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 1.499 by root, Tue Jun 23 23:40:06 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
85.0 115.14 Wed Jun 24 01:37:48 CEST 2009
9 - ARCHITECTURAL REDESIGN, use 4.912 for the stable version. 12 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
13 as this seems to have been a difficult concept (poll, sleep, idle,
14 idle_upto).
15 - add Coro::AnyEvent::readable/writable functions.
16 - clarify Coro::EV/Event/AnyEvent manpages.
17 - free per-thread global scalars in the thread calling ->cancel, to
18 avoid crashes when $_, $@ etc., are magical but some of those
19 had already been freed.
20 - "unexperimentalise" the callback interface for Coro::Semaphore.
21 - speed up ready queue management/context switching by using a linked
22 list instead of an array (~5-10%).
23 - implement "watch" command in Coro::Debug shells.
24 - for fun, implement time-slicing as an example in the manpage.
25 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
26 Coro::Event, use the normal AnyEvent handling instead of dieing
27 (the same is true for Coro::Handle).
28 - properly document Coro::EV::timed_io_once.
29 - avoid unneccessary ->cancel calls in Coro::Handle.
30 - maybe make it work on mingw32 with win32 backend
31 (based on patch by Yasuhiro Matsumoto).
32
335.132 Fri May 29 09:00:39 CEST 2009
34 - do not keep a reference to the argument itself in
35 Coro::Semaphore::guard, as it could change later.
36 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
37 should support a prepare callback.
38
395.131 Mon Mar 16 23:20:37 CET 2009
40 - implement and document Coro->suspend, Coro->resume.
41 - fix Coro::Select implementation to not (often) close
42 the passed file descriptors (testcase provided by pippijn).
43
445.13 Mon Dec 15 21:51:42 CET 2008
45 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
46 - don't set diehook to C<undef> but instead to NULL, to avoid
47 spurious warnings.
48 - fix a lot of bugs in Coro::SemaphoreSet.
49 - Coro::SemaphoreSet will less often create a semaphore needlessly.
50 - add Coro::SemaphoreSet::count and wait methods.
51 - take advantage of the new Guard module.
52 - deprecate Coro::guard.
53 - try to fix the dreaded 01_unblock tests once more. I hate it when
54 testsuites need more fixing than the code they are supposed to test.
55 - croak in more cases when a required callback isn't resolvable.
56 - fix some minor issues in Coro::State->call/eval.
57 - use current coroutine context instead of a temporary one
58 when temporarily switching to another coroutine.
59 - do not call C-level on_destroy handlers during global destruction,
60 to avoid needless segfaults.
61
625.12 Sun Dec 7 13:30:38 CET 2008
63 - add default config for MirOS, which seems to be a bug-to-bug
64 compatible fork of openbsd ("world domination by releasing
65 openbsd cvs before the openbsd folks do it" or so :).
66 - free_padlist did destroy the names pad, not good, but didn't
67 seem to bother perl - this could fix issues such as eval ""
68 inside a function called from multiple coroutines.
69 - use a different method to detect destruction time.
70 - be more careful when freeing padlists just before global
71 destruction.
72 - fixed and expanded the call/cc example.
73 - renamed _terminate to _coro_run.
74 - new method Coro::Channel->shutdown.
75 - try pthreads on openbsd <4.4 (broken sigaltstack, will
76 pthreads fare better?).
77 - be less picky about destroying "a" running coroutine.
78
795.11 Tue Nov 25 21:49:05 CET 2008
80 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
81 working code with failed assertions, introducing more bugs than
82 it fixes, requiring elaborate workarounds :(
83
845.1 Mon Nov 24 08:54:59 CET 2008
85 - wrote a small introductory tutorial - Coro::Intro.
86 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
87 - Coro::Select did errornously dup the file descriptors
88 and didn't work with all AnyEvent backends.
89 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
90 LWP data transfers.
91 - disassociate c contexts from coro objects - this is agruably more
92 correct, but mostly allows sharing of cctxs between coro and state
93 objects, for added memory savings and speed increases.
94 - bumped $Coro::POOL_RSS up to 32kb by default.
95 - no longer set the optype to OP_CUSTOM, as B::* understandably
96 doesn't like this very much (and we *are* a type of entersub).
97 - implement state cloning, just to prove that call/cc can be done.
98 - automatically load Coro::AnyEvent in Coro::Handle.
99 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
100 do not have this method (and don't need it either).
101 - speed up generic anyevent methods in Coro::Handle by using rouse
102 callbacks.
103 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
104 others. It also makes the debugger happier, as you can trace
105 through the idle threads now.
106 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
107 - eg/event was pretty broken.
108 - better 5.8.6 compatibility.
109
1105.0 Thu Nov 20 10:35:05 CET 2008
111 - NEW ARCHITECTURE: use the latest 4.x version if you experience
112 stability issues.
113 - bump API version to 7 - all dependents must be recompiled.
114 - removed timed_* functions - they were not being used anyways
115 and should be replaced by a more generic mechanism -
116 and were annoying to support anyways :)
117 - removed SemaphoreSet's waiter method - use sem method instead.
118 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
119 - async_pool did free a scalar value twice
120 ("Attempt to unreference...").
121 - fix a longstanding bug where calling terminate on a coro that
122 was waiting for a semaphore that was just becoming available
123 would cause a deadlock (semaphore would get into a state where
124 it was available but waiters were still blocked).
125 - calling throw on a coroutine that is waiting for a semaphore will
126 no longer make it acquire the semaphore (and thus leak a count).
127 - perl's process emulation is now not even theoretically supported
128 anymore.
129 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
130 conversion of callback-style to blocking-style.
131 - new methods $coro->schedule_to and ->cede_to, to specifically
132 schedule or cede to a specific coroutine.
133 - new function Coro::Semaphore::wait.
134 - use named constants in Coro::Channel (Richard Hundt).
10 - directly patch the entersub opcode calling ssl functions (cede, 135 - directly patch the entersub opcode calling SLF functions (cede,
11 transfer and so on). this does speed up context switching by 136 transfer and so on). this does speed up context switching, but
12 up to 10%. More importanly, it frees us from the hardcoded 137 more importanly, it frees us from the hardcoded behaviour of
13 behaviour of entersub, so we might actually be able to return 138 entersub, so we might actually be able to return something from
14 something from those functions. 139 those functions and atcually create new ones.
140 - take advantage of __builtin_frame_address on gcc.
141 - expose THX in coroapi (not sure whether this was a wise decision,
142 as "threaded" perls are running at half speed anyways).
143 - implement execute_slf (schedule-like-function) interface that makes
144 it possible to implement schedule-like-functions in XS.
145 - use new SLF interface to massively speed up Coro::EV by roughly a
146 factor of two.
147 - used new SLF interface to massively speed up Coro::Semaphore by a
148 factor of three.
149 - used new SLF interface to speed up Coro::AIO by roughly a factor of
150 four and reduce its memory usage considerably.
151 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
152 for a nice speedup and vastly more correct behaviour. Also implement
153 a new method "sem" to get at the underlying semaphore object.
154 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
155 (in comparison) 20-40% speedup.
156 - used new SLF interface to reimplement Coro::Signal gaining
157 some unknown (because I was too lazy), but certain, speedup, and also
158 making signals reliable for the first time.
159 - used new SLF interface and other optimisations to speed up async_pool
160 by a factor of two. It also doesn't rely on perl's exception mechanism
161 to exit anymore. The overhead for terminating an async_pool, coro over
162 a normal async is now very small.
163 - sped up coroutine creation/destruction by 40%.
164 - forgot to include Coro/libcoro/README in the dist for all these years.
165 - work around a freebsd pthreads bug (manual testcancel is required as
166 pthread_cond_wait isn't a cancellation point on freebsd).
167 - use new rouse functions to speed up and simplify Coro::BDB.
168 - make "prefer perl native functions" work with threaded perls.
169 - condense Coro::Debug ps output, hint at v and w flags.
170 - (libcoro) lots of minor cleanups and portability improvements.
171
1724.914 Wed Nov 19 12:54:18 CET 2008
173 - fix a disastrous bug in the readline optimisation
174 introduced in 4.801.
175
1764.913 Sat Nov 15 07:58:28 CET 2008
177 - async_pool did free a scalar value twice
178 ("Attempt to unreference...").
15 179
164.912 Thu Nov 13 18:31:23 CET 2008 1804.912 Thu Nov 13 18:31:23 CET 2008
17 - minor cleanups. 181 - minor cleanups.
18 - use much larger stacks on linux and perl < 5.8.8. 182 - use much larger stacks on linux and perl < 5.8.8.
19 - Coro::Debug::new_unix_server did not unlink the socket 183 - Coro::Debug::new_unix_server did not unlink the socket

Diff Legend

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