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

Comparing Coro/Changes (file contents):
Revision 1.402 by root, Tue Nov 11 03:26:58 2008 UTC vs.
Revision 1.473 by root, Thu Dec 4 17:29:40 2008 UTC

… …
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)).
7 7
85.12
9 - add default config for MirOS, which seems to be a bug-to-bug
10 compatible fork of openbsd ("world domination by releasing
11 openbsd cvs before the openbsd folks do it" or so :).
12 - free_padlist did destroy the names pad, not good, but didn't
13 seem to bother perl - this could fix issues such as eval ""
14 inside a function called form multiple coroutines.
15 - use a different method to detect destruction time.
16 - be more careful when freeing padlists just before global
17 destruction.
18 - fixed and expanded the call/cc example.
19 - renamed _terminate to _coro_run.
20 - new method Coro::Channel->shutdown.
21 - try pthreads on openbsd <4.4 (broken sigaltstack, will
22 pthreads fare better?).
23 - be less picky about destroying "a" running coroutine.
24
255.11 Tue Nov 25 21:49:05 CET 2008
26 - DEBUGGING in 5.10.0 is a mess: it constantly flags perfectly
27 working code with failed assertions, introducing more bugs than
28 it fixes, requiring elaborate workarounds :(
29
305.1 Mon Nov 24 08:54:59 CET 2008
31 - wrote a small introductory tutorial - Coro::Intro.
32 - convert Coro::Timer, Coro::Select and Coro::Util to rouse API.
33 - Coro::Select did errornously dup the file descriptors
34 and didn't work with all AnyEvent backends.
35 - Coro::Select wasn't imported correctly form Coro::LWP, causing blocking
36 LWP data transfers.
37 - disassociate c contexts from coro objects - this is agruably more
38 correct, but mostly allows sharing of cctxs between coro and state
39 objects, for added memory savings and speed increases.
40 - bumped $Coro::POOL_RSS up to 32kb by default.
41 - no longer set the optype to OP_CUSTOM, as B::* understandably
42 doesn't like this very much (and we *are* a type of entersub).
43 - implement state cloning, just to prove that call/cc can be done.
44 - automatically load Coro::AnyEvent in Coro::Handle.
45 - wrap ->cancel calls in eval inside Coro::Handle as EV watchers
46 do not have this method (and don't need it either).
47 - speed up generic anyevent methods in Coro::Handle by using rouse
48 callbacks.
49 - allow coroutines in $Coro::IDLE, speeding up Coro::AnyEvent and
50 others. It also makes the debugger happier, as you can trace
51 through the idle threads now.
52 - add comppad_name* and hints ($^H, %^H) to per-thread variables.
53 - eg/event was pretty broken.
54 - better 5.8.6 compatibility.
55
565.0 Thu Nov 20 10:35:05 CET 2008
57 - NEW ARCHITECTURE: use the latest 4.x version if you experience
58 stability issues.
59 - bump API version to 7 - all dependents must be recompiled.
60 - removed timed_* functions - they were not being used anyways
61 and should be replaced by a more generic mechanism -
62 and were annoying to support anyways :)
63 - removed SemaphoreSet's waiter method - use sem method instead.
64 - Coro::Semaphore->adjust didn't correctly wake up enough waiters.
65 - async_pool did free a scalar value twice
66 ("Attempt to unreference...").
67 - fix a longstanding bug where calling terminate on a coro that
68 was waiting for a semaphore that was just becoming available
69 would cause a deadlock (semaphore would get into a state where
70 it was available but waiters were still blocked).
71 - calling throw on a coroutine that is waiting for a semaphore will
72 no longer make it acquire the semaphore (and thus leak a count).
73 - perl's process emulation is now not even theoretically supported
74 anymore.
75 - new functions Coro::rouse_cb and Coro::rouse_wait for easier
76 conversion of callback-style to blocking-style.
77 - new methods $coro->schedule_to and ->cede_to, to specifically
78 schedule or cede to a specific coroutine.
79 - new function Coro::Semaphore::wait.
80 - use named constants in Coro::Channel (Richard Hundt).
81 - directly patch the entersub opcode calling SLF functions (cede,
82 transfer and so on). this does speed up context switching, but
83 more importanly, it frees us from the hardcoded behaviour of
84 entersub, so we might actually be able to return something from
85 those functions and atcually create new ones.
86 - take advantage of __builtin_frame_address on gcc.
87 - expose THX in coroapi (not sure whether this was a wise decision,
88 as "threaded" perls are running at half speed anyways).
89 - implement execute_slf (schedule-like-function) interface that makes
90 it possible to implement schedule-like-functions in XS.
91 - use new SLF interface to massively speed up Coro::EV by roughly a
92 factor of two.
93 - used new SLF interface to massively speed up Coro::Semaphore by a
94 factor of three.
95 - used new SLF interface to speed up Coro::AIO by roughly a factor of
96 four and reduce its memory usage considerably.
97 - implement Coro::SemaphoreSet purely in terms of Coro::Semaphore,
98 for a nice speedup and vastly more correct behaviour. Also implement
99 a new method "sem" to get at the underlying semaphore object.
100 - implement Coro::Channel in terms of Coro::Semaphore, for a moderate
101 (in comparison) 20-40% speedup.
102 - used new SLF interface to reimplement Coro::Signal gaining
103 some unknown (because I was too lazy), but certain, speedup, and also
104 making signals reliable for the first time.
105 - used new SLF interface and other optimisations to speed up async_pool
106 by a factor of two. It also doesn't rely on perl's exception mechanism
107 to exit anymore. The overhead for terminating an async_pool, coro over
108 a normal async is now very small.
109 - sped up coroutine creation/destruction by 40%.
110 - forgot to include Coro/libcoro/README in the dist for all these years.
111 - work around a freebsd pthreads bug (manual testcancel is required as
112 pthread_cond_wait isn't a cancellation point on freebsd).
113 - use new rouse functions to speed up and simplify Coro::BDB.
114 - make "prefer perl native functions" work with threaded perls.
115 - condense Coro::Debug ps output, hint at v and w flags.
116 - (libcoro) lots of minor cleanups and portability improvements.
117
1184.914 Wed Nov 19 12:54:18 CET 2008
119 - fix a disastrous bug in the readline optimisation
120 introduced in 4.801.
121
1224.913 Sat Nov 15 07:58:28 CET 2008
123 - async_pool did free a scalar value twice
124 ("Attempt to unreference...").
125
1264.912 Thu Nov 13 18:31:23 CET 2008
127 - minor cleanups.
128 - use much larger stacks on linux and perl < 5.8.8.
129 - Coro::Debug::new_unix_server did not unlink the socket
130 when destroyed.
131
84.901 Tue Nov 11 04:26:17 CET 2008 1324.911 Tue Nov 11 04:26:17 CET 2008
9 - "port" to threaded perls. 133 - "port" to threaded perls.
10 134
114.91 Mon Nov 10 05:36:38 CET 2008 1354.91 Mon Nov 10 05:36:38 CET 2008
12 - the ->throw exception object no longer leaks. 136 - the ->throw exception object no longer leaks.
13 - creating a new cctx leaked a scopestack entry (memleak). 137 - creating a new cctx leaked a scopestack entry (memleak).

Diff Legend

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