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

Comparing Coro/Changes (file contents):
Revision 1.378 by root, Tue Sep 30 17:12:34 2008 UTC vs.
Revision 1.510 by root, Thu Jul 9 18:45:57 2009 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::coroaio. 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
84.8 11 - considerably speed up Coro::Select by taking avdantage
12 of AnyEvent > 4.8 and some other optimisations.
13 - someone stole my EXTRA_META!!!
14
155.151 Mon Jul 6 05:41:57 CEST 2009
16 - backport to windows process emulation code again (patch by
17 Yasuhiro MATSUMOTO).
18 - slightly update Coro::MakeMaker.
19
205.15 Tue Jun 30 10:28:06 CEST 2009
21 - deprecate Coro::Socket, document how to get ipv6 support via
22 AnyEvent::Socket instead.
23 - implement signal->wait ($cb) interface, similar to semaphores.
24 - work around SvOK not supporting getmagic, so we have to getmagic
25 to test for undef :/ (reported by Matthias Waldorf).
26 - load Coro::AnyEvent in all modules using AnyEvent.
27 - work around perl corrupting our internal data structures,
28 reported by Tokuhiro Matsuno.
29 - enable per-coroutine real and cpu time gathering
30 (Coro::State::enable_times).
31
325.14 Wed Jun 24 01:37:48 CEST 2009
33 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
34 as this seems to have been a difficult concept (poll, sleep, idle,
35 idle_upto).
36 - add Coro::AnyEvent::readable/writable functions.
37 - clarify Coro::EV/Event/AnyEvent manpages.
38 - free per-thread global scalars in the thread calling ->cancel, to
39 avoid crashes when $_, $@ etc., are magical but some of those
40 had already been freed.
41 - "unexperimentalise" the callback interface for Coro::Semaphore.
42 - speed up ready queue management/context switching by using a linked
43 list instead of an array (~5-10%).
44 - implement "watch" command in Coro::Debug shells.
45 - for fun, implement time-slicing as an example in the manpage.
46 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
47 Coro::Event, use the normal AnyEvent handling instead of dieing
48 (the same is true for Coro::Handle).
49 - properly document Coro::EV::timed_io_once.
50 - avoid unneccessary ->cancel calls in Coro::Handle.
51 - maybe make it work on mingw32 with win32 backend
52 (based on patch by Yasuhiro Matsumoto).
53
545.132 Fri May 29 09:00:39 CEST 2009
55 - do not keep a reference to the argument itself in
56 Coro::Semaphore::guard, as it could change later.
57 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
58 should support a prepare callback.
59
605.131 Mon Mar 16 23:20:37 CET 2009
61 - implement and document Coro->suspend, Coro->resume.
62 - fix Coro::Select implementation to not (often) close
63 the passed file descriptors (testcase provided by pippijn).
64
655.13 Mon Dec 15 21:51:42 CET 2008
66 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
67 - don't set diehook to C<undef> but instead to NULL, to avoid
68 spurious warnings.
69 - fix a lot of bugs in Coro::SemaphoreSet.
70 - Coro::SemaphoreSet will less often create a semaphore needlessly.
71 - add Coro::SemaphoreSet::count and wait methods.
72 - take advantage of the new Guard module.
73 - deprecate Coro::guard.
74 - try to fix the dreaded 01_unblock tests once more. I hate it when
75 testsuites need more fixing than the code they are supposed to test.
76 - croak in more cases when a required callback isn't resolvable.
77 - fix some minor issues in Coro::State->call/eval.
78 - use current coroutine context instead of a temporary one
79 when temporarily switching to another coroutine.
80 - do not call C-level on_destroy handlers during global destruction,
81 to avoid needless segfaults.
82
835.12 Sun Dec 7 13:30:38 CET 2008
84 - add default config for MirOS, which seems to be a bug-to-bug
85 compatible fork of openbsd ("world domination by releasing
86 openbsd cvs before the openbsd folks do it" or so :).
87 - free_padlist did destroy the names pad, not good, but didn't
88 seem to bother perl - this could fix issues such as eval ""
89 inside a function called from multiple coroutines.
90 - use a different method to detect destruction time.
91 - be more careful when freeing padlists just before global
92 destruction.
93 - fixed and expanded the call/cc example.
94 - renamed _terminate to _coro_run.
95 - new method Coro::Channel->shutdown.
96 - try pthreads on openbsd <4.4 (broken sigaltstack, will
97 pthreads fare better?).
98 - be less picky about destroying "a" running coroutine.
99
1005.11 Tue Nov 25 21:49:05 CET 2008
101 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
102 working code with failed assertions, introducing more bugs than
103 it fixes, requiring elaborate workarounds :(
104
1055.1 Mon Nov 24 08:54:59 CET 2008
106 - wrote a small introductory tutorial - Coro::Intro.
107 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
108 - Coro::Select did errornously dup the file descriptors
109 and didn't work with all AnyEvent backends.
110 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
111 LWP data transfers.
112 - disassociate c contexts from coro objects - this is agruably more
113 correct, but mostly allows sharing of cctxs between coro and state
114 objects, for added memory savings and speed increases.
115 - bumped $Coro::POOL_RSS up to 32kb by default.
116 - no longer set the optype to OP_CUSTOM, as B::* understandably
117 doesn't like this very much (and we *are* a type of entersub).
118 - implement state cloning, just to prove that call/cc can be done.
119 - automatically load Coro::AnyEvent in Coro::Handle.
120 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
121 do not have this method (and don't need it either).
122 - speed up generic anyevent methods in Coro::Handle by using rouse
123 callbacks.
124 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
125 others. It also makes the debugger happier, as you can trace
126 through the idle threads now.
127 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
128 - eg/event was pretty broken.
129 - better 5.8.6 compatibility.
130
1315.0 Thu Nov 20 10:35:05 CET 2008
132 - NEW ARCHITECTURE: use the latest 4.x version if you experience
133 stability issues.
134 - bump API version to 7 - all dependents must be recompiled.
135 - removed timed_* functions - they were not being used anyways
136 and should be replaced by a more generic mechanism -
137 and were annoying to support anyways :)
138 - removed SemaphoreSet's waiter method - use sem method instead.
139 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
140 - async_pool did free a scalar value twice
141 ("Attempt to unreference...").
142 - fix a longstanding bug where calling terminate on a coro that
143 was waiting for a semaphore that was just becoming available
144 would cause a deadlock (semaphore would get into a state where
145 it was available but waiters were still blocked).
146 - calling throw on a coroutine that is waiting for a semaphore will
147 no longer make it acquire the semaphore (and thus leak a count).
148 - perl's process emulation is now not even theoretically supported
149 anymore.
150 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
151 conversion of callback-style to blocking-style.
152 - new methods $coro->schedule_to and ->cede_to, to specifically
153 schedule or cede to a specific coroutine.
154 - new function Coro::Semaphore::wait.
155 - use named constants in Coro::Channel (Richard Hundt).
156 - directly patch the entersub opcode calling SLF functions (cede,
157 transfer and so on). this does speed up context switching, but
158 more importanly, it frees us from the hardcoded behaviour of
159 entersub, so we might actually be able to return something from
160 those functions and atcually create new ones.
161 - take advantage of __builtin_frame_address on gcc.
162 - expose THX in coroapi (not sure whether this was a wise decision,
163 as "threaded" perls are running at half speed anyways).
164 - implement execute_slf (schedule-like-function) interface that makes
165 it possible to implement schedule-like-functions in XS.
166 - use new SLF interface to massively speed up Coro::EV by roughly a
167 factor of two.
168 - used new SLF interface to massively speed up Coro::Semaphore by a
169 factor of three.
170 - used new SLF interface to speed up Coro::AIO by roughly a factor of
171 four and reduce its memory usage considerably.
172 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
173 for a nice speedup and vastly more correct behaviour. Also implement
174 a new method "sem" to get at the underlying semaphore object.
175 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
176 (in comparison) 20-40% speedup.
177 - used new SLF interface to reimplement Coro::Signal gaining
178 some unknown (because I was too lazy), but certain, speedup, and also
179 making signals reliable for the first time.
180 - used new SLF interface and other optimisations to speed up async_pool
181 by a factor of two. It also doesn't rely on perl's exception mechanism
182 to exit anymore. The overhead for terminating an async_pool, coro over
183 a normal async is now very small.
184 - sped up coroutine creation/destruction by 40%.
185 - forgot to include Coro/libcoro/README in the dist for all these years.
186 - work around a freebsd pthreads bug (manual testcancel is required as
187 pthread_cond_wait isn't a cancellation point on freebsd).
188 - use new rouse functions to speed up and simplify Coro::BDB.
189 - make "prefer perl native functions" work with threaded perls.
190 - condense Coro::Debug ps output, hint at v and w flags.
191 - (libcoro) lots of minor cleanups and portability improvements.
192
1934.914 Wed Nov 19 12:54:18 CET 2008
194 - fix a disastrous bug in the readline optimisation
195 introduced in 4.801.
196
1974.913 Sat Nov 15 07:58:28 CET 2008
198 - async_pool did free a scalar value twice
199 ("Attempt to unreference...").
200
2014.912 Thu Nov 13 18:31:23 CET 2008
202 - minor cleanups.
203 - use much larger stacks on linux and perl < 5.8.8.
204 - Coro::Debug::new_unix_server did not unlink the socket
205 when destroyed.
206
2074.911 Tue Nov 11 04:26:17 CET 2008
208 - "port" to threaded perls.
209
2104.91 Mon Nov 10 05:36:38 CET 2008
211 - the ->throw exception object no longer leaks.
212 - creating a new cctx leaked a scopestack entry (memleak).
213 - new coroutines didn't get created with a zero flags field
214 (unknown impact).
215 - calling ->throw on a not-yet-started coroutine should now work
216 instead of being ignored.
217 - ->throw is now supported on Coro::State objects.
218 - clean up cctx creation code a bit.
219 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
220
2214.9 Sat Nov 8 17:45:27 CET 2008
222 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
223 - (libcoro) no longer leak threads in the experimental pthread backend,
224 also speed it up considerably.
225 - (libcoro) do not rely on makecontext passing void *'s unscathed.
226 - fix compiletime dependencies on libcoro in the Makefile.
227 - cctx_count wasn't always updated properly.
228 - Coro::State::cctx_stacksize wasn't applied correctly.
229 - new function Coro::State::cctx_max_idle.
230 - the default max number of idle C contexts is now 4.
231 - (libcoro) try harder to get _setjmp/_longjmp.
232 - (libcoro) cleanup and extend the libcoro API to officially
233 allow the creation of empty source contexts.
234 - very experimental workaround for the totally broken netbsd platform.
235 - tried to hack around openbsd bugs.
236
2374.804 Wed Nov 5 16:36:00 CET 2008
238 - Coro::Debug::new_unix_server would not create a non-blocking listening
239 socket, sometimes causing freezes.
240 - (libcoro) fix misaligned stack points for setjmp and assembly
241 methods, which can cause crashes on x86/x86_64 with a sufficiently
242 aggressive compiler.
243 - new function: Coro::Debug::new_tcp_server.
244 - move ->throw into the Coro class because it only works on coro objects.
245
2464.803 Mon Nov 3 17:16:12 CET 2008
247 - (libcoro) use a global asm statement to become independent of gcc
248 otpimisations for CORO_ASM (thanks to pippijn for the idea).
249 - try to workaround yet another broken bsd, this time dragonfly.
250
2514.802 Thu Oct 30 10:56:12 CET 2008
252 - support -fno-omit-frame-pointer on x86 with the assembly method.
253 - tune 01_unblock.t tests a bit.
254
2554.801 Wed Oct 22 18:33:37 CEST 2008
256 - improve readline speed for very long "lines".
257 - backport to 5.8.8.
258
2594.8 Thu Oct 2 13:34:40 CEST 2008
9 - new function Coro::AIO::aio_wait. 260 - new function Coro::AIO::aio_wait.
10 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent". 261 - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent".
11 - greatly speed up and reduce memory usage of Coro::AIO requests. 262 - greatly speed up and reduce memory usage of Coro::AIO requests.
263 - implement some other ยต-optimisations.
12 264
134.749 Mon Sep 29 14:40:12 CEST 2008 2654.749 Mon Sep 29 14:40:12 CEST 2008
14 - port to slow and broken pseudo-threaded perls. (courtesy pippijn). 266 - port to slow and broken pseudo-threaded perls. (courtesy pippijn).
15 267
164.748 Sat Sep 27 14:03:19 CEST 2008 2684.748 Sat Sep 27 14:03:19 CEST 2008

Diff Legend

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