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

Comparing Coro/Changes (file contents):
Revision 1.418 by root, Sun Nov 16 00:55:41 2008 UTC vs.
Revision 1.554 by root, Thu Mar 31 04:24:19 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
85.0 7 - be a bit more crash-resistant when calling (buggy) on_destroy
9 - NEW ARCHITECTURE: use 4.912 for the stable version. 8 callbacks (bets effort).
9
105.372 Wed Feb 23 06:14:30 CET 2011
11 - apparently mingw doesn't provide a working gettimeofday, try to
12 work around that by relying on Time::HiRes (indirectly brought to
13 my attention by Max Maischein).
14 - fix some portability issues when Time::HiRes was used.
15
165.371 Mon Feb 21 14:36:08 CET 2011
17 - backport to windows process emulation code again.
18
195.37 Sat Feb 19 07:49:44 CET 2011
20 - add a big "Coro thread life cycle" section to "man Coro".
21 - try a tentative workaround against the breakage that 5.13 has
22 introduced without depreciation period. sigh.
23 - no longer use Time::HiRes if gettimeofday is available, which
24 saves quite a lot of memory.
25
265.36 Sun Feb 13 05:33:41 CET 2011
27 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
28 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
29 method call.
30 - undocument Coro::Timer::sleep and obsolete whole module.
31 - optimise Coro::Timer::timeout memory and cpu usage.
32 - slightly updated Coro::Intro for recent changes.
33 - do not initialise PL_dirty anymore.
34
355.25 Thu Nov 11 01:08:39 CET 2010
36 - try a different approach on netbsd - netbsd 5 finally has marginally
37 working pthreads, but still broken ucontext/sigaltstack.
38 - openbsd 4.8 finally got their act together, Coro works out of the box
39 with asm, setjmp and pthreads (no change, just informational).
40
415.24 Sat Oct 23 11:27:12 CEST 2010
42 - port to the EV 4.0 API.
43 - work around bugs in mingw32, making strawberry perl work
44 out of the box.
45 - correctly modify Coro::AIO function prototypes
46 so that they reflect the "no optional parameters" rule.
47 - "ported" libcoro to C++.
48
495.23 Mon May 17 18:50:42 CEST 2010
50 - be more resistant to ordering changes when initialising
51 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
52 Waldorf).
53 - document that perl 5.12 deliberately removed support for cloning.
54
555.22 Wed Apr 14 03:55:35 CEST 2010
56 - correctly return udnef on errors in Coro::Handle::read/write
57 (testcase by Marc Mims).
58 - convert Coro::Util into a "perl compatibility wrapper" - the functions
59 are less useful now, but are drop-in replacements for existing
60 functions, listing better alternatives in the documentation. This also
61 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
62 with Coro::Util::inet_aton.
63 - do not override $Coro::idle unconditionally in Coro.pm, as other
64 modules could have provided their own idle coro already
65 (for exmaple, Coro::AnyEvent).
66 - fix Coro::Util::gethost* functions.
67 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
68
695.21 Wed Dec 16 07:19:51 CET 2009
70 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
71 together.
72 - add some examples on how to combine other event loops with Coro in
73 Coro::AnyEvent, and how to run it (and not to block). Seems to be
74 the most common source of confusion.
75 - try to catch people naively blocking in an event callback.
76 - work around the perl filehandle bug issue in conjunction with
77 older common::sense (as indirectly pointed out by ZSystem).
78 - clarify the "not from signal handlers" section.
79
805.2 Sun Oct 4 14:54:24 CEST 2009
81 - Coro::Storable destroyed the prototypes of the functions it wrapped.
82 - export rouse_cb and rouse_wait by default now.
83 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
84 - new method $state->swap_sv.
85 - added section on "windows process emulation" to the manpage,
86 after a not-so-fruitful (nor-friendly) "discussion" with chip
87 salzenberg (discussion implies arguments, but his only arguments
88 were ad-hominems, one wonders why he started it in the first
89 place). I hope this explains it well enough for him to understand,
90 and maybe well enough for others to understand.
91 - use common::sense everywhere now.
92 - idle callbacks are no longer supported, use idle coros instead.
93 - print a thread listing when a deadlock is detected.
94
955.17 Sat Aug 22 23:09:31 CEST 2009
96 - work around a bug in the perl debugger causing crashes
97 when running under the debugger by marking _pool_handler as nodebug.
98 - speed up Coro::async considerably.
99 - try some hacks to get netbsd to work "more often" - their broken
100 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
101 - convert Coro to AE 5.0 API.
102
1035.162 Tue Jul 28 04:04:03 CEST 2009
104 - perl 5.8.2 is now minimum requirement.
105 - skip t/19_handle.t on broken windows perls.
106
1075.161 Wed Jul 22 04:47:38 CEST 2009
108 - Coro::AnyEvent::poll could have a different prototype when EV was
109 used as backend (analyzed by Tatsuhiko Miyagawa).
110 - Coro::AnyEvent errornously initialised the event loop when loaded,
111 not on demand.
112 - try to workaround rare */t/01_unblock.t failures.
113
1145.16 Tue Jul 21 01:44:37 CEST 2009
115 - Coro::AnyEvent failed to hook into the event loop
116 when no threads had been readied between detecting
117 the event loop and actually running it.
118 - considerably speed up Coro::Select by taking avdantage
119 of AnyEvent > 4.8 and some other optimisations.
120 - implement paragraph readline mode in Coro::Handle
121 (based on patches by Zsbán Ambrus).
122 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
123 and analyzed by Yasuhiro MATSUMOTO).
124 - clarified libcoro license and copyright.
125 - someone stole my EXTRA_META!!!
126 - implement Coro::Select::patch_pp_sselect and it's brother,
127 for hardcode select overriding.
128
1295.151 Mon Jul 6 05:41:57 CEST 2009
130 - backport to windows process emulation code again (patch by
131 Yasuhiro MATSUMOTO).
132 - slightly update Coro::MakeMaker.
133
1345.15 Tue Jun 30 10:28:06 CEST 2009
135 - deprecate Coro::Socket, document how to get ipv6 support via
136 AnyEvent::Socket instead.
137 - implement signal->wait ($cb) interface, similar to semaphores.
138 - work around SvOK not supporting getmagic, so we have to getmagic
139 to test for undef :/ (reported by Matthias Waldorf).
140 - load Coro::AnyEvent in all modules using AnyEvent.
141 - work around perl corrupting our internal data structures,
142 reported by Tokuhiro Matsuno.
143 - enable per-coroutine real and cpu time gathering
144 (Coro::State::enable_times).
145
1465.14 Wed Jun 24 01:37:48 CEST 2009
147 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
148 as this seems to have been a difficult concept (poll, sleep, idle,
149 idle_upto).
150 - add Coro::AnyEvent::readable/writable functions.
151 - clarify Coro::EV/Event/AnyEvent manpages.
152 - free per-thread global scalars in the thread calling ->cancel, to
153 avoid crashes when $_, $@ etc., are magical but some of those
154 had already been freed.
155 - "unexperimentalise" the callback interface for Coro::Semaphore.
156 - speed up ready queue management/context switching by using a linked
157 list instead of an array (~5-10%).
158 - implement "watch" command in Coro::Debug shells.
159 - for fun, implement time-slicing as an example in the manpage.
160 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
161 Coro::Event, use the normal AnyEvent handling instead of dieing
162 (the same is true for Coro::Handle).
163 - properly document Coro::EV::timed_io_once.
164 - avoid unneccessary ->cancel calls in Coro::Handle.
165 - maybe make it work on mingw32 with win32 backend
166 (based on patch by Yasuhiro Matsumoto).
167
1685.132 Fri May 29 09:00:39 CEST 2009
169 - do not keep a reference to the argument itself in
170 Coro::Semaphore::guard, as it could change later.
171 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
172 should support a prepare callback.
173
1745.131 Mon Mar 16 23:20:37 CET 2009
175 - implement and document Coro->suspend, Coro->resume.
176 - fix Coro::Select implementation to not (often) close
177 the passed file descriptors (testcase provided by pippijn).
178
1795.13 Mon Dec 15 21:51:42 CET 2008
180 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
181 - don't set diehook to C<undef> but instead to NULL, to avoid
182 spurious warnings.
183 - fix a lot of bugs in Coro::SemaphoreSet.
184 - Coro::SemaphoreSet will less often create a semaphore needlessly.
185 - add Coro::SemaphoreSet::count and wait methods.
186 - take advantage of the new Guard module.
187 - deprecate Coro::guard.
188 - try to fix the dreaded 01_unblock tests once more. I hate it when
189 testsuites need more fixing than the code they are supposed to test.
190 - croak in more cases when a required callback isn't resolvable.
191 - fix some minor issues in Coro::State->call/eval.
192 - use current coroutine context instead of a temporary one
193 when temporarily switching to another coroutine.
194 - do not call C-level on_destroy handlers during global destruction,
195 to avoid needless segfaults.
196
1975.12 Sun Dec 7 13:30:38 CET 2008
198 - add default config for MirOS, which seems to be a bug-to-bug
199 compatible fork of openbsd ("world domination by releasing
200 openbsd cvs before the openbsd folks do it" or so :).
201 - free_padlist did destroy the names pad, not good, but didn't
202 seem to bother perl - this could fix issues such as eval ""
203 inside a function called from multiple coroutines.
204 - use a different method to detect destruction time.
205 - be more careful when freeing padlists just before global
206 destruction.
207 - fixed and expanded the call/cc example.
208 - renamed _terminate to _coro_run.
209 - new method Coro::Channel->shutdown.
210 - try pthreads on openbsd <4.4 (broken sigaltstack, will
211 pthreads fare better?).
212 - be less picky about destroying "a" running coroutine.
213
2145.11 Tue Nov 25 21:49:05 CET 2008
215 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
216 working code with failed assertions, introducing more bugs than
217 it fixes, requiring elaborate workarounds :(
218
2195.1 Mon Nov 24 08:54:59 CET 2008
220 - wrote a small introductory tutorial - Coro::Intro.
221 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
222 - Coro::Select did errornously dup the file descriptors
223 and didn't work with all AnyEvent backends.
224 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
225 LWP data transfers.
226 - disassociate c contexts from coro objects - this is agruably more
227 correct, but mostly allows sharing of cctxs between coro and state
228 objects, for added memory savings and speed increases.
229 - bumped $Coro::POOL_RSS up to 32kb by default.
230 - no longer set the optype to OP_CUSTOM, as B::* understandably
231 doesn't like this very much (and we *are* a type of entersub).
232 - implement state cloning, just to prove that call/cc can be done.
233 - automatically load Coro::AnyEvent in Coro::Handle.
234 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
235 do not have this method (and don't need it either).
236 - speed up generic anyevent methods in Coro::Handle by using rouse
237 callbacks.
238 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
239 others. It also makes the debugger happier, as you can trace
240 through the idle threads now.
241 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
242 - eg/event was pretty broken.
243 - better 5.8.6 compatibility.
244
2455.0 Thu Nov 20 10:35:05 CET 2008
246 - NEW ARCHITECTURE: use the latest 4.x version if you experience
247 stability issues.
10 - bump API version to 7 - all dependents must be recompiled. 248 - bump API version to 7 - all dependents must be recompiled.
249 - removed timed_* functions - they were not being used anyways
250 and should be replaced by a more generic mechanism -
251 and were annoying to support anyways :)
252 - removed SemaphoreSet's waiter method - use sem method instead.
11 - Coro::Semaphore->adjust didn't correctly wake up enough waiters. 253 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
12 - async_pool did free a scalar value twice 254 - async_pool did free a scalar value twice
13 ("Attempt to unreference..."). 255 ("Attempt to unreference...").
256 - fix a longstanding bug where calling terminate on a coro that
257 was waiting for a semaphore that was just becoming available
258 would cause a deadlock (semaphore would get into a state where
259 it was available but waiters were still blocked).
260 - calling throw on a coroutine that is waiting for a semaphore will
261 no longer make it acquire the semaphore (and thus leak a count).
14 - perl's process emulation is now not even theoretically supported 262 - perl's process emulation is now not even theoretically supported
15 anymore. 263 anymore.
264 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
265 conversion of callback-style to blocking-style.
266 - new methods $coro->schedule_to and ->cede_to, to specifically
267 schedule or cede to a specific coroutine.
268 - new function Coro::Semaphore::wait.
269 - use named constants in Coro::Channel (Richard Hundt).
16 - directly patch the entersub opcode calling SLF functions (cede, 270 - directly patch the entersub opcode calling SLF functions (cede,
17 transfer and so on). this does speed up context switching, but 271 transfer and so on). this does speed up context switching, but
18 more importanly, it frees us from the hardcoded behaviour of 272 more importanly, it frees us from the hardcoded behaviour of
19 entersub, so we might actually be able to return something from 273 entersub, so we might actually be able to return something from
20 those functions and atcually create new ones. 274 those functions and atcually create new ones.
21 - take advantage of __builtin_frame_address on gcc. 275 - take advantage of __builtin_frame_address on gcc.
22 - expose THX in coroapi (not sure whether this was a wise decision, 276 - expose THX in coroapi (not sure whether this was a wise decision,
23 as "threaded" perls are running at half speed anyways). 277 as "threaded" perls are running at half speed anyways).
24 - implement execute_slf (schedule-like-function) interface that makes 278 - implement execute_slf (schedule-like-function) interface that makes
25 it possible to implement schedule-like-functions in XS. 279 it possible to implement schedule-like-functions in XS.
26 - use new SLF interface to massively speed up Coro::EV by roughly a factor 280 - use new SLF interface to massively speed up Coro::EV by roughly a
27 of two. 281 factor of two.
28 - removed timed_down and timed_guard functions - they were not being used
29 anyways and should be replaced by a more generic mechanism - and were
30 annoying to support anyways :
31 - used new SLF interface to massively speed up Coro::Semaphore by a factor 282 - used new SLF interface to massively speed up Coro::Semaphore by a
32 of three. 283 factor of three.
284 - used new SLF interface to speed up Coro::AIO by roughly a factor of
285 four and reduce its memory usage considerably.
286 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
287 for a nice speedup and vastly more correct behaviour. Also implement
288 a new method "sem" to get at the underlying semaphore object.
289 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
290 (in comparison) 20-40% speedup.
291 - used new SLF interface to reimplement Coro::Signal gaining
292 some unknown (because I was too lazy), but certain, speedup, and also
293 making signals reliable for the first time.
294 - used new SLF interface and other optimisations to speed up async_pool
295 by a factor of two. It also doesn't rely on perl's exception mechanism
296 to exit anymore. The overhead for terminating an async_pool, coro over
297 a normal async is now very small.
298 - sped up coroutine creation/destruction by 40%.
33 - forgot to include Coro/libcoro/README in the dist for all these years. 299 - forgot to include Coro/libcoro/README in the dist for all these years.
34 - work around a freebsd pthreads bug (manual testcancel is required as 300 - work around a freebsd pthreads bug (manual testcancel is required as
35 pthread_cond_wait isn't a cancellation point on freebsd). 301 pthread_cond_wait isn't a cancellation point on freebsd).
302 - use new rouse functions to speed up and simplify Coro::BDB.
303 - make "prefer perl native functions" work with threaded perls.
304 - condense Coro::Debug ps output, hint at v and w flags.
305 - (libcoro) lots of minor cleanups and portability improvements.
306
3074.914 Wed Nov 19 12:54:18 CET 2008
308 - fix a disastrous bug in the readline optimisation
309 introduced in 4.801.
36 310
374.913 Sat Nov 15 07:58:28 CET 2008 3114.913 Sat Nov 15 07:58:28 CET 2008
38 - async_pool did free a scalar value twice 312 - async_pool did free a scalar value twice
39 ("Attempt to unreference..."). 313 ("Attempt to unreference...").
40 314

Diff Legend

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