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

Comparing Coro/Changes (file contents):
Revision 1.412 by root, Fri Nov 14 20:35:49 2008 UTC vs.
Revision 1.559 by root, Fri May 6 21:15:17 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)).
7TODO: MAP_STACK
8 6
9// tobench 1.500 7 - croak when a coro being cancelled tries to block (e.g. while
10// evbench 305436 8 executing a guard block).
115.0 9 - implement a Coro::State->safe_cancel method that might fail, but
12 - NEW ARCHITECTURE: use 4.912 for the stable version. 10 cancels in a "safer" way.
11 - implement Coro->cancel in XS for a 20% speed improvement, and to
12 be able to implement mutual cancellation.
13 - cancelling a coro while it itself is cancelling another coro is
14 now supported and working, instead of triggering an assertion.
15 - be a bit more crash-resistant when calling (buggy) on_destroy
16 callbacks (best effort).
17 - move on_destroy into the slf_frame, to allow extension slf
18 functions to have destructors.
19
205.372 Wed Feb 23 06:14:30 CET 2011
21 - apparently mingw doesn't provide a working gettimeofday, try to
22 work around that by relying on Time::HiRes (indirectly brought to
23 my attention by Max Maischein).
24 - fix some portability issues when Time::HiRes was used.
25
265.371 Mon Feb 21 14:36:08 CET 2011
27 - backport to windows process emulation code again.
28
295.37 Sat Feb 19 07:49:44 CET 2011
30 - add a big "Coro thread life cycle" section to "man Coro".
31 - try a tentative workaround against the breakage that 5.13 has
32 introduced without depreciation period. sigh.
33 - no longer use Time::HiRes if gettimeofday is available, which
34 saves quite a lot of memory.
35
365.36 Sun Feb 13 05:33:41 CET 2011
37 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
38 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
39 method call.
40 - undocument Coro::Timer::sleep and obsolete whole module.
41 - optimise Coro::Timer::timeout memory and cpu usage.
42 - slightly updated Coro::Intro for recent changes.
43 - do not initialise PL_dirty anymore.
44
455.25 Thu Nov 11 01:08:39 CET 2010
46 - try a different approach on netbsd - netbsd 5 finally has marginally
47 working pthreads, but still broken ucontext/sigaltstack.
48 - openbsd 4.8 finally got their act together, Coro works out of the box
49 with asm, setjmp and pthreads (no change, just informational).
50
515.24 Sat Oct 23 11:27:12 CEST 2010
52 - port to the EV 4.0 API.
53 - work around bugs in mingw32, making strawberry perl work
54 out of the box.
55 - correctly modify Coro::AIO function prototypes
56 so that they reflect the "no optional parameters" rule.
57 - "ported" libcoro to C++.
58
595.23 Mon May 17 18:50:42 CEST 2010
60 - be more resistant to ordering changes when initialising
61 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
62 Waldorf).
63 - document that perl 5.12 deliberately removed support for cloning.
64
655.22 Wed Apr 14 03:55:35 CEST 2010
66 - correctly return udnef on errors in Coro::Handle::read/write
67 (testcase by Marc Mims).
68 - convert Coro::Util into a "perl compatibility wrapper" - the functions
69 are less useful now, but are drop-in replacements for existing
70 functions, listing better alternatives in the documentation. This also
71 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
72 with Coro::Util::inet_aton.
73 - do not override $Coro::idle unconditionally in Coro.pm, as other
74 modules could have provided their own idle coro already
75 (for exmaple, Coro::AnyEvent).
76 - fix Coro::Util::gethost* functions.
77 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
78
795.21 Wed Dec 16 07:19:51 CET 2009
80 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
81 together.
82 - add some examples on how to combine other event loops with Coro in
83 Coro::AnyEvent, and how to run it (and not to block). Seems to be
84 the most common source of confusion.
85 - try to catch people naively blocking in an event callback.
86 - work around the perl filehandle bug issue in conjunction with
87 older common::sense (as indirectly pointed out by ZSystem).
88 - clarify the "not from signal handlers" section.
89
905.2 Sun Oct 4 14:54:24 CEST 2009
91 - Coro::Storable destroyed the prototypes of the functions it wrapped.
92 - export rouse_cb and rouse_wait by default now.
93 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
94 - new method $state->swap_sv.
95 - added section on "windows process emulation" to the manpage,
96 after a not-so-fruitful (nor-friendly) "discussion" with chip
97 salzenberg (discussion implies arguments, but his only arguments
98 were ad-hominems, one wonders why he started it in the first
99 place). I hope this explains it well enough for him to understand,
100 and maybe well enough for others to understand.
101 - use common::sense everywhere now.
102 - idle callbacks are no longer supported, use idle coros instead.
103 - print a thread listing when a deadlock is detected.
104
1055.17 Sat Aug 22 23:09:31 CEST 2009
106 - work around a bug in the perl debugger causing crashes
107 when running under the debugger by marking _pool_handler as nodebug.
108 - speed up Coro::async considerably.
109 - try some hacks to get netbsd to work "more often" - their broken
110 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
111 - convert Coro to AE 5.0 API.
112
1135.162 Tue Jul 28 04:04:03 CEST 2009
114 - perl 5.8.2 is now minimum requirement.
115 - skip t/19_handle.t on broken windows perls.
116
1175.161 Wed Jul 22 04:47:38 CEST 2009
118 - Coro::AnyEvent::poll could have a different prototype when EV was
119 used as backend (analyzed by Tatsuhiko Miyagawa).
120 - Coro::AnyEvent errornously initialised the event loop when loaded,
121 not on demand.
122 - try to workaround rare */t/01_unblock.t failures.
123
1245.16 Tue Jul 21 01:44:37 CEST 2009
125 - Coro::AnyEvent failed to hook into the event loop
126 when no threads had been readied between detecting
127 the event loop and actually running it.
128 - considerably speed up Coro::Select by taking avdantage
129 of AnyEvent > 4.8 and some other optimisations.
130 - implement paragraph readline mode in Coro::Handle
131 (based on patches by Zsbán Ambrus).
132 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
133 and analyzed by Yasuhiro MATSUMOTO).
134 - clarified libcoro license and copyright.
135 - someone stole my EXTRA_META!!!
136 - implement Coro::Select::patch_pp_sselect and it's brother,
137 for hardcode select overriding.
138
1395.151 Mon Jul 6 05:41:57 CEST 2009
140 - backport to windows process emulation code again (patch by
141 Yasuhiro MATSUMOTO).
142 - slightly update Coro::MakeMaker.
143
1445.15 Tue Jun 30 10:28:06 CEST 2009
145 - deprecate Coro::Socket, document how to get ipv6 support via
146 AnyEvent::Socket instead.
147 - implement signal->wait ($cb) interface, similar to semaphores.
148 - work around SvOK not supporting getmagic, so we have to getmagic
149 to test for undef :/ (reported by Matthias Waldorf).
150 - load Coro::AnyEvent in all modules using AnyEvent.
151 - work around perl corrupting our internal data structures,
152 reported by Tokuhiro Matsuno.
153 - enable per-coroutine real and cpu time gathering
154 (Coro::State::enable_times).
155
1565.14 Wed Jun 24 01:37:48 CEST 2009
157 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
158 as this seems to have been a difficult concept (poll, sleep, idle,
159 idle_upto).
160 - add Coro::AnyEvent::readable/writable functions.
161 - clarify Coro::EV/Event/AnyEvent manpages.
162 - free per-thread global scalars in the thread calling ->cancel, to
163 avoid crashes when $_, $@ etc., are magical but some of those
164 had already been freed.
165 - "unexperimentalise" the callback interface for Coro::Semaphore.
166 - speed up ready queue management/context switching by using a linked
167 list instead of an array (~5-10%).
168 - implement "watch" command in Coro::Debug shells.
169 - for fun, implement time-slicing as an example in the manpage.
170 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
171 Coro::Event, use the normal AnyEvent handling instead of dieing
172 (the same is true for Coro::Handle).
173 - properly document Coro::EV::timed_io_once.
174 - avoid unneccessary ->cancel calls in Coro::Handle.
175 - maybe make it work on mingw32 with win32 backend
176 (based on patch by Yasuhiro Matsumoto).
177
1785.132 Fri May 29 09:00:39 CEST 2009
179 - do not keep a reference to the argument itself in
180 Coro::Semaphore::guard, as it could change later.
181 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
182 should support a prepare callback.
183
1845.131 Mon Mar 16 23:20:37 CET 2009
185 - implement and document Coro->suspend, Coro->resume.
186 - fix Coro::Select implementation to not (often) close
187 the passed file descriptors (testcase provided by pippijn).
188
1895.13 Mon Dec 15 21:51:42 CET 2008
190 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
191 - don't set diehook to C<undef> but instead to NULL, to avoid
192 spurious warnings.
193 - fix a lot of bugs in Coro::SemaphoreSet.
194 - Coro::SemaphoreSet will less often create a semaphore needlessly.
195 - add Coro::SemaphoreSet::count and wait methods.
196 - take advantage of the new Guard module.
197 - deprecate Coro::guard.
198 - try to fix the dreaded 01_unblock tests once more. I hate it when
199 testsuites need more fixing than the code they are supposed to test.
200 - croak in more cases when a required callback isn't resolvable.
201 - fix some minor issues in Coro::State->call/eval.
202 - use current coroutine context instead of a temporary one
203 when temporarily switching to another coroutine.
204 - do not call C-level on_destroy handlers during global destruction,
205 to avoid needless segfaults.
206
2075.12 Sun Dec 7 13:30:38 CET 2008
208 - add default config for MirOS, which seems to be a bug-to-bug
209 compatible fork of openbsd ("world domination by releasing
210 openbsd cvs before the openbsd folks do it" or so :).
211 - free_padlist did destroy the names pad, not good, but didn't
212 seem to bother perl - this could fix issues such as eval ""
213 inside a function called from multiple coroutines.
214 - use a different method to detect destruction time.
215 - be more careful when freeing padlists just before global
216 destruction.
217 - fixed and expanded the call/cc example.
218 - renamed _terminate to _coro_run.
219 - new method Coro::Channel->shutdown.
220 - try pthreads on openbsd <4.4 (broken sigaltstack, will
221 pthreads fare better?).
222 - be less picky about destroying "a" running coroutine.
223
2245.11 Tue Nov 25 21:49:05 CET 2008
225 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
226 working code with failed assertions, introducing more bugs than
227 it fixes, requiring elaborate workarounds :(
228
2295.1 Mon Nov 24 08:54:59 CET 2008
230 - wrote a small introductory tutorial - Coro::Intro.
231 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
232 - Coro::Select did errornously dup the file descriptors
233 and didn't work with all AnyEvent backends.
234 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
235 LWP data transfers.
236 - disassociate c contexts from coro objects - this is agruably more
237 correct, but mostly allows sharing of cctxs between coro and state
238 objects, for added memory savings and speed increases.
239 - bumped $Coro::POOL_RSS up to 32kb by default.
240 - no longer set the optype to OP_CUSTOM, as B::* understandably
241 doesn't like this very much (and we *are* a type of entersub).
242 - implement state cloning, just to prove that call/cc can be done.
243 - automatically load Coro::AnyEvent in Coro::Handle.
244 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
245 do not have this method (and don't need it either).
246 - speed up generic anyevent methods in Coro::Handle by using rouse
247 callbacks.
248 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
249 others. It also makes the debugger happier, as you can trace
250 through the idle threads now.
251 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
252 - eg/event was pretty broken.
253 - better 5.8.6 compatibility.
254
2555.0 Thu Nov 20 10:35:05 CET 2008
256 - NEW ARCHITECTURE: use the latest 4.x version if you experience
257 stability issues.
13 - bump API version to 7 - all dependents must be recompiled. 258 - bump API version to 7 - all dependents must be recompiled.
259 - removed timed_* functions - they were not being used anyways
260 and should be replaced by a more generic mechanism -
261 and were annoying to support anyways :)
262 - removed SemaphoreSet's waiter method - use sem method instead.
263 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
264 - async_pool did free a scalar value twice
265 ("Attempt to unreference...").
266 - fix a longstanding bug where calling terminate on a coro that
267 was waiting for a semaphore that was just becoming available
268 would cause a deadlock (semaphore would get into a state where
269 it was available but waiters were still blocked).
270 - calling throw on a coroutine that is waiting for a semaphore will
271 no longer make it acquire the semaphore (and thus leak a count).
14 - perl's process emulation is now not even theoretically supported 272 - perl's process emulation is now not even theoretically supported
15 anymore. 273 anymore.
274 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
275 conversion of callback-style to blocking-style.
276 - new methods $coro->schedule_to and ->cede_to, to specifically
277 schedule or cede to a specific coroutine.
278 - new function Coro::Semaphore::wait.
279 - use named constants in Coro::Channel (Richard Hundt).
16 - directly patch the entersub opcode calling SLF functions (cede, 280 - directly patch the entersub opcode calling SLF functions (cede,
17 transfer and so on). this does speed up context switching, but 281 transfer and so on). this does speed up context switching, but
18 more importanly, it frees us from the hardcoded behaviour of 282 more importanly, it frees us from the hardcoded behaviour of
19 entersub, so we might actually be able to return something from 283 entersub, so we might actually be able to return something from
20 those functions and atcually create new ones. 284 those functions and atcually create new ones.
21 - take advantage of __builtin_frame_address on gcc. 285 - take advantage of __builtin_frame_address on gcc.
22 - expose THX in coroapi (not sure whether this was a wise decision, 286 - expose THX in coroapi (not sure whether this was a wise decision,
23 as "threaded" perls are running at half speed anyways). 287 as "threaded" perls are running at half speed anyways).
24 - implement execute_slf (schedule-like-function) interface that makes 288 - implement execute_slf (schedule-like-function) interface that makes
25 it possible to implement schedule-like-functions in XS. 289 it possible to implement schedule-like-functions in XS.
290 - use new SLF interface to massively speed up Coro::EV by roughly a
291 factor of two.
292 - used new SLF interface to massively speed up Coro::Semaphore by a
293 factor of three.
294 - used new SLF interface to speed up Coro::AIO by roughly a factor of
295 four and reduce its memory usage considerably.
296 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
297 for a nice speedup and vastly more correct behaviour. Also implement
298 a new method "sem" to get at the underlying semaphore object.
299 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
300 (in comparison) 20-40% speedup.
301 - used new SLF interface to reimplement Coro::Signal gaining
302 some unknown (because I was too lazy), but certain, speedup, and also
303 making signals reliable for the first time.
304 - used new SLF interface and other optimisations to speed up async_pool
305 by a factor of two. It also doesn't rely on perl's exception mechanism
306 to exit anymore. The overhead for terminating an async_pool, coro over
307 a normal async is now very small.
308 - sped up coroutine creation/destruction by 40%.
309 - forgot to include Coro/libcoro/README in the dist for all these years.
310 - work around a freebsd pthreads bug (manual testcancel is required as
311 pthread_cond_wait isn't a cancellation point on freebsd).
312 - use new rouse functions to speed up and simplify Coro::BDB.
313 - make "prefer perl native functions" work with threaded perls.
314 - condense Coro::Debug ps output, hint at v and w flags.
315 - (libcoro) lots of minor cleanups and portability improvements.
316
3174.914 Wed Nov 19 12:54:18 CET 2008
318 - fix a disastrous bug in the readline optimisation
319 introduced in 4.801.
320
3214.913 Sat Nov 15 07:58:28 CET 2008
322 - async_pool did free a scalar value twice
323 ("Attempt to unreference...").
26 324
274.912 Thu Nov 13 18:31:23 CET 2008 3254.912 Thu Nov 13 18:31:23 CET 2008
28 - minor cleanups. 326 - minor cleanups.
29 - use much larger stacks on linux and perl < 5.8.8. 327 - use much larger stacks on linux and perl < 5.8.8.
30 - Coro::Debug::new_unix_server did not unlink the socket 328 - Coro::Debug::new_unix_server did not unlink the socket

Diff Legend

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