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

Comparing Coro/Changes (file contents):
Revision 1.413 by root, Fri Nov 14 23:48:09 2008 UTC vs.
Revision 1.534 by root, Sun Apr 11 19:23:18 2010 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: unready_all
5 actor model (which is cool in a lot of important aspects (failures!), 5TODO: myhttpd header parsing
6 but very lacking in others (higher level ipc)). 6TODO: channel->maxsize(newsize)?
7TODO: MAP_STACK
8 7
95.0 8TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
10 - NEW ARCHITECTURE: use 4.912 for the stable version. 9
10 - convert Coro::Util into a "perl compatibility wrapper" - the functions
11 are less useful now, but are drop-in replacements for existing
12 functions, listing better alternatives in the documentation. This also
13 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
14 with Coro::Util::inet_aton.
15 - do not override $Coro::idle unconditionally in Coro.pm, as other
16 modules could have provided their own idle coro already
17 (for exmaple, Coro::AnyEvent).
18 - fix Coro::Util::gethost* functions.
19 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
20
215.21 Wed Dec 16 07:19:51 CET 2009
22 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
23 together.
24 - add some examples on how to combine other event loops with Coro in
25 Coro::AnyEvent, and how to run it (and not to block). Seems to be
26 the most common source of confusion.
27 - try to catch people naively blocking in an event callback.
28 - work around the perl filehandle bug issue in conjunction with
29 older common::sense (as indirectly pointed out by ZSystem).
30 - clarify the "not from signal handlers" section.
31
325.2 Sun Oct 4 14:54:24 CEST 2009
33 - Coro::Storable destroyed the prototypes of the functions it wrapped.
34 - export rouse_cb and rouse_wait by default now.
35 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
36 - new method $state->swap_sv.
37 - added section on "windows process emulation" to the manpage,
38 after a not-so-fruitful (nor-friendly) "discussion" with chip
39 salzenberg (discussion implies arguments, but his only arguments
40 were ad-hominems, one wonders why he started it in the first
41 place). I hope this explains it well enough for him to understand,
42 and maybe well enough for others to understand.
43 - use common::sense everywhere now.
44 - idle callbacks are no longer supported, use idle coros instead.
45 - print a thread listing when a deadlock is detected.
46
475.17 Sat Aug 22 23:09:31 CEST 2009
48 - work around a bug in the perl debugger causing crashes
49 when running under the debugger by marking _pool_handler as nodebug.
50 - speed up Coro::async considerably.
51 - try some hacks to get netbsd to work "more often" - their broken
52 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
53 - convert Coro to AE 5.0 API.
54
555.162 Tue Jul 28 04:04:03 CEST 2009
56 - perl 5.8.2 is now minimum requirement.
57 - skip t/19_handle.t on broken windows perls.
58
595.161 Wed Jul 22 04:47:38 CEST 2009
60 - Coro::AnyEvent::poll could have a different prototype when EV was
61 used as backend (analyzed by Tatsuhiko Miyagawa).
62 - Coro::AnyEvent errornously initialised the event loop when loaded,
63 not on demand.
64 - try to workaround rare */t/01_unblock.t failures.
65
665.16 Tue Jul 21 01:44:37 CEST 2009
67 - Coro::AnyEvent failed to hook into the event loop
68 when no threads had been readied between detecting
69 the event loop and actually running it.
70 - considerably speed up Coro::Select by taking avdantage
71 of AnyEvent > 4.8 and some other optimisations.
72 - implement paragraph readline mode in Coro::Handle
73 (based on patches by Zsbán Ambrus).
74 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
75 and analyzed by Yasuhiro MATSUMOTO).
76 - clarified libcoro license and copyright.
77 - someone stole my EXTRA_META!!!
78 - implement Coro::Select::patch_pp_sselect and it's brother,
79 for hardcode select overriding.
80
815.151 Mon Jul 6 05:41:57 CEST 2009
82 - backport to windows process emulation code again (patch by
83 Yasuhiro MATSUMOTO).
84 - slightly update Coro::MakeMaker.
85
865.15 Tue Jun 30 10:28:06 CEST 2009
87 - deprecate Coro::Socket, document how to get ipv6 support via
88 AnyEvent::Socket instead.
89 - implement signal->wait ($cb) interface, similar to semaphores.
90 - work around SvOK not supporting getmagic, so we have to getmagic
91 to test for undef :/ (reported by Matthias Waldorf).
92 - load Coro::AnyEvent in all modules using AnyEvent.
93 - work around perl corrupting our internal data structures,
94 reported by Tokuhiro Matsuno.
95 - enable per-coroutine real and cpu time gathering
96 (Coro::State::enable_times).
97
985.14 Wed Jun 24 01:37:48 CEST 2009
99 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
100 as this seems to have been a difficult concept (poll, sleep, idle,
101 idle_upto).
102 - add Coro::AnyEvent::readable/writable functions.
103 - clarify Coro::EV/Event/AnyEvent manpages.
104 - free per-thread global scalars in the thread calling ->cancel, to
105 avoid crashes when $_, $@ etc., are magical but some of those
106 had already been freed.
107 - "unexperimentalise" the callback interface for Coro::Semaphore.
108 - speed up ready queue management/context switching by using a linked
109 list instead of an array (~5-10%).
110 - implement "watch" command in Coro::Debug shells.
111 - for fun, implement time-slicing as an example in the manpage.
112 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
113 Coro::Event, use the normal AnyEvent handling instead of dieing
114 (the same is true for Coro::Handle).
115 - properly document Coro::EV::timed_io_once.
116 - avoid unneccessary ->cancel calls in Coro::Handle.
117 - maybe make it work on mingw32 with win32 backend
118 (based on patch by Yasuhiro Matsumoto).
119
1205.132 Fri May 29 09:00:39 CEST 2009
121 - do not keep a reference to the argument itself in
122 Coro::Semaphore::guard, as it could change later.
123 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
124 should support a prepare callback.
125
1265.131 Mon Mar 16 23:20:37 CET 2009
127 - implement and document Coro->suspend, Coro->resume.
128 - fix Coro::Select implementation to not (often) close
129 the passed file descriptors (testcase provided by pippijn).
130
1315.13 Mon Dec 15 21:51:42 CET 2008
132 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
133 - don't set diehook to C<undef> but instead to NULL, to avoid
134 spurious warnings.
135 - fix a lot of bugs in Coro::SemaphoreSet.
136 - Coro::SemaphoreSet will less often create a semaphore needlessly.
137 - add Coro::SemaphoreSet::count and wait methods.
138 - take advantage of the new Guard module.
139 - deprecate Coro::guard.
140 - try to fix the dreaded 01_unblock tests once more. I hate it when
141 testsuites need more fixing than the code they are supposed to test.
142 - croak in more cases when a required callback isn't resolvable.
143 - fix some minor issues in Coro::State->call/eval.
144 - use current coroutine context instead of a temporary one
145 when temporarily switching to another coroutine.
146 - do not call C-level on_destroy handlers during global destruction,
147 to avoid needless segfaults.
148
1495.12 Sun Dec 7 13:30:38 CET 2008
150 - add default config for MirOS, which seems to be a bug-to-bug
151 compatible fork of openbsd ("world domination by releasing
152 openbsd cvs before the openbsd folks do it" or so :).
153 - free_padlist did destroy the names pad, not good, but didn't
154 seem to bother perl - this could fix issues such as eval ""
155 inside a function called from multiple coroutines.
156 - use a different method to detect destruction time.
157 - be more careful when freeing padlists just before global
158 destruction.
159 - fixed and expanded the call/cc example.
160 - renamed _terminate to _coro_run.
161 - new method Coro::Channel->shutdown.
162 - try pthreads on openbsd <4.4 (broken sigaltstack, will
163 pthreads fare better?).
164 - be less picky about destroying "a" running coroutine.
165
1665.11 Tue Nov 25 21:49:05 CET 2008
167 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
168 working code with failed assertions, introducing more bugs than
169 it fixes, requiring elaborate workarounds :(
170
1715.1 Mon Nov 24 08:54:59 CET 2008
172 - wrote a small introductory tutorial - Coro::Intro.
173 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
174 - Coro::Select did errornously dup the file descriptors
175 and didn't work with all AnyEvent backends.
176 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
177 LWP data transfers.
178 - disassociate c contexts from coro objects - this is agruably more
179 correct, but mostly allows sharing of cctxs between coro and state
180 objects, for added memory savings and speed increases.
181 - bumped $Coro::POOL_RSS up to 32kb by default.
182 - no longer set the optype to OP_CUSTOM, as B::* understandably
183 doesn't like this very much (and we *are* a type of entersub).
184 - implement state cloning, just to prove that call/cc can be done.
185 - automatically load Coro::AnyEvent in Coro::Handle.
186 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
187 do not have this method (and don't need it either).
188 - speed up generic anyevent methods in Coro::Handle by using rouse
189 callbacks.
190 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
191 others. It also makes the debugger happier, as you can trace
192 through the idle threads now.
193 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
194 - eg/event was pretty broken.
195 - better 5.8.6 compatibility.
196
1975.0 Thu Nov 20 10:35:05 CET 2008
198 - NEW ARCHITECTURE: use the latest 4.x version if you experience
199 stability issues.
11 - bump API version to 7 - all dependents must be recompiled. 200 - bump API version to 7 - all dependents must be recompiled.
201 - removed timed_* functions - they were not being used anyways
202 and should be replaced by a more generic mechanism -
203 and were annoying to support anyways :)
204 - removed SemaphoreSet's waiter method - use sem method instead.
205 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
206 - async_pool did free a scalar value twice
207 ("Attempt to unreference...").
208 - fix a longstanding bug where calling terminate on a coro that
209 was waiting for a semaphore that was just becoming available
210 would cause a deadlock (semaphore would get into a state where
211 it was available but waiters were still blocked).
212 - calling throw on a coroutine that is waiting for a semaphore will
213 no longer make it acquire the semaphore (and thus leak a count).
12 - perl's process emulation is now not even theoretically supported 214 - perl's process emulation is now not even theoretically supported
13 anymore. 215 anymore.
216 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
217 conversion of callback-style to blocking-style.
218 - new methods $coro->schedule_to and ->cede_to, to specifically
219 schedule or cede to a specific coroutine.
220 - new function Coro::Semaphore::wait.
221 - use named constants in Coro::Channel (Richard Hundt).
14 - directly patch the entersub opcode calling SLF functions (cede, 222 - directly patch the entersub opcode calling SLF functions (cede,
15 transfer and so on). this does speed up context switching, but 223 transfer and so on). this does speed up context switching, but
16 more importanly, it frees us from the hardcoded behaviour of 224 more importanly, it frees us from the hardcoded behaviour of
17 entersub, so we might actually be able to return something from 225 entersub, so we might actually be able to return something from
18 those functions and atcually create new ones. 226 those functions and atcually create new ones.
19 - take advantage of __builtin_frame_address on gcc. 227 - take advantage of __builtin_frame_address on gcc.
20 - expose THX in coroapi (not sure whether this was a wise decision, 228 - expose THX in coroapi (not sure whether this was a wise decision,
21 as "threaded" perls are running at half speed anyways). 229 as "threaded" perls are running at half speed anyways).
22 - implement execute_slf (schedule-like-function) interface that makes 230 - implement execute_slf (schedule-like-function) interface that makes
23 it possible to implement schedule-like-functions in XS. 231 it possible to implement schedule-like-functions in XS.
24 - use new SLF interface to massively speed up Coro::EV (it's roughly 232 - use new SLF interface to massively speed up Coro::EV by roughly a
25 twice as fast now). 233 factor of two.
234 - used new SLF interface to massively speed up Coro::Semaphore by a
235 factor of three.
236 - used new SLF interface to speed up Coro::AIO by roughly a factor of
237 four and reduce its memory usage considerably.
238 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
239 for a nice speedup and vastly more correct behaviour. Also implement
240 a new method "sem" to get at the underlying semaphore object.
241 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
242 (in comparison) 20-40% speedup.
243 - used new SLF interface to reimplement Coro::Signal gaining
244 some unknown (because I was too lazy), but certain, speedup, and also
245 making signals reliable for the first time.
246 - used new SLF interface and other optimisations to speed up async_pool
247 by a factor of two. It also doesn't rely on perl's exception mechanism
248 to exit anymore. The overhead for terminating an async_pool, coro over
249 a normal async is now very small.
250 - sped up coroutine creation/destruction by 40%.
251 - forgot to include Coro/libcoro/README in the dist for all these years.
252 - work around a freebsd pthreads bug (manual testcancel is required as
253 pthread_cond_wait isn't a cancellation point on freebsd).
254 - use new rouse functions to speed up and simplify Coro::BDB.
255 - make "prefer perl native functions" work with threaded perls.
256 - condense Coro::Debug ps output, hint at v and w flags.
257 - (libcoro) lots of minor cleanups and portability improvements.
258
2594.914 Wed Nov 19 12:54:18 CET 2008
260 - fix a disastrous bug in the readline optimisation
261 introduced in 4.801.
262
2634.913 Sat Nov 15 07:58:28 CET 2008
264 - async_pool did free a scalar value twice
265 ("Attempt to unreference...").
26 266
274.912 Thu Nov 13 18:31:23 CET 2008 2674.912 Thu Nov 13 18:31:23 CET 2008
28 - minor cleanups. 268 - minor cleanups.
29 - use much larger stacks on linux and perl < 5.8.8. 269 - use much larger stacks on linux and perl < 5.8.8.
30 - Coro::Debug::new_unix_server did not unlink the socket 270 - Coro::Debug::new_unix_server did not unlink the socket

Diff Legend

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