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

Comparing Coro/Changes (file contents):
Revision 1.408 by root, Fri Nov 14 02:29:09 2008 UTC vs.
Revision 1.526 by root, Thu Oct 1 23:51:33 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
85.0 11 - Coro::Storable destroyed the prototypes of the functions it wrapped.
9 - ARCHITECTURAL REDESIGN, use 4.912 for the stable version. 12 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
13 - added section on "windows process emulation" to the manpage,
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 - idle callbacks are no longer supported, use idle coros instead.
21 - print a thread listing when a deadlock is detected.
22
235.17 Sat Aug 22 23:09:31 CEST 2009
24 - work around a bug in the perl debugger causing crashes
25 when running under the debugger by marking _pool_handler as nodebug.
26 - speed up Coro::async considerably.
27 - try some hacks to get netbsd to work "more often" - their broken
28 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
29 - convert Coro to AE 5.0 API.
30
315.162 Tue Jul 28 04:04:03 CEST 2009
32 - perl 5.8.2 is now minimum requirement.
33 - skip t/19_handle.t on broken windows perls.
34
355.161 Wed Jul 22 04:47:38 CEST 2009
36 - Coro::AnyEvent::poll could have a different prototype when EV was
37 used as backend (analyzed by Tatsuhiko Miyagawa).
38 - Coro::AnyEvent errornously initialised the event loop when loaded,
39 not on demand.
40 - try to workaround rare */t/01_unblock.t failures.
41
425.16 Tue Jul 21 01:44:37 CEST 2009
43 - Coro::AnyEvent failed to hook into the event loop
44 when no threads had been readied between detecting
45 the event loop and actually running it.
46 - considerably speed up Coro::Select by taking avdantage
47 of AnyEvent > 4.8 and some other optimisations.
48 - implement paragraph readline mode in Coro::Handle
49 (based on patches by Zsbán Ambrus).
50 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
51 and analyzed by Yasuhiro MATSUMOTO).
52 - clarified libcoro license and copyright.
53 - someone stole my EXTRA_META!!!
54 - implement Coro::Select::patch_pp_sselect and it's brother,
55 for hardcode select overriding.
56
575.151 Mon Jul 6 05:41:57 CEST 2009
58 - backport to windows process emulation code again (patch by
59 Yasuhiro MATSUMOTO).
60 - slightly update Coro::MakeMaker.
61
625.15 Tue Jun 30 10:28:06 CEST 2009
63 - deprecate Coro::Socket, document how to get ipv6 support via
64 AnyEvent::Socket instead.
65 - implement signal->wait ($cb) interface, similar to semaphores.
66 - work around SvOK not supporting getmagic, so we have to getmagic
67 to test for undef :/ (reported by Matthias Waldorf).
68 - load Coro::AnyEvent in all modules using AnyEvent.
69 - work around perl corrupting our internal data structures,
70 reported by Tokuhiro Matsuno.
71 - enable per-coroutine real and cpu time gathering
72 (Coro::State::enable_times).
73
745.14 Wed Jun 24 01:37:48 CEST 2009
75 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
76 as this seems to have been a difficult concept (poll, sleep, idle,
77 idle_upto).
78 - add Coro::AnyEvent::readable/writable functions.
79 - clarify Coro::EV/Event/AnyEvent manpages.
80 - free per-thread global scalars in the thread calling ->cancel, to
81 avoid crashes when $_, $@ etc., are magical but some of those
82 had already been freed.
83 - "unexperimentalise" the callback interface for Coro::Semaphore.
84 - speed up ready queue management/context switching by using a linked
85 list instead of an array (~5-10%).
86 - implement "watch" command in Coro::Debug shells.
87 - for fun, implement time-slicing as an example in the manpage.
88 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
89 Coro::Event, use the normal AnyEvent handling instead of dieing
90 (the same is true for Coro::Handle).
91 - properly document Coro::EV::timed_io_once.
92 - avoid unneccessary ->cancel calls in Coro::Handle.
93 - maybe make it work on mingw32 with win32 backend
94 (based on patch by Yasuhiro Matsumoto).
95
965.132 Fri May 29 09:00:39 CEST 2009
97 - do not keep a reference to the argument itself in
98 Coro::Semaphore::guard, as it could change later.
99 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
100 should support a prepare callback.
101
1025.131 Mon Mar 16 23:20:37 CET 2009
103 - implement and document Coro->suspend, Coro->resume.
104 - fix Coro::Select implementation to not (often) close
105 the passed file descriptors (testcase provided by pippijn).
106
1075.13 Mon Dec 15 21:51:42 CET 2008
108 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
109 - don't set diehook to C<undef> but instead to NULL, to avoid
110 spurious warnings.
111 - fix a lot of bugs in Coro::SemaphoreSet.
112 - Coro::SemaphoreSet will less often create a semaphore needlessly.
113 - add Coro::SemaphoreSet::count and wait methods.
114 - take advantage of the new Guard module.
115 - deprecate Coro::guard.
116 - try to fix the dreaded 01_unblock tests once more. I hate it when
117 testsuites need more fixing than the code they are supposed to test.
118 - croak in more cases when a required callback isn't resolvable.
119 - fix some minor issues in Coro::State->call/eval.
120 - use current coroutine context instead of a temporary one
121 when temporarily switching to another coroutine.
122 - do not call C-level on_destroy handlers during global destruction,
123 to avoid needless segfaults.
124
1255.12 Sun Dec 7 13:30:38 CET 2008
126 - add default config for MirOS, which seems to be a bug-to-bug
127 compatible fork of openbsd ("world domination by releasing
128 openbsd cvs before the openbsd folks do it" or so :).
129 - free_padlist did destroy the names pad, not good, but didn't
130 seem to bother perl - this could fix issues such as eval ""
131 inside a function called from multiple coroutines.
132 - use a different method to detect destruction time.
133 - be more careful when freeing padlists just before global
134 destruction.
135 - fixed and expanded the call/cc example.
136 - renamed _terminate to _coro_run.
137 - new method Coro::Channel->shutdown.
138 - try pthreads on openbsd <4.4 (broken sigaltstack, will
139 pthreads fare better?).
140 - be less picky about destroying "a" running coroutine.
141
1425.11 Tue Nov 25 21:49:05 CET 2008
143 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
144 working code with failed assertions, introducing more bugs than
145 it fixes, requiring elaborate workarounds :(
146
1475.1 Mon Nov 24 08:54:59 CET 2008
148 - wrote a small introductory tutorial - Coro::Intro.
149 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
150 - Coro::Select did errornously dup the file descriptors
151 and didn't work with all AnyEvent backends.
152 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
153 LWP data transfers.
154 - disassociate c contexts from coro objects - this is agruably more
155 correct, but mostly allows sharing of cctxs between coro and state
156 objects, for added memory savings and speed increases.
157 - bumped $Coro::POOL_RSS up to 32kb by default.
158 - no longer set the optype to OP_CUSTOM, as B::* understandably
159 doesn't like this very much (and we *are* a type of entersub).
160 - implement state cloning, just to prove that call/cc can be done.
161 - automatically load Coro::AnyEvent in Coro::Handle.
162 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
163 do not have this method (and don't need it either).
164 - speed up generic anyevent methods in Coro::Handle by using rouse
165 callbacks.
166 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
167 others. It also makes the debugger happier, as you can trace
168 through the idle threads now.
169 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
170 - eg/event was pretty broken.
171 - better 5.8.6 compatibility.
172
1735.0 Thu Nov 20 10:35:05 CET 2008
174 - NEW ARCHITECTURE: use the latest 4.x version if you experience
175 stability issues.
176 - bump API version to 7 - all dependents must be recompiled.
177 - removed timed_* functions - they were not being used anyways
178 and should be replaced by a more generic mechanism -
179 and were annoying to support anyways :)
180 - removed SemaphoreSet's waiter method - use sem method instead.
181 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
182 - async_pool did free a scalar value twice
183 ("Attempt to unreference...").
184 - fix a longstanding bug where calling terminate on a coro that
185 was waiting for a semaphore that was just becoming available
186 would cause a deadlock (semaphore would get into a state where
187 it was available but waiters were still blocked).
188 - calling throw on a coroutine that is waiting for a semaphore will
189 no longer make it acquire the semaphore (and thus leak a count).
190 - perl's process emulation is now not even theoretically supported
191 anymore.
192 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
193 conversion of callback-style to blocking-style.
194 - new methods $coro->schedule_to and ->cede_to, to specifically
195 schedule or cede to a specific coroutine.
196 - new function Coro::Semaphore::wait.
197 - use named constants in Coro::Channel (Richard Hundt).
10 - directly patch the entersub opcode calling ssl functions (cede, 198 - directly patch the entersub opcode calling SLF functions (cede,
11 transfer and so on). this does speed up context switching by 199 transfer and so on). this does speed up context switching, but
12 up to 10%. More importanly, it frees us from the hardcoded 200 more importanly, it frees us from the hardcoded behaviour of
13 behaviour of entersub, so we might actually be able to return 201 entersub, so we might actually be able to return something from
14 something from those functions. 202 those functions and atcually create new ones.
203 - take advantage of __builtin_frame_address on gcc.
204 - expose THX in coroapi (not sure whether this was a wise decision,
205 as "threaded" perls are running at half speed anyways).
206 - implement execute_slf (schedule-like-function) interface that makes
207 it possible to implement schedule-like-functions in XS.
208 - use new SLF interface to massively speed up Coro::EV by roughly a
209 factor of two.
210 - used new SLF interface to massively speed up Coro::Semaphore by a
211 factor of three.
212 - used new SLF interface to speed up Coro::AIO by roughly a factor of
213 four and reduce its memory usage considerably.
214 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
215 for a nice speedup and vastly more correct behaviour. Also implement
216 a new method "sem" to get at the underlying semaphore object.
217 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
218 (in comparison) 20-40% speedup.
219 - used new SLF interface to reimplement Coro::Signal gaining
220 some unknown (because I was too lazy), but certain, speedup, and also
221 making signals reliable for the first time.
222 - used new SLF interface and other optimisations to speed up async_pool
223 by a factor of two. It also doesn't rely on perl's exception mechanism
224 to exit anymore. The overhead for terminating an async_pool, coro over
225 a normal async is now very small.
226 - sped up coroutine creation/destruction by 40%.
227 - forgot to include Coro/libcoro/README in the dist for all these years.
228 - work around a freebsd pthreads bug (manual testcancel is required as
229 pthread_cond_wait isn't a cancellation point on freebsd).
230 - use new rouse functions to speed up and simplify Coro::BDB.
231 - make "prefer perl native functions" work with threaded perls.
232 - condense Coro::Debug ps output, hint at v and w flags.
233 - (libcoro) lots of minor cleanups and portability improvements.
234
2354.914 Wed Nov 19 12:54:18 CET 2008
236 - fix a disastrous bug in the readline optimisation
237 introduced in 4.801.
238
2394.913 Sat Nov 15 07:58:28 CET 2008
240 - async_pool did free a scalar value twice
241 ("Attempt to unreference...").
15 242
164.912 Thu Nov 13 18:31:23 CET 2008 2434.912 Thu Nov 13 18:31:23 CET 2008
17 - minor cleanups. 244 - minor cleanups.
18 - use much larger stacks on linux and perl < 5.8.8. 245 - use much larger stacks on linux and perl < 5.8.8.
19 - Coro::Debug::new_unix_server did not unlink the socket 246 - Coro::Debug::new_unix_server did not unlink the socket

Diff Legend

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