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

Comparing Coro/Changes (file contents):
Revision 1.151 by root, Thu Nov 23 23:54:02 2006 UTC vs.
Revision 1.260 by root, Sat Oct 6 00:08:04 2007 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: check wether stack-sharing is still effective 3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO.
4TODO: more effective stack sharing by having one "default stack" for "simple" coroutines? 4TODO: maybe implement a default message channel, very much like Erlang
5TODO: verify exception frame changes and force stack copies? 5 (which is cool in a lot of important aspects (failures!),
6#2 0x00002ac4a63f2622 in api_schedule () at State.xs:895 6 but very lacking in others (higher level ipc)).
7#3 0x00002ac4a63f26a9 in XS_Coro_schedule (cv=0x911700) at State.xs:1095
8#4 0x00000000004941b8 in Perl_pp_entersub () at pp_hot.c:2869
9#5 0x0000000000489fee in Perl_runops_standard () at run.c:37
10#6 0x000000000041dec7 in Perl_call_sv (sv=0xbfd8ea0, flags=<value optimized out>) at perl.c:2806
11#7 0x00002ac4a6606ad3 in pe_event_invoke (ev=0x1ada6e0) at c/ev.c:308
12#8 0x00002ac4a6608878 in one_event (tm=60) at c/queue.c:91
13#9 0x00002ac4a6608f4f in XS_Event_one_event (cv=<value optimized out>) at c/queue.c:234
14#10 0x00000000004941b8 in Perl_pp_entersub () at pp_hot.c:2869
15#11 0x0000000000489fee in Perl_runops_standard () at run.c:37
16#12 0x00000000004c4287 in S_docatch (o=<value optimized out>) at pp_ctl.c:2566
17#13 0x0000000000489fee in Perl_runops_standard () at run.c:37
18#14 0x00000000004c4287 in S_docatch (o=<value optimized out>) at pp_ctl.c:2566
19#15 0x0000000000489fee in Perl_runops_standard () at run.c:37
20#16 0x000000000041e38c in Perl_call_sv (sv=<value optimized out>, flags=140) at perl.c:2806
21#17 0x00002ac4a63f1d72 in setup_coro (arg=0x881d80) at State.xs:620
22#18 0x00002ac4a63f28b7 in coro_init () at libcoro/coro.c:78
23#19 0x0000000000000000 in ?? ()
24 7
25 - minor documentation cleanup. 84.02
9 - fix a very minor per-coroutine memleak (a single codereference).
10 - fixed a bug where the currently in-use c context would be freed
11 prematurely (can happen only when programs change the stacksize
12 or use tracing).
13 - tracing can no longer keep a coro alive after it terminated.
14 - do static branch prediction in the common path for gcc. gives
15 about 2-5% speed improvement here.
16
174.01 Fri Oct 5 22:10:49 CEST 2007
18 - instead of recreating *a* standard output handle we simply
19 use STDOUT, which is faster and hopefully more robust.
20
214.0 Fri Oct 5 12:56:00 CEST 2007
22 - incompatibly changed Coro::Storable::freeze.
23 - major new feature: added Coro::Debug, for interactive coroutine
24 debugging, tracing and much more.
25 - major bug fix: unbelievable, but true: $_, $/ and many other
26 "saved" variables actually weren't being saved. This has been fixed,
27 of course, while increasing performance while losing all the save
28 flags.
29 - save flags are gone, and all the api functions dealing with them.
30 - added Coro::Semaphore::adjust.
31 - added Coro::Util::fork_eval.
32 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
33 - added Coro::killall.
34 - reduce initial stack sizes to allow for "micro-coroutines".
35 - better async_pool resource management, moved parts of async_pool
36 handling to XS (major speed improvement).
37 - actually croak before modifying important data structures.
38 - refuse to transfer while compiling.
39 - possibly support eval EXPR better now.
40 - enable assembly per default on linux+bsd x86+amd64.
41 - all internal members were renamed _something for easier subclassing.
42 - many minor tweaks.
43
443.63 Wed May 16 14:10:06 CEST 2007
45 - implement handcoded assembly for x86/amd64 SVR ABI.
46
473.62 Fri Apr 27 21:36:06 CEST 2007
48 - upgrade libcoro (which might set unwind info correctly).
49 - change default on linux to setjmp/longjmp.
50
513.61 Thu Apr 19 12:36:18 CEST 2007
52 - Coro::Storable caused an endless loop when thawing invalid
53 pst's sometimes.
54 - use a Semaphore in Coro::Storable, as Storable doesn't
55 seem to be reentrant (although it is documented to
56 be threadsafe...).
57 - fix Coro::Signal to bring back the original unreliable
58 but stateful semantics.
59 - fixed a lot of typos in Coro.pm (patch submitted by David
60 Steinbrunner, which applied flawlessly).
61
623.6 Sat Apr 14 17:13:31 CEST 2007
63 - added some bugfixes to get eg/myhttpd working again.
64 - added Coro::Storable for often-cede'ing freeze/thaw.
65 - try to do a clean exit when a coroutine calls exit
66 (EXPERIMENTAL).
67 - got rid of indirect call through _coro_init.
68 - updated the partly antique examples in eg/ to
69 work again and be a bit less magic, too.
70 - fixed Coro::Signal semantics to work as documented again.
71
723.55 Sun Mar 25 01:20:47 CET 2007
73 - add SAVE_DEFFH to save the default filehandle and enable
74 it by default.
75 - finally move socket-operations from Coro::Socket to Coro::Handle
76 to be able to unblock foreign sockets.
77 - add Coro::State::save_also and guarded_save.
78 - add count accessor to Coro::Semaphore.
79 - add Coro::State::cctx_stacksize.
80 - just for the fun of it, do not rely on implicit context,
81 which can dramatically improve performance, but people
82 using windows-process-emulation perls do not care much about
83 performance.
84
853.51 Sun Mar 4 14:18:04 CET 2007
86 - fixed a problem when you weakened references to Coro::State's
87 (patch by Yuval Kogman).
88
893.501 Wed Feb 28 12:44:07 CET 2007
90 - rename some global symbols as macosx from hell redefines
91 them without asking.
92
933.5 Tue Feb 13 20:22:53 CET 2007
94 - do AnyEvent model detection earlier, avoiding problems
95 caused by first using AnyEvent and later Coro::Event.
96 - implement and document Coro::Event event objects.
97 - fix a potential problem in Coro::Event causing crashes.
98 - initialise PL_comppad when creating a new coroutine,
99 avoids crashes on early coro destruction.
100
1013.41 Mon Jan 22 19:19:49 CET 2007
102 - readline on Coro::Handle did not support undefined $/,
103 nor did it deliver partial lines on EOF or error.
104 - implement malloc fallback for stack allocation because
105 stupid broken idiotic OSX has a stupid broken
106 idiotic fits-the-whole-os mmap "implementation" and
107 my dick feels longer if Coro is portable even to
108 obsolete platforms.
109
1103.4 Fri Jan 19 21:52:54 CET 2007
111 - remove t/09_timer.t, as it isn't really testing much
112 but was rather flaky in practise.
113 - async_pool coro would keep arguments and callback alive until
114 it was reused.
115 - cancellation of a coroutine could cause spurious idle calls
116 in cede_notself.
117
1183.3 Sat Jan 6 03:45:00 CET 2007
119 - implement $coro->on_destroy.
120 - Coro::Event blocking semantics have been changed,
121 documented and - hopefully - improved.
122 - fix nice adding, not subtracting, from priority.
123 - fix ->prio and api_is_ready (patch by Mark Hinds).
124 - fixed an assert ("... == prev__cctx->idle_te")
125 that could errronously trigger.
126 - fix various large and small memleaks.
127 - use a (hopefully) more stable cancel implementation
128 that immediately frees the coroutine data.
129 - cede/cede_notself return a status now.
130 - added Coro::guard function.
131 - added a global coroutine pool for jobs (on my machine,
132 I can create and execute 48k simple coros/s with async,
133 and 128k coros with async_pool).
134 - Coro::AIO now uses the coroutine priority as io priority.
135
1363.2 Fri Dec 22 05:07:09 CET 2006
137 - improve portability to slightly older perls.
138 - use cleaner coroutine destruction.
139 - simplify configuration for users.
140 - optionally (unrecommended) prefer perl functions over
141 their coro replacements.
142
1433.11 Tue Dec 5 13:11:24 CET 2006
144 - fixed some bogus assert's, but as perl.h disables assert even
145 without NDEBUG (thank you very much), not too many people should
146 notice (that did include myself). Andreas König noticed, though :)
147 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
148 it. Also noticed by Andreas König :)
149 - save/restore tainted status.
150 - verified to pass the testsuite on my 5.9.5.
151
1523.1 Mon Dec 4 23:03:40 CET 2006
153 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
154 - incompatible change: transfer flags are now per-state.
155 - give a better error message on deadlock.
156 - document Coro::nready.
157 - enhanced testsuite.
158
1593.01 Sun Dec 3 23:47:42 CET 2006
160 - forgot to include Coro::Timer.
161
1623.0 Sun Dec 3 22:57:25 CET 2006
163 - the "FINALLY COMPLETELY STABLE" release (coming soon:
164 the "FAMOUS LAST WORDS" release).
165 - implement a new stack sharing algorithm, which uses a stack
166 pool (size currently hardcoded).
167 - make stack sharing mandatory (it no longer uses a heuristic).
168 - eval/die no longer cause weird problems under heavy use.
169 - Coro::Event could cause livelocks if it was used but
170 no Coro::Event watchers were used.
171 - Coro::Event now uses asynccheck as prepare does not
172 check for changed watchers.
173 - Coro::Event allows multiple waiting coros and will wake up one
174 per event.
175 - Coro::Event should be cleaner and more efficient now.
176 - new utility function Coro::unblock_sub.
177 - document the sad fact that Event is no longer reentrant.
178 - putting a coroutine into the ready queue twice could under
179 some circumstances lead to stack corruption.
180 - minor incompatibility: subclassing Coro::State is supported
181 directly now without going through a _coro_state member.
182 - state/coro switching is much faster now.
26 - verys minor optimisations and cleanups. 183 - very minor optimisations and code/documentation cleanup.
184 - avoid problems due to compiler inlining.
185 - removed timers from Coro::Timer -> use AnyEvent instead.
186 - replaced Coro::idle coroutine by (cleaner) idle handler.
187 - updated to newest libcoro.
188 - implement enhanced support for valgrind.
189 - implement is_ready and return value for ready.
190 - removed Coro::Cont, it was a misnomer (it's generators),
191 and it was somewhat annoying to get it right. Will come back
192 if somebody asks for it :)
193 - many ->wait methods and Coro::Event could return
194 spuriously without the event having happened.
27 195
282.5 Tue Nov 7 12:22:33 CET 2006 1962.5 Tue Nov 7 12:22:33 CET 2006
29 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket 197 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
30 use AnyEvent, moved them to Coro/. 198 use AnyEvent, moved them to Coro/.
31 - added Coro::LWP which contains all the uglyness required to 199 - added Coro::LWP which contains all the uglyness required to

Diff Legend

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