apr 14 03:55:35 CEST 2010
62 - correctly return udnef on errors in Coro::Handle::read/write
63 (testcase by Marc Mims).
64 - convert Coro::Util into a "perl compatibility wrapper" - the functions
65 are less useful now, but are drop-in replacements for existing
66 functions, listing better alternatives in the documentation. This also
67 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
68 with Coro::Util::inet_aton.
69 - do not override $Coro::idle unconditionally in Coro.pm, as other
70 modules could have provided their own idle coro already
71 (for exmaple, Coro::AnyEvent).
72 - fix Coro::Util::gethost* functions.
73 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
74
755.21 Wed Dec 16 07:19:51 CET 2009
76 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
77 together.
78 - add some examples on how to combine other event loops with Coro in
79 Coro::AnyEvent, and how to run it (and not to block). Seems to be
80 the most common source of confusion.
81 - try to catch people naively blocking in an event callback.
82 - work around the perl filehandle bug issue in conjunction with
83 older common::sense (as indirectly pointed out by ZSystem).
84 - clarify the "not from signal handlers" section.
85
865.2 Sun Oct 4 14:54:24 CEST 2009
87 - Coro::Storable destroyed the prototypes of the functions it wrapped.
88 - export rouse_cb and rouse_wait by default now.
89 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
90 - new method $state->swap_sv.
91 - added section on "windows process emulation" to the manpage,
92 after a not-so-fruitful (nor-friendly) "discussion" with chip
93 salzenberg (discussion implies arguments, but his only arguments
94 were ad-hominems, one wonders why he started it in the first
95 place). I hope this explains it well enough for him to understand,
96 and maybe well enough for others to understand.
97 - use common::sense everywhere now.
98 - idle callbacks are no longer supported, use idle coros instead.
99 - print a thread listing when a deadlock is detected.
100
1015.17 Sat Aug 22 23:09:31 CEST 2009
102 - work around a bug in the perl debugger causing crashes
103 when running under the debugger by marking _pool_handler as nodebug.
104 - speed up Coro::async considerably.
105 - try some hacks to get netbsd to work "more often" - their broken
106 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
107 - convert Coro to AE 5.0 API.
108
1095.162 Tue Jul 28 04:04:03 CEST 2009
110 - perl 5.8.2 is now minimum requirement.
111 - skip t/19_handle.t on broken windows perls.
112
1135.161 Wed Jul 22 04:47:38 CEST 2009
114 - Coro::AnyEvent::poll could have a different prototype when EV was
115 used as backend (analyzed by Tatsuhiko Miyagawa).
116 - Coro::AnyEvent errornously initialised the event loop when loaded,
117 not on demand.
118 - try to workaround rare */t/01_unblock.t failures.
119
1205.16 Tue Jul 21 01:44:37 CEST 2009
121 - Coro::AnyEvent failed to hook into the event loop
122 when no threads had been readied between detecting
123 the event loop and actually running it.
124 - considerably speed up Coro::Select by taking avdantage
125 of AnyEvent > 4.8 and some other optimisations.
126 - implement paragraph readline mode in Coro::Handle
127 (based on patches by Zsbán Ambrus).
128 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
129 and analyzed by Yasuhiro MATSUMOTO).
130 - clarified libcoro license and copyright.
131 - someone stole my EXTRA_META!!!
132 - implement Coro::Select::patch_pp_sselect and it's brother,
133 for hardcode select overriding.
134
1355.151 Mon Jul 6 05:41:57 CEST 2009
136 - backport to windows process emulation code again (patch by
137 Yasuhiro MATSUMOTO).
138 - slightly update Coro::MakeMaker.
139
1405.15 Tue Jun 30 10:28:06 CEST 2009
141 - deprecate Coro::Socket, document how to get ipv6 support via
142 AnyEvent::Socket instead.
143 - implement signal->wait ($cb) interface, similar to semaphores.
144 - work around SvOK not supporting getmagic, so we have to getmagic
145 to test for undef :/ (reported by Matthias Waldorf).
146 - load Coro::AnyEvent in all modules using AnyEvent.
147 - work around perl corrupting our internal data structures,
148 reported by Tokuhiro Matsuno.
149 - enable per-coroutine real and cpu time gathering
150 (Coro::State::enable_times).
151
1525.14 Wed Jun 24 01:37:48 CEST 2009
153 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
154 as this seems to have been a difficult concept (poll, sleep, idle,
155 idle_upto).
156 - add Coro::AnyEvent::readable/writable functions.
157 - clarify Coro::EV/Event/AnyEvent manpages.
158 - free per-thread global scalars in the thread calling ->cancel, to
159 avoid crashes when $_, $@ etc., are magical but some of those
160 had already been freed.
161 - "unexperimentalise" the callback interface for Coro::Semaphore.
162 - speed up ready queue management/context switching by using a linked
163 list instead of an array (~5-10%).
164 - implement "watch" command in Coro::Debug shells.
165 - for fun, implement time-slicing as an example in the manpage.
166 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
167 Coro::Event, use the normal AnyEvent handling instead of dieing
168 (the same is true for Coro::Handle).
169 - properly document Coro::EV::timed_io_once.
170 - avoid unneccessary ->cancel calls in Coro::Handle.
171 - maybe make it work on mingw32 with win32 backend
172 (based on patch by Yasuhiro Matsumoto).
173
1745.132 Fri May 29 09:00:39 CEST 2009
175 - do not keep a reference to the argument itself in
176 Coro::Semaphore::guard, as it could change later.
177 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
178 should support a prepare callback.
179
1805.131 Mon Mar 16 23:20:37 CET 2009
181 - implement and document Coro->suspend, Coro->resume.
182 - fix Coro::Select implementation to not (often) close
183 the passed file descriptors (testcase provided by pippijn).
184
1855.13 Mon Dec 15 21:51:42 CET 2008
186 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
187 - don't set diehook to C<undef> but instead to NULL, to avoid
188 spurious warnings.
189 - fix a lot of bugs in Coro::SemaphoreSet.
190 - Coro::SemaphoreSet will less often create a semaphore needlessly.
191 - add Coro::SemaphoreSet::count and wait methods.
192 - take advantage of the new Guard module.
193 - deprecate Coro::guard.
194 - try to fix the dreaded 01_unblock tests once more. I hate it when
195 testsuites need more fixing than the code they are supposed to test.
196 - croak in more cases when a required callback isn't resolvable.
197 - fix some minor issues in Coro::State->call/eval.
198 - use current coroutine context instead of a temporary one
199 when temporarily switching to another coroutine.
200 - do not call C-level on_destroy handlers during global destruction,
201 to avoid needless segfaults.
202
2035.12 Sun Dec 7 13:30:38 CET 2008
204 - add default config for MirOS, which seems to be a bug-to-bug
205 compatible fork of openbsd ("world domination by releasing
206 openbsd cvs before the openbsd folks do it" or so :).
207 - free_padlist did destroy the names pad, not good, but didn't
208 seem to bother perl - this could fix issues such as eval ""
209 inside a function called from multiple coroutines.
210 - use a different method to detect destruction time.
211 - be more careful when freeing padlists just before global
212 destruction.
213 - fixed and expanded the call/cc example.
214 - renamed _terminate to _coro_run.
215 - new method Coro::Channel->shutdown.
216 - try pthreads on openbsd <4.4 (broken sigaltstack, will
217 pthreads fare better?).
218 - be less picky about destroying "a" running coroutine.
219
2205.11 Tue Nov 25 21:49:05 CET 2008
221 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
222 working code with failed assertions, introducing more bugs than
223 it fixes, requiring elaborate workarounds :(
224
2255.1 Mon Nov 24 08:54:59 CET 2008
226 - wrote a small introductory tutorial - Coro::Intro.
227 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
228 - Coro::Select did errornously dup the file descriptors
229 and didn't work with all AnyEvent backends.
230 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
231 LWP data transfers.
232 - disassociate c contexts from coro objects - this is agruably more
233 correct, but mostly allows sharing of cctxs between coro and state
234 objects, for added memory savings and speed increases.
235 - bumped $Coro::POOL_RSS up to 32kb by default.
236 - no longer set the optype to OP_CUSTOM, as B::* understandably
237 doesn't like this very much (and we *are* a type of entersub).
238 - implement state cloning, just to prove that call/cc can be done.
239 - automatically load Coro::AnyEvent in Coro::Handle.
240 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
241 do not have this method (and don't need it either).
242 - speed up generic anyevent methods in Coro::Handle by using rouse
243 callbacks.
244 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
245 others. It also makes the debugger happier, as you can trace
246 through the idle threads now.
247 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
248 - eg/event was pretty broken.
249 - better 5.8.6 compatibility.
250
2515.0 Thu Nov 20 10:35:05 CET 2008
13 - NEW ARCHITECTURE: use the latest 4.x version if you experience 252 - NEW ARCHITECTURE: use the latest 4.x version if you experience
14 stability issues. 253 stability issues.
15 - bump API version to 7 - all dependents must be recompiled. 254 - bump API version to 7 - all dependents must be recompiled.
16 - removed timed_down and timed_guard functions - they were not being 255 - removed timed_* functions - they were not being used anyways
17 used anyways and should be replaced by a more generic mechanism - 256 and should be replaced by a more generic mechanism -
18 and were annoying to support anyways :) 257 and were annoying to support anyways :)
19 - removed SemaphoreSet's waiter method - use sem instead. 258 - removed SemaphoreSet's waiter method - use sem method instead.
20 - Coro::Semaphore->adjust didn't correctly wake up enough waiters. 259 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
21 - async_pool did free a scalar value twice 260 - async_pool did free a scalar value twice
22 ("Attempt to unreference..."). 261 ("Attempt to unreference...").
23 - fix a longstanding bug where calling terminate on a coro that 262 - fix a longstanding bug where calling terminate on a coro that
24 was waiting for a semaphore that was just becoming available 263 was waiting for a semaphore that was just becoming available
25 would cause a deadlock (semaphore would get into a state where 264 would cause a deadlock (semaphore would get into a state where
26 it was available but waiters were waiting(. 265 it was available but waiters were still blocked).
266 - calling throw on a coroutine that is waiting for a semaphore will
267 no longer make it acquire the semaphore (and thus leak a count).
27 - perl's process emulation is now not even theoretically supported 268 - perl's process emulation is now not even theoretically supported
28 anymore. 269 anymore.
270 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
271 conversion of callback-style to blocking-style.
272 - new methods $coro->schedule_to and ->cede_to, to specifically
273 schedule or cede to a specific coroutine.
274 - new function Coro::Semaphore::wait.
29 - use named constants in Coro::Channel (Richard Hundt). 275 - use named constants in Coro::Channel (Richard Hundt).
30 - directly patch the entersub opcode calling SLF functions (cede, 276 - directly patch the entersub opcode calling SLF functions (cede,
31 transfer and so on). this does speed up context switching, but 277 transfer and so on). this does speed up context switching, but
32 more importanly, it frees us from the hardcoded behaviour of 278 more importanly, it frees us from the hardcoded behaviour of
33 entersub, so we might actually be able to return something from 279 entersub, so we might actually be able to return something from
… …
44 - used new SLF interface to speed up Coro::AIO by roughly a factor of 290 - used new SLF interface to speed up Coro::AIO by roughly a factor of
45 four and reduce its memory usage considerably. 291 four and reduce its memory usage considerably.
46 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore, 292 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
47 for a nice speedup and vastly more correct behaviour. Also implement 293 for a nice speedup and vastly more correct behaviour. Also implement
48 a new method "sem" to get at the underlying semaphore object. 294 a new method "sem" to get at the underlying semaphore object.
295 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
296 (in comparison) 20-40% speedup.
297 - used new SLF interface to reimplement Coro::Signal gaining
298 some unknown (because I was too lazy), but certain, speedup, and also
299 making signals reliable for the first time.
300 - used new SLF interface and other optimisations to speed up async_pool
301 by a factor of two. It also doesn't rely on perl's exception mechanism
302 to exit anymore. The overhead for terminating an async_pool, coro over
303 a normal async is now very small.
304 - sped up coroutine creation/destruction by 40%.
49 - forgot to include Coro/libcoro/README in the dist for all these years. 305 - forgot to include Coro/libcoro/README in the dist for all these years.
50 - work around a freebsd pthreads bug (manual testcancel is required as 306 - work around a freebsd pthreads bug (manual testcancel is required as
51 pthread_cond_wait isn't a cancellation point on freebsd). 307 pthread_cond_wait isn't a cancellation point on freebsd).
308 - use new rouse functions to speed up and simplify Coro::BDB.
309 - make "prefer perl native functions" work with threaded perls.
310 - condense Coro::Debug ps output, hint at v and w flags.
311 - (libcoro) lots of minor cleanups and portability improvements.
312
3134.914 Wed Nov 19 12:54:18 CET 2008
314 - fix a disastrous bug in the readline optimisation
315 introduced in 4.801.
52 316
534.913 Sat Nov 15 07:58:28 CET 2008 3174.913 Sat Nov 15 07:58:28 CET 2008
54 - async_pool did free a scalar value twice 318 - async_pool did free a scalar value twice
55 ("Attempt to unreference..."). 319 ("Attempt to unreference...").
56 320

Diff Legend

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