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

Comparing Coro/Changes (file contents):
Revision 1.117 by root, Mon Dec 12 19:43:41 2005 UTC vs.
Revision 1.271 by root, Tue Oct 9 22:29:26 2007 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO.
4TODO: maybe implement a default message channel, very much like Erlang
5 (which is cool in a lot of important aspects (failures!),
6 but very lacking in others (higher level ipc)).
7
84.1
9 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
10 be local to each coro (see Coro::State).
11 - possibly bring back 5.10 compatibility (untested).
12 - work around stupid (and wrong) warning on 5.10 :(.
13 - overlay the saved state over the context stack. This saves
14 a few hundred bytes per coroutine on average and also
15 speeds up context switching a bit.
16 - further tune default stack sizes.
17 - (more) correctly calculate stack usage in coro_rss.
18 - Coro::Storable::blocking_* did not properly lock
19 resulting in races between coroutines.
20 - added Coro::Storable::guard.
21 - stopping to trace a coroutine could destroy the cctx of
22 an unrelated coroutine.
23 - explain the relationship between Perl and C coroutines in
24 more detail in Coro::State.
25 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
26 causing a fork per resolve. This also affected Coro::Socket.
27 - sped up Coro::Event next event handling by 15%, uses less memory.
28
294.03 Sat Oct 6 21:24:00 CEST 2007
30 - added Coro::throw method.
31 - minor code cleanups.
32
334.02 Sat Oct 6 02:36:47 CEST 2007
34 - fix a very minor per-coroutine memleak (a single codereference).
35 - fixed a bug where the currently in-use c context would be freed
36 prematurely (can happen only when programs change the stacksize
37 or use tracing).
38 - tracing can no longer keep a coro alive after it terminated.
39 - do static branch prediction in the common path for gcc. gives
40 about 2-5% speed improvement here.
41
424.01 Fri Oct 5 22:10:49 CEST 2007
43 - instead of recreating *a* standard output handle we simply
44 use STDOUT, which is faster and hopefully more robust.
45
464.0 Fri Oct 5 12:56:00 CEST 2007
47 - incompatibly changed Coro::Storable::freeze.
48 - major new feature: added Coro::Debug, for interactive coroutine
49 debugging, tracing and much more.
50 - major bug fix: unbelievable, but true: $_, $/ and many other
51 "saved" variables actually weren't being saved. This has been fixed,
52 of course, while increasing performance while losing all the save
53 flags.
54 - save flags are gone, and all the api functions dealing with them.
55 - added Coro::Semaphore::adjust.
56 - added Coro::Util::fork_eval.
57 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
58 - added Coro::killall.
59 - reduce initial stack sizes to allow for "micro-coroutines".
60 - better async_pool resource management, moved parts of async_pool
61 handling to XS (major speed improvement).
62 - actually croak before modifying important data structures.
63 - refuse to transfer while compiling.
64 - possibly support eval EXPR better now.
65 - enable assembly per default on linux+bsd x86+amd64.
66 - all internal members were renamed _something for easier subclassing.
67 - many minor tweaks.
68
693.63 Wed May 16 14:10:06 CEST 2007
70 - implement handcoded assembly for x86/amd64 SVR ABI.
71
723.62 Fri Apr 27 21:36:06 CEST 2007
73 - upgrade libcoro (which might set unwind info correctly).
74 - change default on linux to setjmp/longjmp.
75
763.61 Thu Apr 19 12:36:18 CEST 2007
77 - Coro::Storable caused an endless loop when thawing invalid
78 pst's sometimes.
79 - use a Semaphore in Coro::Storable, as Storable doesn't
80 seem to be reentrant (although it is documented to
81 be threadsafe...).
82 - fix Coro::Signal to bring back the original unreliable
83 but stateful semantics.
84 - fixed a lot of typos in Coro.pm (patch submitted by David
85 Steinbrunner, which applied flawlessly).
86
873.6 Sat Apr 14 17:13:31 CEST 2007
88 - added some bugfixes to get eg/myhttpd working again.
89 - added Coro::Storable for often-cede'ing freeze/thaw.
90 - try to do a clean exit when a coroutine calls exit
91 (EXPERIMENTAL).
92 - got rid of indirect call through _coro_init.
93 - updated the partly antique examples in eg/ to
94 work again and be a bit less magic, too.
95 - fixed Coro::Signal semantics to work as documented again.
96
973.55 Sun Mar 25 01:20:47 CET 2007
98 - add SAVE_DEFFH to save the default filehandle and enable
99 it by default.
100 - finally move socket-operations from Coro::Socket to Coro::Handle
101 to be able to unblock foreign sockets.
102 - add Coro::State::save_also and guarded_save.
103 - add count accessor to Coro::Semaphore.
104 - add Coro::State::cctx_stacksize.
105 - just for the fun of it, do not rely on implicit context,
106 which can dramatically improve performance, but people
107 using windows-process-emulation perls do not care much about
108 performance.
109
1103.51 Sun Mar 4 14:18:04 CET 2007
111 - fixed a problem when you weakened references to Coro::State's
112 (patch by Yuval Kogman).
113
1143.501 Wed Feb 28 12:44:07 CET 2007
115 - rename some global symbols as macosx from hell redefines
116 them without asking.
117
1183.5 Tue Feb 13 20:22:53 CET 2007
119 - do AnyEvent model detection earlier, avoiding problems
120 caused by first using AnyEvent and later Coro::Event.
121 - implement and document Coro::Event event objects.
122 - fix a potential problem in Coro::Event causing crashes.
123 - initialise PL_comppad when creating a new coroutine,
124 avoids crashes on early coro destruction.
125
1263.41 Mon Jan 22 19:19:49 CET 2007
127 - readline on Coro::Handle did not support undefined $/,
128 nor did it deliver partial lines on EOF or error.
129 - implement malloc fallback for stack allocation because
130 stupid broken idiotic OSX has a stupid broken
131 idiotic fits-the-whole-os mmap "implementation" and
132 my dick feels longer if Coro is portable even to
133 obsolete platforms.
134
1353.4 Fri Jan 19 21:52:54 CET 2007
136 - remove t/09_timer.t, as it isn't really testing much
137 but was rather flaky in practise.
138 - async_pool coro would keep arguments and callback alive until
139 it was reused.
140 - cancellation of a coroutine could cause spurious idle calls
141 in cede_notself.
142
1433.3 Sat Jan 6 03:45:00 CET 2007
144 - implement $coro->on_destroy.
145 - Coro::Event blocking semantics have been changed,
146 documented and - hopefully - improved.
147 - fix nice adding, not subtracting, from priority.
148 - fix ->prio and api_is_ready (patch by Mark Hinds).
149 - fixed an assert ("... == prev__cctx->idle_te")
150 that could errronously trigger.
151 - fix various large and small memleaks.
152 - use a (hopefully) more stable cancel implementation
153 that immediately frees the coroutine data.
154 - cede/cede_notself return a status now.
155 - added Coro::guard function.
156 - added a global coroutine pool for jobs (on my machine,
157 I can create and execute 48k simple coros/s with async,
158 and 128k coros with async_pool).
159 - Coro::AIO now uses the coroutine priority as io priority.
160
1613.2 Fri Dec 22 05:07:09 CET 2006
162 - improve portability to slightly older perls.
163 - use cleaner coroutine destruction.
164 - simplify configuration for users.
165 - optionally (unrecommended) prefer perl functions over
166 their coro replacements.
167
1683.11 Tue Dec 5 13:11:24 CET 2006
169 - fixed some bogus assert's, but as perl.h disables assert even
170 without NDEBUG (thank you very much), not too many people should
171 notice (that did include myself). Andreas König noticed, though :)
172 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
173 it. Also noticed by Andreas König :)
174 - save/restore tainted status.
175 - verified to pass the testsuite on my 5.9.5.
176
1773.1 Mon Dec 4 23:03:40 CET 2006
178 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
179 - incompatible change: transfer flags are now per-state.
180 - give a better error message on deadlock.
181 - document Coro::nready.
182 - enhanced testsuite.
183
1843.01 Sun Dec 3 23:47:42 CET 2006
185 - forgot to include Coro::Timer.
186
1873.0 Sun Dec 3 22:57:25 CET 2006
188 - the "FINALLY COMPLETELY STABLE" release (coming soon:
189 the "FAMOUS LAST WORDS" release).
190 - implement a new stack sharing algorithm, which uses a stack
191 pool (size currently hardcoded).
192 - make stack sharing mandatory (it no longer uses a heuristic).
193 - eval/die no longer cause weird problems under heavy use.
194 - Coro::Event could cause livelocks if it was used but
195 no Coro::Event watchers were used.
196 - Coro::Event now uses asynccheck as prepare does not
197 check for changed watchers.
198 - Coro::Event allows multiple waiting coros and will wake up one
199 per event.
200 - Coro::Event should be cleaner and more efficient now.
201 - new utility function Coro::unblock_sub.
202 - document the sad fact that Event is no longer reentrant.
203 - putting a coroutine into the ready queue twice could under
204 some circumstances lead to stack corruption.
205 - minor incompatibility: subclassing Coro::State is supported
206 directly now without going through a _coro_state member.
207 - state/coro switching is much faster now.
208 - very minor optimisations and code/documentation cleanup.
209 - avoid problems due to compiler inlining.
210 - removed timers from Coro::Timer -> use AnyEvent instead.
211 - replaced Coro::idle coroutine by (cleaner) idle handler.
212 - updated to newest libcoro.
213 - implement enhanced support for valgrind.
214 - implement is_ready and return value for ready.
215 - removed Coro::Cont, it was a misnomer (it's generators),
216 and it was somewhat annoying to get it right. Will come back
217 if somebody asks for it :)
218 - many ->wait methods and Coro::Event could return
219 spuriously without the event having happened.
220
2212.5 Tue Nov 7 12:22:33 CET 2006
222 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
223 use AnyEvent, moved them to Coro/.
224 - added Coro::LWP which contains all the uglyness required to
225 make LWP non-blocking.
226 - should work with perl 5.9.x now (Andreas König made me do it).
227 - fixed another bug in Coro::Select when the timeout was undef.
228 - reuse PL_start_env for all coros, saves some memory per coroutine.
229 - manage PL_top_env differently, hopefully to avoid panic: top_env.
230 - timeout argument was not properly used in Coro::Socket.
231 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
232 - emulate undocumented(!) functionality of IO::Socket required
233 by LWP(!!).
234 - updated eg/lwp to work with newer lwp's.
235 - remove "FATAL: uncaught exception" prefix. Coroutines that die
236 kill the whole process, just as exceptions in the main "coroutine"
237 did already.
238
2392.1 Wed Nov 1 23:01:13 CET 2006
240 - fix a long-standing bug in Coro::Select where select with
241 zero timeout would instead change the current default filehandle.
242 - use a simpler and hopefully more robust way to clone padlists
243 (uses less memory and a perl function instead of our own).
244 - coro can now create a stack guard area on many architectures.
245 - Coro::AIO properly reexports additional functions from IO::AIO.
246 - updated libcoro with a workaround for OS X,
247 pach and testing by Michael Schwern.
248
2492.0 Tue Oct 24 05:47:17 CEST 2006
250 - support additional aio requests in Coro::AIO.
251
2521.9 (never properly released due to a glitch)
253
2541.8 Thu Feb 2 00:59:06 CET 2006
255 - applied suggested patch by SAMV to avoid problems during stupid
256 mark & sweep gc run.
257 - applied patch by Scott Walters for 5.9.3 compatibility.
258
2591.7 Tue Dec 27 01:41:58 CET 2005
260 - added Coro::AIO, a thin wrapper around IO::AIO.
261 - improved Makefile.PL explanations.
262
2631.6 Mon Dec 12 21:30:05 CET 2005
3 - additionally save PM_curpm might fix as-of-yet 264 - additionally save PM_curpm might fix as-of-yet
4 unseen problems with regex matches being attributes to the 265 unseen problems with regex matches being attributed to the
5 wrong package. 266 wrong package.
267 - add t/10_bugs.t, which currently checks against imho broken
268 perls that use 0.26MB of stack space per Perl_magic_get
269 invocation (newer linuxes) as opposed to the 0.0002MB perl
270 normally uses.
271 - make stacksize configurable for the ultimate debian experience.
6 272
71.51 Mon Dec 12 18:48:36 CET 2005 2731.51 Mon Dec 12 18:48:36 CET 2005
8 - remove debugging warn() accidentally left in Coro::Select. 274 - remove debugging warn() accidentally left in Coro::Select.
9 275
101.5 Tue Nov 29 13:32:44 CET 2005 2761.5 Tue Nov 29 13:32:44 CET 2005

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines