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

Comparing Coro/Changes (file contents):
Revision 1.107 by root, Sat Aug 20 01:09:19 2005 UTC vs.
Revision 1.215 by root, Mon Mar 19 15:50:48 2007 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: better (earlier) error messages on broken transfer's
4TODO: destroy gets only called on cancel: fix(?)
5
63.55
7 - add SAVE_DEFFH to save the default filehandle and enable
8 it by default.
9 - finally move socket-operations from Coro::Socket to Coro::Handle
10 to be able to unblock foreign sockets.
11 - add Coro::State::save_also and guarded_save.
12 - add count accessor to Coro::Semaphore.
13 - add Coro::State::cctx_stacksize.
14 - just for the fun of it, do not rely on implicit context,
15 which can dramatically improve performance, but people
16 using windows-process-emulation perls do not care much about
17 performance.
18
193.51 Sun Mar 4 14:18:04 CET 2007
20 - fixed a problem when you weakened references to Coro::State's
21 (patch by Yuval Kogman).
22
233.501 Wed Feb 28 12:44:07 CET 2007
24 - rename some global symbols as macosx from hell redefines
25 them without asking.
26
273.5 Tue Feb 13 20:22:53 CET 2007
28 - do AnyEvent model detection earlier, avoiding problems
29 caused by first using AnyEvent and later Coro::Event.
30 - implement and document Coro::Event event objects.
31 - fix a potential problem in Coro::Event causing crashes.
32 - initialise PL_comppad when creating a new coroutine,
33 avoids crashes on early coro destruction.
34
353.41 Mon Jan 22 19:19:49 CET 2007
36 - readline on Coro::Handle did not support undefined $/,
37 nor did it deliver partial lines on EOF or error.
38 - implement malloc fallback for stack allocation because
39 stupid broken idiotic OSX has a stupid broken
40 idiotic fits-the-whole-os mmap "implementation" and
41 my dick feels longer if Coro is portable even to
42 obsolete platforms.
43
443.4 Fri Jan 19 21:52:54 CET 2007
45 - remove t/09_timer.t, as it isn't really testing much
46 but was rather flaky in practise.
47 - async_pool coro would keep arguments and callback alive until
48 it was reused.
49 - cancellation of a coroutine could cause spurious idle calls
50 in cede_notself.
51
523.3 Sat Jan 6 03:45:00 CET 2007
53 - implement $coro->on_destroy.
54 - Coro::Event blocking semantics have been changed,
55 documented and - hopefully - improved.
56 - fix nice adding, not subtracting, from priority.
57 - fix ->prio and api_is_ready (patch by Mark Hinds).
58 - fixed an assert ("... == prev__cctx->idle_te")
59 that could errronously trigger.
60 - fix various large and small memleaks.
61 - use a (hopefully) more stable cancel implementation
62 that immediately frees the coroutine data.
63 - cede/cede_notself return a status now.
64 - added Coro::guard function.
65 - added a global coroutine pool for jobs (on my machine,
66 I can create and execute 48k simple coros/s with async,
67 and 128k coros with async_pool).
68 - Coro::AIO now uses the coroutine priority as io priority.
69
703.2 Fri Dec 22 05:07:09 CET 2006
71 - improve portability to slightly older perls.
72 - use cleaner coroutine destruction.
73 - simplify configuration for users.
74 - optionally (unrecommended) prefer perl functions over
75 their coro replacements.
76
773.11 Tue Dec 5 13:11:24 CET 2006
78 - fixed some bogus assert's, but as perl.h disables assert even
79 without NDEBUG (thank you very much), not too many people should
80 notice (that did include myself). Andreas König noticed, though :)
81 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
82 it. Also noticed by Andreas König :)
83 - save/restore tainted status.
84 - verified to pass the testsuite on my 5.9.5.
85
863.1 Mon Dec 4 23:03:40 CET 2006
87 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
88 - incompatible change: transfer flags are now per-state.
89 - give a better error message on deadlock.
90 - document Coro::nready.
91 - enhanced testsuite.
92
933.01 Sun Dec 3 23:47:42 CET 2006
94 - forgot to include Coro::Timer.
95
963.0 Sun Dec 3 22:57:25 CET 2006
97 - the "FINALLY COMPLETELY STABLE" release (coming soon:
98 the "FAMOUS LAST WORDS" release).
99 - implement a new stack sharing algorithm, which uses a stack
100 pool (size currently hardcoded).
101 - make stack sharing mandatory (it no longer uses a heuristic).
102 - eval/die no longer cause weird problems under heavy use.
103 - Coro::Event could cause livelocks if it was used but
104 no Coro::Event watchers were used.
105 - Coro::Event now uses asynccheck as prepare does not
106 check for changed watchers.
107 - Coro::Event allows multiple waiting coros and will wake up one
108 per event.
109 - Coro::Event should be cleaner and more efficient now.
110 - new utility function Coro::unblock_sub.
111 - document the sad fact that Event is no longer reentrant.
112 - putting a coroutine into the ready queue twice could under
113 some circumstances lead to stack corruption.
114 - minor incompatibility: subclassing Coro::State is supported
115 directly now without going through a _coro_state member.
116 - state/coro switching is much faster now.
117 - very minor optimisations and code/documentation cleanup.
118 - avoid problems due to compiler inlining.
119 - removed timers from Coro::Timer -> use AnyEvent instead.
120 - replaced Coro::idle coroutine by (cleaner) idle handler.
121 - updated to newest libcoro.
122 - implement enhanced support for valgrind.
123 - implement is_ready and return value for ready.
124 - removed Coro::Cont, it was a misnomer (it's generators),
125 and it was somewhat annoying to get it right. Will come back
126 if somebody asks for it :)
127 - many ->wait methods and Coro::Event could return
128 spuriously without the event having happened.
129
1302.5 Tue Nov 7 12:22:33 CET 2006
131 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
132 use AnyEvent, moved them to Coro/.
133 - added Coro::LWP which contains all the uglyness required to
134 make LWP non-blocking.
135 - should work with perl 5.9.x now (Andreas König made me do it).
136 - fixed another bug in Coro::Select when the timeout was undef.
137 - reuse PL_start_env for all coros, saves some memory per coroutine.
138 - manage PL_top_env differently, hopefully to avoid panic: top_env.
139 - timeout argument was not properly used in Coro::Socket.
140 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
141 - emulate undocumented(!) functionality of IO::Socket required
142 by LWP(!!).
143 - updated eg/lwp to work with newer lwp's.
144 - remove "FATAL: uncaught exception" prefix. Coroutines that die
145 kill the whole process, just as exceptions in the main "coroutine"
146 did already.
147
1482.1 Wed Nov 1 23:01:13 CET 2006
149 - fix a long-standing bug in Coro::Select where select with
150 zero timeout would instead change the current default filehandle.
151 - use a simpler and hopefully more robust way to clone padlists
152 (uses less memory and a perl function instead of our own).
153 - coro can now create a stack guard area on many architectures.
154 - Coro::AIO properly reexports additional functions from IO::AIO.
155 - updated libcoro with a workaround for OS X,
156 pach and testing by Michael Schwern.
157
1582.0 Tue Oct 24 05:47:17 CEST 2006
159 - support additional aio requests in Coro::AIO.
160
1611.9 (never properly released due to a glitch)
162
1631.8 Thu Feb 2 00:59:06 CET 2006
164 - applied suggested patch by SAMV to avoid problems during stupid
165 mark & sweep gc run.
166 - applied patch by Scott Walters for 5.9.3 compatibility.
167
1681.7 Tue Dec 27 01:41:58 CET 2005
169 - added Coro::AIO, a thin wrapper around IO::AIO.
170 - improved Makefile.PL explanations.
171
1721.6 Mon Dec 12 21:30:05 CET 2005
173 - additionally save PM_curpm might fix as-of-yet
174 unseen problems with regex matches being attributed to the
175 wrong package.
176 - add t/10_bugs.t, which currently checks against imho broken
177 perls that use 0.26MB of stack space per Perl_magic_get
178 invocation (newer linuxes) as opposed to the 0.0002MB perl
179 normally uses.
180 - make stacksize configurable for the ultimate debian experience.
181
1821.51 Mon Dec 12 18:48:36 CET 2005
183 - remove debugging warn() accidentally left in Coro::Select.
184
1851.5 Tue Nov 29 13:32:44 CET 2005
186 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
187 - fix Coro::Select 'select'.
188 - strict'ify some modules.
189 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
190 - use XSLoader in selected modules.
191 - remove some 5.6 compatibility cruft.
192
1931.4 Tue Sep 6 00:11:05 CEST 2005
194 - libcoro did not take into account the trampoline on amd64,
195 when 'l'inux method was used.
196
1971.31 Tue Aug 30 23:31:33 CEST 2005
198 - some portability fixes/workarounds.
199
31.3 Sat Aug 20 03:08:56 CEST 2005 2001.3 Sat Aug 20 03:08:56 CEST 2005
201 - no code changes. module seems to work fine.
4 - improve Coro::State docs, remove reference to nonexisting 202 - improve Coro::State docs, remove reference to nonexisting
5 Coro::State::flush method. 203 Coro::State::flush method.
6 - no longer autodetect windows, present it as an option instead. 204 - no longer autodetect windows, present it as an option instead.
7 205
81.2 Mon May 16 02:00:55 CEST 2005 2061.2 Mon May 16 02:00:55 CEST 2005
151.11 Thu Mar 3 18:00:52 CET 2005 2131.11 Thu Mar 3 18:00:52 CET 2005
16 - change of contact address. 214 - change of contact address.
17 215
181.1 Tue Feb 22 20:51:16 CET 2005 2161.1 Tue Feb 22 20:51:16 CET 2005
19 - support [l]inux method on amd64. 217 - support [l]inux method on amd64.
20 - make some context switches while compiling/eval'ing work. 218 - allow some context switches while compiling/eval'ing.
21 219
221.0 Tue Aug 10 03:47:27 CEST 2004 2201.0 Tue Aug 10 03:47:27 CEST 2004
23 - Event 0.89 fixes the bug again. 221 - Event 0.89 fixes the bug again.
24 - fix the remaining(??) memleak. I didn't free the cache-AV for 222 - fix the remaining(??) memleak. I didn't free the cache-AV for
25 padlists, so one AV per created anon-sub leaked. 223 padlists, so one AV per created anon-sub leaked.

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines