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

Comparing Coro/Changes (file contents):
Revision 1.446 by root, Thu Nov 20 03:11:29 2008 UTC vs.
Revision 1.580 by root, Wed Aug 3 09:43:03 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: async_pool to SLF 76.03 Wed Aug 3 11:41:30 CEST 2011
9TODO: _cancel to xs 8 - change how Coro patches AnyEvent condvars for compatibility to
105.0 9 AnyEvent 6.x.
10 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
11 hasn't been included.
12
136.02 Wed Jul 13 04:35:19 CEST 2011
14 - "improve portability to Gentoo" - gentoo manages to put perl variables
15 in memory areas that are farther than 2gb apart, which the jit couldn't
16 handle and barfed. now it's just a bit slower on gentoo and similar
17 systems.
18
196.01 Sun Jul 3 12:31:14 CEST 2011
20 - workarounds are good, but the test for whether pthreads are used
21 was not good. thisone should be better.
22 - check differently whether gcc generates cfi instructions itself.
23
246.0 Wed Jun 29 19:43:35 CEST 2011
25 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
26 destroyed and cleaned up automatically (e.g. async { schedule }).
27 - implement a JIT compiler for part of the thread switch code,
28 which gives a 50% speed improvement on threaded perls, and
29 about 4% on non-thraeded perls (so threaded perls now finally
30 reach about half the speed of non-threaded perls).
31 - slightly modernise Coro::Intro, add section about rouse functions.
32 - avoid DEFSV and ERRSV, giving another 10% improvement
33 in thread switching.
34 - Coro::State->is_destroyed is now called is_zombie.
35 - implement a Coro->safe_cancel method that might fail, but
36 cancels in a "safer" way if it succeeds.
37 - add preliminary support for DEBUGGING perls.
38 - get rid of two hash-accesses when initialising a new Coro - this
39 speeds up coro creation by almost a factor of two.
40 - croak when a coro that is being cancelled tries to block
41 (e.g. while executing a guard block), instead of crashing or
42 deadlocking.
43 - use a more robust and also faster method to identify Coro::State
44 objects - speeds up everything a bit.
45 - implement Coro->cancel in XS for a 20% speed improvement, and to
46 be able to implement mutual cancellation.
47 - speed up context switches by a percent or two by more efficiently
48 allocating context stack entries.
49 - implement Coro->join and Coro->on_destroy in XS for a speedup and
50 a reduction in memory use.
51 - cancelling a coro while it itself is cancelling another coro is
52 now supported and working, instead of triggering an assertion.
53 - be a bit more crash-resistant when calling (buggy) on_destroy
54 callbacks (best effort).
55 - move on_destroy into the slf_frame, to allow extension slf
56 functions to have destructors.
57 - get rid if coro refcounting - simply crash in other interpreter
58 threads by nulling the pointers on clone.
59 - simplify warn/die hook handling when loading Coro - the convoluted
60 logic seems to be no longer neccessary.
61 - use libecb instead of our own home-grown gcc hacks.
62 - document alternatives to Coro::LWP. Please use them :)
63 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
64 sigaltstack. Really. wine suffers from it, erlang suffers from it,
65 and it's known since at least 2006.
66
675.372 Wed Feb 23 06:14:30 CET 2011
68 - apparently mingw doesn't provide a working gettimeofday, try to
69 work around that by relying on Time::HiRes (indirectly brought to
70 my attention by Max Maischein).
71 - fix some portability issues when Time::HiRes was used.
72
735.371 Mon Feb 21 14:36:08 CET 2011
74 - backport to windows process emulation code again.
75
765.37 Sat Feb 19 07:49:44 CET 2011
77 - add a big "Coro thread life cycle" section to "man Coro".
78 - try a tentative workaround against the breakage that 5.13 has
79 introduced without depreciation period. sigh.
80 - no longer use Time::HiRes if gettimeofday is available, which
81 saves quite a lot of memory.
82
835.36 Sun Feb 13 05:33:41 CET 2011
84 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
85 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
86 method call.
87 - undocument Coro::Timer::sleep and obsolete whole module.
88 - optimise Coro::Timer::timeout memory and cpu usage.
89 - slightly updated Coro::Intro for recent changes.
90 - do not initialise PL_dirty anymore.
91
925.25 Thu Nov 11 01:08:39 CET 2010
93 - try a different approach on netbsd - netbsd 5 finally has marginally
94 working pthreads, but still broken ucontext/sigaltstack.
95 - openbsd 4.8 finally got their act together, Coro works out of the box
96 with asm, setjmp and pthreads (no change, just informational).
97
985.24 Sat Oct 23 11:27:12 CEST 2010
99 - port to the EV 4.0 API.
100 - work around bugs in mingw32, making strawberry perl work
101 out of the box.
102 - correctly modify Coro::AIO function prototypes
103 so that they reflect the "no optional parameters" rule.
104 - "ported" libcoro to C++.
105
1065.23 Mon May 17 18:50:42 CEST 2010
107 - be more resistant to ordering changes when initialising
108 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
109 Waldorf).
110 - document that perl 5.12 deliberately removed support for cloning.
111
1125.22 Wed Apr 14 03:55:35 CEST 2010
113 - correctly return udnef on errors in Coro::Handle::read/write
114 (testcase by Marc Mims).
115 - convert Coro::Util into a "perl compatibility wrapper" - the functions
116 are less useful now, but are drop-in replacements for existing
117 functions, listing better alternatives in the documentation. This also
118 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
119 with Coro::Util::inet_aton.
120 - do not override $Coro::idle unconditionally in Coro.pm, as other
121 modules could have provided their own idle coro already
122 (for exmaple, Coro::AnyEvent).
123 - fix Coro::Util::gethost* functions.
124 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
125
1265.21 Wed Dec 16 07:19:51 CET 2009
127 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
128 together.
129 - add some examples on how to combine other event loops with Coro in
130 Coro::AnyEvent, and how to run it (and not to block). Seems to be
131 the most common source of confusion.
132 - try to catch people naively blocking in an event callback.
133 - work around the perl filehandle bug issue in conjunction with
134 older common::sense (as indirectly pointed out by ZSystem).
135 - clarify the "not from signal handlers" section.
136
1375.2 Sun Oct 4 14:54:24 CEST 2009
138 - Coro::Storable destroyed the prototypes of the functions it wrapped.
139 - export rouse_cb and rouse_wait by default now.
140 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
141 - new method $state->swap_sv.
142 - added section on "windows process emulation" to the manpage,
143 after a not-so-fruitful (nor-friendly) "discussion" with chip
144 salzenberg (discussion implies arguments, but his only arguments
145 were ad-hominems, one wonders why he started it in the first
146 place). I hope this explains it well enough for him to understand,
147 and maybe well enough for others to understand.
148 - use common::sense everywhere now.
149 - idle callbacks are no longer supported, use idle coros instead.
150 - print a thread listing when a deadlock is detected.
151
1525.17 Sat Aug 22 23:09:31 CEST 2009
153 - work around a bug in the perl debugger causing crashes
154 when running under the debugger by marking _pool_handler as nodebug.
155 - speed up Coro::async considerably.
156 - try some hacks to get netbsd to work "more often" - their broken
157 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
158 - convert Coro to AE 5.0 API.
159
1605.162 Tue Jul 28 04:04:03 CEST 2009
161 - perl 5.8.2 is now minimum requirement.
162 - skip t/19_handle.t on broken windows perls.
163
1645.161 Wed Jul 22 04:47:38 CEST 2009
165 - Coro::AnyEvent::poll could have a different prototype when EV was
166 used as backend (analyzed by Tatsuhiko Miyagawa).
167 - Coro::AnyEvent errornously initialised the event loop when loaded,
168 not on demand.
169 - try to workaround rare */t/01_unblock.t failures.
170
1715.16 Tue Jul 21 01:44:37 CEST 2009
172 - Coro::AnyEvent failed to hook into the event loop
173 when no threads had been readied between detecting
174 the event loop and actually running it.
175 - considerably speed up Coro::Select by taking avdantage
176 of AnyEvent > 4.8 and some other optimisations.
177 - implement paragraph readline mode in Coro::Handle
178 (based on patches by Zsbán Ambrus).
179 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
180 and analyzed by Yasuhiro MATSUMOTO).
181 - clarified libcoro license and copyright.
182 - someone stole my EXTRA_META!!!
183 - implement Coro::Select::patch_pp_sselect and it's brother,
184 for hardcode select overriding.
185
1865.151 Mon Jul 6 05:41:57 CEST 2009
187 - backport to windows process emulation code again (patch by
188 Yasuhiro MATSUMOTO).
189 - slightly update Coro::MakeMaker.
190
1915.15 Tue Jun 30 10:28:06 CEST 2009
192 - deprecate Coro::Socket, document how to get ipv6 support via
193 AnyEvent::Socket instead.
194 - implement signal->wait ($cb) interface, similar to semaphores.
195 - work around SvOK not supporting getmagic, so we have to getmagic
196 to test for undef :/ (reported by Matthias Waldorf).
197 - load Coro::AnyEvent in all modules using AnyEvent.
198 - work around perl corrupting our internal data structures,
199 reported by Tokuhiro Matsuno.
200 - enable per-coroutine real and cpu time gathering
201 (Coro::State::enable_times).
202
2035.14 Wed Jun 24 01:37:48 CEST 2009
204 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
205 as this seems to have been a difficult concept (poll, sleep, idle,
206 idle_upto).
207 - add Coro::AnyEvent::readable/writable functions.
208 - clarify Coro::EV/Event/AnyEvent manpages.
209 - free per-thread global scalars in the thread calling ->cancel, to
210 avoid crashes when $_, $@ etc., are magical but some of those
211 had already been freed.
212 - "unexperimentalise" the callback interface for Coro::Semaphore.
213 - speed up ready queue management/context switching by using a linked
214 list instead of an array (~5-10%).
215 - implement "watch" command in Coro::Debug shells.
216 - for fun, implement time-slicing as an example in the manpage.
217 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
218 Coro::Event, use the normal AnyEvent handling instead of dieing
219 (the same is true for Coro::Handle).
220 - properly document Coro::EV::timed_io_once.
221 - avoid unneccessary ->cancel calls in Coro::Handle.
222 - maybe make it work on mingw32 with win32 backend
223 (based on patch by Yasuhiro Matsumoto).
224
2255.132 Fri May 29 09:00:39 CEST 2009
226 - do not keep a reference to the argument itself in
227 Coro::Semaphore::guard, as it could change later.
228 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
229 should support a prepare callback.
230
2315.131 Mon Mar 16 23:20:37 CET 2009
232 - implement and document Coro->suspend, Coro->resume.
233 - fix Coro::Select implementation to not (often) close
234 the passed file descriptors (testcase provided by pippijn).
235
2365.13 Mon Dec 15 21:51:42 CET 2008
237 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
238 - don't set diehook to C<undef> but instead to NULL, to avoid
239 spurious warnings.
240 - fix a lot of bugs in Coro::SemaphoreSet.
241 - Coro::SemaphoreSet will less often create a semaphore needlessly.
242 - add Coro::SemaphoreSet::count and wait methods.
243 - take advantage of the new Guard module.
244 - deprecate Coro::guard.
245 - try to fix the dreaded 01_unblock tests once more. I hate it when
246 testsuites need more fixing than the code they are supposed to test.
247 - croak in more cases when a required callback isn't resolvable.
248 - fix some minor issues in Coro::State->call/eval.
249 - use current coroutine context instead of a temporary one
250 when temporarily switching to another coroutine.
251 - do not call C-level on_destroy handlers during global destruction,
252 to avoid needless segfaults.
253
2545.12 Sun Dec 7 13:30:38 CET 2008
255 - add default config for MirOS, which seems to be a bug-to-bug
256 compatible fork of openbsd ("world domination by releasing
257 openbsd cvs before the openbsd folks do it" or so :).
258 - free_padlist did destroy the names pad, not good, but didn't
259 seem to bother perl - this could fix issues such as eval ""
260 inside a function called from multiple coroutines.
261 - use a different method to detect destruction time.
262 - be more careful when freeing padlists just before global
263 destruction.
264 - fixed and expanded the call/cc example.
265 - renamed _terminate to _coro_run.
266 - new method Coro::Channel->shutdown.
267 - try pthreads on openbsd <4.4 (broken sigaltstack, will
268 pthreads fare better?).
269 - be less picky about destroying "a" running coroutine.
270
2715.11 Tue Nov 25 21:49:05 CET 2008
272 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
273 working code with failed assertions, introducing more bugs than
274 it fixes, requiring elaborate workarounds :(
275
2765.1 Mon Nov 24 08:54:59 CET 2008
277 - wrote a small introductory tutorial - Coro::Intro.
278 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
279 - Coro::Select did errornously dup the file descriptors
280 and didn't work with all AnyEvent backends.
281 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
282 LWP data transfers.
283 - disassociate c contexts from coro objects - this is agruably more
284 correct, but mostly allows sharing of cctxs between coro and state
285 objects, for added memory savings and speed increases.
286 - bumped $Coro::POOL_RSS up to 32kb by default.
287 - no longer set the optype to OP_CUSTOM, as B::* understandably
288 doesn't like this very much (and we *are* a type of entersub).
289 - implement state cloning, just to prove that call/cc can be done.
290 - automatically load Coro::AnyEvent in Coro::Handle.
291 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
292 do not have this method (and don't need it either).
293 - speed up generic anyevent methods in Coro::Handle by using rouse
294 callbacks.
295 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
296 others. It also makes the debugger happier, as you can trace
297 through the idle threads now.
298 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
299 - eg/event was pretty broken.
300 - better 5.8.6 compatibility.
301
3025.0 Thu Nov 20 10:35:05 CET 2008
11 - NEW ARCHITECTURE: use the latest 4.x version if you experience 303 - NEW ARCHITECTURE: use the latest 4.x version if you experience
12 stability issues. 304 stability issues.
13 - bump API version to 7 - all dependents must be recompiled. 305 - bump API version to 7 - all dependents must be recompiled.
14 - removed timed_* functions - they were not being used anyways 306 - removed timed_* functions - they were not being used anyways
15 and should be replaced by a more generic mechanism - 307 and should be replaced by a more generic mechanism -
16 and were annoying to support anyways :) 308 and were annoying to support anyways :)
17 - removed SemaphoreSet's waiter method - use sem instead. 309 - removed SemaphoreSet's waiter method - use sem method instead.
18 - Coro::Semaphore->adjust didn't correctly wake up enough waiters. 310 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
19 - async_pool did free a scalar value twice 311 - async_pool did free a scalar value twice
20 ("Attempt to unreference..."). 312 ("Attempt to unreference...").
21 - fix a disastrous bug in the readline optimisation
22 introduced in 4.801.
23 - fix a longstanding bug where calling terminate on a coro that 313 - fix a longstanding bug where calling terminate on a coro that
24 was waiting for a semaphore that was just becoming available 314 was waiting for a semaphore that was just becoming available
25 would cause a deadlock (semaphore would get into a state where 315 would cause a deadlock (semaphore would get into a state where
26 it was available but waiters were still blocked). 316 it was available but waiters were still blocked).
27 - calling throw on a coroutine that is waiting for a semaphore will 317 - calling throw on a coroutine that is waiting for a semaphore will
28 no longer make it acquire the semaphore (and thus leak a count). 318 no longer make it acquire the semaphore (and thus leak a count).
29 - perl's process emulation is now not even theoretically supported 319 - perl's process emulation is now not even theoretically supported
30 anymore. 320 anymore.
31 - new functions Coro::rouse_cb and Coro::rouse_wait. 321 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
322 conversion of callback-style to blocking-style.
323 - new methods $coro->schedule_to and ->cede_to, to specifically
324 schedule or cede to a specific coroutine.
32 - new function Coro::Semaphore::wait. 325 - new function Coro::Semaphore::wait.
33 - use named constants in Coro::Channel (Richard Hundt). 326 - use named constants in Coro::Channel (Richard Hundt).
34 - directly patch the entersub opcode calling SLF functions (cede, 327 - directly patch the entersub opcode calling SLF functions (cede,
35 transfer and so on). this does speed up context switching, but 328 transfer and so on). this does speed up context switching, but
36 more importanly, it frees us from the hardcoded behaviour of 329 more importanly, it frees us from the hardcoded behaviour of
… …
66 - use new rouse functions to speed up and simplify Coro::BDB. 359 - use new rouse functions to speed up and simplify Coro::BDB.
67 - make "prefer perl native functions" work with threaded perls. 360 - make "prefer perl native functions" work with threaded perls.
68 - condense Coro::Debug ps output, hint at v and w flags. 361 - condense Coro::Debug ps output, hint at v and w flags.
69 - (libcoro) lots of minor cleanups and portability improvements. 362 - (libcoro) lots of minor cleanups and portability improvements.
70 363
3644.914 Wed Nov 19 12:54:18 CET 2008
365 - fix a disastrous bug in the readline optimisation
366 introduced in 4.801.
367
714.913 Sat Nov 15 07:58:28 CET 2008 3684.913 Sat Nov 15 07:58:28 CET 2008
72 - async_pool did free a scalar value twice 369 - async_pool did free a scalar value twice
73 ("Attempt to unreference..."). 370 ("Attempt to unreference...").
74 371
754.912 Thu Nov 13 18:31:23 CET 2008 3724.912 Thu Nov 13 18:31:23 CET 2008

Diff Legend

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