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

Comparing Coro/Changes (file contents):
Revision 1.432 by root, Tue Nov 18 10:42:43 2008 UTC vs.
Revision 1.525 by root, Thu Oct 1 23:50:23 2009 UTC

… …
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: maybe implement a default message channel, very much like Erlang's
5 actor model (which is cool in a lot of important aspects (failures!), 5 actor model (which is cool in a lot of important aspects (failures!),
6 but very lacking in others (higher level ipc)). 6 but very lacking in others (higher level ipc)).
7TODO: unready_all
8TODO: myhttpd header parsing
9TODO: channel->maxsize(newsize)?
7 10
8TODO: testcancel 11 - Coro::Storable destroyed the prototypes of the functions it wrapped.
9TODO: guard as SLF - to avoid deadlocks. 12 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
10TODO: Coro::Signal as SLF. 13 - added section on "windows process emulation" to the manpage,
11 - avoid deadlock in Coro::Channel if maxsize == 1 (Richard Hundt). 14 after a not-so-fruitful (nor-friendly) "discussion" with chip
15 salzenberg (discussion implies arguments, but his only arguments
16 were ad-hominems, one wonders why he started it in the first
17 place). I hope this explains it well enough for him to understand,
18 and maybe well enough for others to understand.
19 - use common::sense everywhere now.
20 - mark all SLF functions as nodebug, to avoid more debugger bugs.
21 - idle callbacks are no longer supported, use idle coros instead.
22 - print a thread listing when a deadlock is detected.
12 23
135.0 245.17 Sat Aug 22 23:09:31 CEST 2009
25 - work around a bug in the perl debugger causing crashes
26 when running under the debugger by marking _pool_handler as nodebug.
27 - speed up Coro::async considerably.
28 - try some hacks to get netbsd to work "more often" - their broken
29 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
30 - convert Coro to AE 5.0 API.
31
325.162 Tue Jul 28 04:04:03 CEST 2009
33 - perl 5.8.2 is now minimum requirement.
34 - skip t/19_handle.t on broken windows perls.
35
365.161 Wed Jul 22 04:47:38 CEST 2009
37 - Coro::AnyEvent::poll could have a different prototype when EV was
38 used as backend (analyzed by Tatsuhiko Miyagawa).
39 - Coro::AnyEvent errornously initialised the event loop when loaded,
40 not on demand.
41 - try to workaround rare */t/01_unblock.t failures.
42
435.16 Tue Jul 21 01:44:37 CEST 2009
44 - Coro::AnyEvent failed to hook into the event loop
45 when no threads had been readied between detecting
46 the event loop and actually running it.
47 - considerably speed up Coro::Select by taking avdantage
48 of AnyEvent > 4.8 and some other optimisations.
49 - implement paragraph readline mode in Coro::Handle
50 (based on patches by Zsbán Ambrus).
51 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
52 and analyzed by Yasuhiro MATSUMOTO).
53 - clarified libcoro license and copyright.
54 - someone stole my EXTRA_META!!!
55 - implement Coro::Select::patch_pp_sselect and it's brother,
56 for hardcode select overriding.
57
585.151 Mon Jul 6 05:41:57 CEST 2009
59 - backport to windows process emulation code again (patch by
60 Yasuhiro MATSUMOTO).
61 - slightly update Coro::MakeMaker.
62
635.15 Tue Jun 30 10:28:06 CEST 2009
64 - deprecate Coro::Socket, document how to get ipv6 support via
65 AnyEvent::Socket instead.
66 - implement signal->wait ($cb) interface, similar to semaphores.
67 - work around SvOK not supporting getmagic, so we have to getmagic
68 to test for undef :/ (reported by Matthias Waldorf).
69 - load Coro::AnyEvent in all modules using AnyEvent.
70 - work around perl corrupting our internal data structures,
71 reported by Tokuhiro Matsuno.
72 - enable per-coroutine real and cpu time gathering
73 (Coro::State::enable_times).
74
755.14 Wed Jun 24 01:37:48 CEST 2009
76 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
77 as this seems to have been a difficult concept (poll, sleep, idle,
78 idle_upto).
79 - add Coro::AnyEvent::readable/writable functions.
80 - clarify Coro::EV/Event/AnyEvent manpages.
81 - free per-thread global scalars in the thread calling ->cancel, to
82 avoid crashes when $_, $@ etc., are magical but some of those
83 had already been freed.
84 - "unexperimentalise" the callback interface for Coro::Semaphore.
85 - speed up ready queue management/context switching by using a linked
86 list instead of an array (~5-10%).
87 - implement "watch" command in Coro::Debug shells.
88 - for fun, implement time-slicing as an example in the manpage.
89 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
90 Coro::Event, use the normal AnyEvent handling instead of dieing
91 (the same is true for Coro::Handle).
92 - properly document Coro::EV::timed_io_once.
93 - avoid unneccessary ->cancel calls in Coro::Handle.
94 - maybe make it work on mingw32 with win32 backend
95 (based on patch by Yasuhiro Matsumoto).
96
975.132 Fri May 29 09:00:39 CEST 2009
98 - do not keep a reference to the argument itself in
99 Coro::Semaphore::guard, as it could change later.
100 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
101 should support a prepare callback.
102
1035.131 Mon Mar 16 23:20:37 CET 2009
104 - implement and document Coro->suspend, Coro->resume.
105 - fix Coro::Select implementation to not (often) close
106 the passed file descriptors (testcase provided by pippijn).
107
1085.13 Mon Dec 15 21:51:42 CET 2008
109 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
110 - don't set diehook to C<undef> but instead to NULL, to avoid
111 spurious warnings.
112 - fix a lot of bugs in Coro::SemaphoreSet.
113 - Coro::SemaphoreSet will less often create a semaphore needlessly.
114 - add Coro::SemaphoreSet::count and wait methods.
115 - take advantage of the new Guard module.
116 - deprecate Coro::guard.
117 - try to fix the dreaded 01_unblock tests once more. I hate it when
118 testsuites need more fixing than the code they are supposed to test.
119 - croak in more cases when a required callback isn't resolvable.
120 - fix some minor issues in Coro::State->call/eval.
121 - use current coroutine context instead of a temporary one
122 when temporarily switching to another coroutine.
123 - do not call C-level on_destroy handlers during global destruction,
124 to avoid needless segfaults.
125
1265.12 Sun Dec 7 13:30:38 CET 2008
127 - add default config for MirOS, which seems to be a bug-to-bug
128 compatible fork of openbsd ("world domination by releasing
129 openbsd cvs before the openbsd folks do it" or so :).
130 - free_padlist did destroy the names pad, not good, but didn't
131 seem to bother perl - this could fix issues such as eval ""
132 inside a function called from multiple coroutines.
133 - use a different method to detect destruction time.
134 - be more careful when freeing padlists just before global
135 destruction.
136 - fixed and expanded the call/cc example.
137 - renamed _terminate to _coro_run.
138 - new method Coro::Channel->shutdown.
139 - try pthreads on openbsd <4.4 (broken sigaltstack, will
140 pthreads fare better?).
141 - be less picky about destroying "a" running coroutine.
142
1435.11 Tue Nov 25 21:49:05 CET 2008
144 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
145 working code with failed assertions, introducing more bugs than
146 it fixes, requiring elaborate workarounds :(
147
1485.1 Mon Nov 24 08:54:59 CET 2008
149 - wrote a small introductory tutorial - Coro::Intro.
150 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
151 - Coro::Select did errornously dup the file descriptors
152 and didn't work with all AnyEvent backends.
153 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
154 LWP data transfers.
155 - disassociate c contexts from coro objects - this is agruably more
156 correct, but mostly allows sharing of cctxs between coro and state
157 objects, for added memory savings and speed increases.
158 - bumped $Coro::POOL_RSS up to 32kb by default.
159 - no longer set the optype to OP_CUSTOM, as B::* understandably
160 doesn't like this very much (and we *are* a type of entersub).
161 - implement state cloning, just to prove that call/cc can be done.
162 - automatically load Coro::AnyEvent in Coro::Handle.
163 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
164 do not have this method (and don't need it either).
165 - speed up generic anyevent methods in Coro::Handle by using rouse
166 callbacks.
167 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
168 others. It also makes the debugger happier, as you can trace
169 through the idle threads now.
170 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
171 - eg/event was pretty broken.
172 - better 5.8.6 compatibility.
173
1745.0 Thu Nov 20 10:35:05 CET 2008
14 - NEW ARCHITECTURE: use the latest 4.x version if you experience 175 - NEW ARCHITECTURE: use the latest 4.x version if you experience
15 stability issues. 176 stability issues.
16 - bump API version to 7 - all dependents must be recompiled. 177 - bump API version to 7 - all dependents must be recompiled.
17 - removed timed_* functions - they were not being used anyways 178 - removed timed_* functions - they were not being used anyways
18 and should be replaced by a more generic mechanism - 179 and should be replaced by a more generic mechanism -
19 and were annoying to support anyways :) 180 and were annoying to support anyways :)
20 - removed SemaphoreSet's waiter method - use sem instead. 181 - removed SemaphoreSet's waiter method - use sem method instead.
21 - Coro::Semaphore->adjust didn't correctly wake up enough waiters. 182 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
22 - async_pool did free a scalar value twice 183 - async_pool did free a scalar value twice
23 ("Attempt to unreference..."). 184 ("Attempt to unreference...").
24 - fix a longstanding bug where calling terminate on a coro that 185 - fix a longstanding bug where calling terminate on a coro that
25 was waiting for a semaphore that was just becoming available 186 was waiting for a semaphore that was just becoming available
26 would cause a deadlock (semaphore would get into a state where 187 would cause a deadlock (semaphore would get into a state where
27 it was available but waiters were waiting(. 188 it was available but waiters were still blocked).
28 - calling throw on a coroutine that is waiting for a semaphore will 189 - calling throw on a coroutine that is waiting for a semaphore will
29 no longer make it acquire the semaphore (and thus leak a count). 190 no longer make it acquire the semaphore (and thus leak a count).
30 - perl's process emulation is now not even theoretically supported 191 - perl's process emulation is now not even theoretically supported
31 anymore. 192 anymore.
193 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
194 conversion of callback-style to blocking-style.
195 - new methods $coro->schedule_to and ->cede_to, to specifically
196 schedule or cede to a specific coroutine.
197 - new function Coro::Semaphore::wait.
32 - use named constants in Coro::Channel (Richard Hundt). 198 - use named constants in Coro::Channel (Richard Hundt).
33 - directly patch the entersub opcode calling SLF functions (cede, 199 - directly patch the entersub opcode calling SLF functions (cede,
34 transfer and so on). this does speed up context switching, but 200 transfer and so on). this does speed up context switching, but
35 more importanly, it frees us from the hardcoded behaviour of 201 more importanly, it frees us from the hardcoded behaviour of
36 entersub, so we might actually be able to return something from 202 entersub, so we might actually be able to return something from
… …
49 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore, 215 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
50 for a nice speedup and vastly more correct behaviour. Also implement 216 for a nice speedup and vastly more correct behaviour. Also implement
51 a new method "sem" to get at the underlying semaphore object. 217 a new method "sem" to get at the underlying semaphore object.
52 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate 218 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
53 (in comparison) 20-40% speedup. 219 (in comparison) 20-40% speedup.
220 - used new SLF interface to reimplement Coro::Signal gaining
221 some unknown (because I was too lazy), but certain, speedup, and also
222 making signals reliable for the first time.
223 - used new SLF interface and other optimisations to speed up async_pool
224 by a factor of two. It also doesn't rely on perl's exception mechanism
225 to exit anymore. The overhead for terminating an async_pool, coro over
226 a normal async is now very small.
227 - sped up coroutine creation/destruction by 40%.
54 - forgot to include Coro/libcoro/README in the dist for all these years. 228 - forgot to include Coro/libcoro/README in the dist for all these years.
55 - work around a freebsd pthreads bug (manual testcancel is required as 229 - work around a freebsd pthreads bug (manual testcancel is required as
56 pthread_cond_wait isn't a cancellation point on freebsd). 230 pthread_cond_wait isn't a cancellation point on freebsd).
231 - use new rouse functions to speed up and simplify Coro::BDB.
232 - make "prefer perl native functions" work with threaded perls.
233 - condense Coro::Debug ps output, hint at v and w flags.
234 - (libcoro) lots of minor cleanups and portability improvements.
235
2364.914 Wed Nov 19 12:54:18 CET 2008
237 - fix a disastrous bug in the readline optimisation
238 introduced in 4.801.
57 239
584.913 Sat Nov 15 07:58:28 CET 2008 2404.913 Sat Nov 15 07:58:28 CET 2008
59 - async_pool did free a scalar value twice 241 - async_pool did free a scalar value twice
60 ("Attempt to unreference..."). 242 ("Attempt to unreference...").
61 243

Diff Legend

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