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 1.562 by root, Tue May 10 19:55:48 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 that is being cancelled tries to block
10// evbench 305436 8 (e.g. while executing a guard block), instead of crashing or
115.0 9 deadlocking.
12 - NEW ARCHITECTURE: use 4.912 for the stable version. 10 - implement a Coro->safe_cancel method that might fail, but
11 cancels in a "safer" way if it succeeds.
12 - implement Coro->cancel in XS for a 20% speed improvement, and to
13 be able to implement mutual cancellation.
14 - speed up context switches by a percent or two by more efficiently
15 allocating context stack entries.
16 - cancelling a coro while it itself is cancelling another coro is
17 now supported and working, instead of triggering an assertion.
18 - be a bit more crash-resistant when calling (buggy) on_destroy
19 callbacks (best effort).
20 - move on_destroy into the slf_frame, to allow extension slf
21 functions to have destructors.
22
235.372 Wed Feb 23 06:14:30 CET 2011
24 - apparently mingw doesn't provide a working gettimeofday, try to
25 work around that by relying on Time::HiRes (indirectly brought to
26 my attention by Max Maischein).
27 - fix some portability issues when Time::HiRes was used.
28
295.371 Mon Feb 21 14:36:08 CET 2011
30 - backport to windows process emulation code again.
31
325.37 Sat Feb 19 07:49:44 CET 2011
33 - add a big "Coro thread life cycle" section to "man Coro".
34 - try a tentative workaround against the breakage that 5.13 has
35 introduced without depreciation period. sigh.
36 - no longer use Time::HiRes if gettimeofday is available, which
37 saves quite a lot of memory.
38
395.36 Sun Feb 13 05:33:41 CET 2011
40 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
41 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
42 method call.
43 - undocument Coro::Timer::sleep and obsolete whole module.
44 - optimise Coro::Timer::timeout memory and cpu usage.
45 - slightly updated Coro::Intro for recent changes.
46 - do not initialise PL_dirty anymore.
47
485.25 Thu Nov 11 01:08:39 CET 2010
49 - try a different approach on netbsd - netbsd 5 finally has marginally
50 working pthreads, but still broken ucontext/sigaltstack.
51 - openbsd 4.8 finally got their act together, Coro works out of the box
52 with asm, setjmp and pthreads (no change, just informational).
53
545.24 Sat Oct 23 11:27:12 CEST 2010
55 - port to the EV 4.0 API.
56 - work around bugs in mingw32, making strawberry perl work
57 out of the box.
58 - correctly modify Coro::AIO function prototypes
59 so that they reflect the "no optional parameters" rule.
60 - "ported" libcoro to C++.
61
625.23 Mon May 17 18:50:42 CEST 2010
63 - be more resistant to ordering changes when initialising
64 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
65 Waldorf).
66 - document that perl 5.12 deliberately removed support for cloning.
67
685.22 Wed Apr 14 03:55:35 CEST 2010
69 - correctly return udnef on errors in Coro::Handle::read/write
70 (testcase by Marc Mims).
71 - convert Coro::Util into a "perl compatibility wrapper" - the functions
72 are less useful now, but are drop-in replacements for existing
73 functions, listing better alternatives in the documentation. This also
74 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
75 with Coro::Util::inet_aton.
76 - do not override $Coro::idle unconditionally in Coro.pm, as other
77 modules could have provided their own idle coro already
78 (for exmaple, Coro::AnyEvent).
79 - fix Coro::Util::gethost* functions.
80 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
81
825.21 Wed Dec 16 07:19:51 CET 2009
83 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
84 together.
85 - add some examples on how to combine other event loops with Coro in
86 Coro::AnyEvent, and how to run it (and not to block). Seems to be
87 the most common source of confusion.
88 - try to catch people naively blocking in an event callback.
89 - work around the perl filehandle bug issue in conjunction with
90 older common::sense (as indirectly pointed out by ZSystem).
91 - clarify the "not from signal handlers" section.
92
935.2 Sun Oct 4 14:54:24 CEST 2009
94 - Coro::Storable destroyed the prototypes of the functions it wrapped.
95 - export rouse_cb and rouse_wait by default now.
96 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
97 - new method $state->swap_sv.
98 - added section on "windows process emulation" to the manpage,
99 after a not-so-fruitful (nor-friendly) "discussion" with chip
100 salzenberg (discussion implies arguments, but his only arguments
101 were ad-hominems, one wonders why he started it in the first
102 place). I hope this explains it well enough for him to understand,
103 and maybe well enough for others to understand.
104 - use common::sense everywhere now.
105 - idle callbacks are no longer supported, use idle coros instead.
106 - print a thread listing when a deadlock is detected.
107
1085.17 Sat Aug 22 23:09:31 CEST 2009
109 - work around a bug in the perl debugger causing crashes
110 when running under the debugger by marking _pool_handler as nodebug.
111 - speed up Coro::async considerably.
112 - try some hacks to get netbsd to work "more often" - their broken
113 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
114 - convert Coro to AE 5.0 API.
115
1165.162 Tue Jul 28 04:04:03 CEST 2009
117 - perl 5.8.2 is now minimum requirement.
118 - skip t/19_handle.t on broken windows perls.
119
1205.161 Wed Jul 22 04:47:38 CEST 2009
121 - Coro::AnyEvent::poll could have a different prototype when EV was
122 used as backend (analyzed by Tatsuhiko Miyagawa).
123 - Coro::AnyEvent errornously initialised the event loop when loaded,
124 not on demand.
125 - try to workaround rare */t/01_unblock.t failures.
126
1275.16 Tue Jul 21 01:44:37 CEST 2009
128 - Coro::AnyEvent failed to hook into the event loop
129 when no threads had been readied between detecting
130 the event loop and actually running it.
131 - considerably speed up Coro::Select by taking avdantage
132 of AnyEvent > 4.8 and some other optimisations.
133 - implement paragraph readline mode in Coro::Handle
134 (based on patches by Zsbán Ambrus).
135 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
136 and analyzed by Yasuhiro MATSUMOTO).
137 - clarified libcoro license and copyright.
138 - someone stole my EXTRA_META!!!
139 - implement Coro::Select::patch_pp_sselect and it's brother,
140 for hardcode select overriding.
141
1425.151 Mon Jul 6 05:41:57 CEST 2009
143 - backport to windows process emulation code again (patch by
144 Yasuhiro MATSUMOTO).
145 - slightly update Coro::MakeMaker.
146
1475.15 Tue Jun 30 10:28:06 CEST 2009
148 - deprecate Coro::Socket, document how to get ipv6 support via
149 AnyEvent::Socket instead.
150 - implement signal->wait ($cb) interface, similar to semaphores.
151 - work around SvOK not supporting getmagic, so we have to getmagic
152 to test for undef :/ (reported by Matthias Waldorf).
153 - load Coro::AnyEvent in all modules using AnyEvent.
154 - work around perl corrupting our internal data structures,
155 reported by Tokuhiro Matsuno.
156 - enable per-coroutine real and cpu time gathering
157 (Coro::State::enable_times).
158
1595.14 Wed Jun 24 01:37:48 CEST 2009
160 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
161 as this seems to have been a difficult concept (poll, sleep, idle,
162 idle_upto).
163 - add Coro::AnyEvent::readable/writable functions.
164 - clarify Coro::EV/Event/AnyEvent manpages.
165 - free per-thread global scalars in the thread calling ->cancel, to
166 avoid crashes when $_, $@ etc., are magical but some of those
167 had already been freed.
168 - "unexperimentalise" the callback interface for Coro::Semaphore.
169 - speed up ready queue management/context switching by using a linked
170 list instead of an array (~5-10%).
171 - implement "watch" command in Coro::Debug shells.
172 - for fun, implement time-slicing as an example in the manpage.
173 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
174 Coro::Event, use the normal AnyEvent handling instead of dieing
175 (the same is true for Coro::Handle).
176 - properly document Coro::EV::timed_io_once.
177 - avoid unneccessary ->cancel calls in Coro::Handle.
178 - maybe make it work on mingw32 with win32 backend
179 (based on patch by Yasuhiro Matsumoto).
180
1815.132 Fri May 29 09:00:39 CEST 2009
182 - do not keep a reference to the argument itself in
183 Coro::Semaphore::guard, as it could change later.
184 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
185 should support a prepare callback.
186
1875.131 Mon Mar 16 23:20:37 CET 2009
188 - implement and document Coro->suspend, Coro->resume.
189 - fix Coro::Select implementation to not (often) close
190 the passed file descriptors (testcase provided by pippijn).
191
1925.13 Mon Dec 15 21:51:42 CET 2008
193 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
194 - don't set diehook to C<undef> but instead to NULL, to avoid
195 spurious warnings.
196 - fix a lot of bugs in Coro::SemaphoreSet.
197 - Coro::SemaphoreSet will less often create a semaphore needlessly.
198 - add Coro::SemaphoreSet::count and wait methods.
199 - take advantage of the new Guard module.
200 - deprecate Coro::guard.
201 - try to fix the dreaded 01_unblock tests once more. I hate it when
202 testsuites need more fixing than the code they are supposed to test.
203 - croak in more cases when a required callback isn't resolvable.
204 - fix some minor issues in Coro::State->call/eval.
205 - use current coroutine context instead of a temporary one
206 when temporarily switching to another coroutine.
207 - do not call C-level on_destroy handlers during global destruction,
208 to avoid needless segfaults.
209
2105.12 Sun Dec 7 13:30:38 CET 2008
211 - add default config for MirOS, which seems to be a bug-to-bug
212 compatible fork of openbsd ("world domination by releasing
213 openbsd cvs before the openbsd folks do it" or so :).
214 - free_padlist did destroy the names pad, not good, but didn't
215 seem to bother perl - this could fix issues such as eval ""
216 inside a function called from multiple coroutines.
217 - use a different method to detect destruction time.
218 - be more careful when freeing padlists just before global
219 destruction.
220 - fixed and expanded the call/cc example.
221 - renamed _terminate to _coro_run.
222 - new method Coro::Channel->shutdown.
223 - try pthreads on openbsd <4.4 (broken sigaltstack, will
224 pthreads fare better?).
225 - be less picky about destroying "a" running coroutine.
226
2275.11 Tue Nov 25 21:49:05 CET 2008
228 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
229 working code with failed assertions, introducing more bugs than
230 it fixes, requiring elaborate workarounds :(
231
2325.1 Mon Nov 24 08:54:59 CET 2008
233 - wrote a small introductory tutorial - Coro::Intro.
234 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
235 - Coro::Select did errornously dup the file descriptors
236 and didn't work with all AnyEvent backends.
237 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
238 LWP data transfers.
239 - disassociate c contexts from coro objects - this is agruably more
240 correct, but mostly allows sharing of cctxs between coro and state
241 objects, for added memory savings and speed increases.
242 - bumped $Coro::POOL_RSS up to 32kb by default.
243 - no longer set the optype to OP_CUSTOM, as B::* understandably
244 doesn't like this very much (and we *are* a type of entersub).
245 - implement state cloning, just to prove that call/cc can be done.
246 - automatically load Coro::AnyEvent in Coro::Handle.
247 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
248 do not have this method (and don't need it either).
249 - speed up generic anyevent methods in Coro::Handle by using rouse
250 callbacks.
251 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
252 others. It also makes the debugger happier, as you can trace
253 through the idle threads now.
254 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
255 - eg/event was pretty broken.
256 - better 5.8.6 compatibility.
257
2585.0 Thu Nov 20 10:35:05 CET 2008
259 - NEW ARCHITECTURE: use the latest 4.x version if you experience
260 stability issues.
13 - bump API version to 7 - all dependents must be recompiled. 261 - bump API version to 7 - all dependents must be recompiled.
262 - removed timed_* functions - they were not being used anyways
263 and should be replaced by a more generic mechanism -
264 and were annoying to support anyways :)
265 - removed SemaphoreSet's waiter method - use sem method instead.
266 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
267 - async_pool did free a scalar value twice
268 ("Attempt to unreference...").
269 - fix a longstanding bug where calling terminate on a coro that
270 was waiting for a semaphore that was just becoming available
271 would cause a deadlock (semaphore would get into a state where
272 it was available but waiters were still blocked).
273 - calling throw on a coroutine that is waiting for a semaphore will
274 no longer make it acquire the semaphore (and thus leak a count).
14 - perl's process emulation is now not even theoretically supported 275 - perl's process emulation is now not even theoretically supported
15 anymore. 276 anymore.
277 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
278 conversion of callback-style to blocking-style.
279 - new methods $coro->schedule_to and ->cede_to, to specifically
280 schedule or cede to a specific coroutine.
281 - new function Coro::Semaphore::wait.
282 - use named constants in Coro::Channel (Richard Hundt).
16 - directly patch the entersub opcode calling SLF functions (cede, 283 - directly patch the entersub opcode calling SLF functions (cede,
17 transfer and so on). this does speed up context switching, but 284 transfer and so on). this does speed up context switching, but
18 more importanly, it frees us from the hardcoded behaviour of 285 more importanly, it frees us from the hardcoded behaviour of
19 entersub, so we might actually be able to return something from 286 entersub, so we might actually be able to return something from
20 those functions and atcually create new ones. 287 those functions and atcually create new ones.
21 - take advantage of __builtin_frame_address on gcc. 288 - take advantage of __builtin_frame_address on gcc.
22 - expose THX in coroapi (not sure whether this was a wise decision, 289 - expose THX in coroapi (not sure whether this was a wise decision,
23 as "threaded" perls are running at half speed anyways). 290 as "threaded" perls are running at half speed anyways).
24 - implement execute_slf (schedule-like-function) interface that makes 291 - implement execute_slf (schedule-like-function) interface that makes
25 it possible to implement schedule-like-functions in XS. 292 it possible to implement schedule-like-functions in XS.
293 - use new SLF interface to massively speed up Coro::EV by roughly a
294 factor of two.
295 - used new SLF interface to massively speed up Coro::Semaphore by a
296 factor of three.
297 - used new SLF interface to speed up Coro::AIO by roughly a factor of
298 four and reduce its memory usage considerably.
299 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
300 for a nice speedup and vastly more correct behaviour. Also implement
301 a new method "sem" to get at the underlying semaphore object.
302 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
303 (in comparison) 20-40% speedup.
304 - used new SLF interface to reimplement Coro::Signal gaining
305 some unknown (because I was too lazy), but certain, speedup, and also
306 making signals reliable for the first time.
307 - used new SLF interface and other optimisations to speed up async_pool
308 by a factor of two. It also doesn't rely on perl's exception mechanism
309 to exit anymore. The overhead for terminating an async_pool, coro over
310 a normal async is now very small.
311 - sped up coroutine creation/destruction by 40%.
312 - forgot to include Coro/libcoro/README in the dist for all these years.
313 - work around a freebsd pthreads bug (manual testcancel is required as
314 pthread_cond_wait isn't a cancellation point on freebsd).
315 - use new rouse functions to speed up and simplify Coro::BDB.
316 - make "prefer perl native functions" work with threaded perls.
317 - condense Coro::Debug ps output, hint at v and w flags.
318 - (libcoro) lots of minor cleanups and portability improvements.
319
3204.914 Wed Nov 19 12:54:18 CET 2008
321 - fix a disastrous bug in the readline optimisation
322 introduced in 4.801.
323
3244.913 Sat Nov 15 07:58:28 CET 2008
325 - async_pool did free a scalar value twice
326 ("Attempt to unreference...").
26 327
274.912 Thu Nov 13 18:31:23 CET 2008 3284.912 Thu Nov 13 18:31:23 CET 2008
28 - minor cleanups. 329 - minor cleanups.
29 - use much larger stacks on linux and perl < 5.8.8. 330 - use much larger stacks on linux and perl < 5.8.8.
30 - Coro::Debug::new_unix_server did not unlink the socket 331 - Coro::Debug::new_unix_server did not unlink the socket

Diff Legend

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