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

Comparing Coro/Changes (file contents):
Revision 1.94 by pcg, Fri May 14 13:25:08 2004 UTC vs.
Revision 1.268 by root, Mon Oct 8 01:23:14 2007 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
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
244.03 Sat Oct 6 21:24:00 CEST 2007
25 - added Coro::throw method.
26 - minor code cleanups.
27
284.02 Sat Oct 6 02:36:47 CEST 2007
29 - fix a very minor per-coroutine memleak (a single codereference).
30 - fixed a bug where the currently in-use c context would be freed
31 prematurely (can happen only when programs change the stacksize
32 or use tracing).
33 - tracing can no longer keep a coro alive after it terminated.
34 - do static branch prediction in the common path for gcc. gives
35 about 2-5% speed improvement here.
36
374.01 Fri Oct 5 22:10:49 CEST 2007
38 - instead of recreating *a* standard output handle we simply
39 use STDOUT, which is faster and hopefully more robust.
40
414.0 Fri Oct 5 12:56:00 CEST 2007
42 - incompatibly changed Coro::Storable::freeze.
43 - major new feature: added Coro::Debug, for interactive coroutine
44 debugging, tracing and much more.
45 - major bug fix: unbelievable, but true: $_, $/ and many other
46 "saved" variables actually weren't being saved. This has been fixed,
47 of course, while increasing performance while losing all the save
48 flags.
49 - save flags are gone, and all the api functions dealing with them.
50 - added Coro::Semaphore::adjust.
51 - added Coro::Util::fork_eval.
52 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
53 - added Coro::killall.
54 - reduce initial stack sizes to allow for "micro-coroutines".
55 - better async_pool resource management, moved parts of async_pool
56 handling to XS (major speed improvement).
57 - actually croak before modifying important data structures.
58 - refuse to transfer while compiling.
59 - possibly support eval EXPR better now.
60 - enable assembly per default on linux+bsd x86+amd64.
61 - all internal members were renamed _something for easier subclassing.
62 - many minor tweaks.
63
643.63 Wed May 16 14:10:06 CEST 2007
65 - implement handcoded assembly for x86/amd64 SVR ABI.
66
673.62 Fri Apr 27 21:36:06 CEST 2007
68 - upgrade libcoro (which might set unwind info correctly).
69 - change default on linux to setjmp/longjmp.
70
713.61 Thu Apr 19 12:36:18 CEST 2007
72 - Coro::Storable caused an endless loop when thawing invalid
73 pst's sometimes.
74 - use a Semaphore in Coro::Storable, as Storable doesn't
75 seem to be reentrant (although it is documented to
76 be threadsafe...).
77 - fix Coro::Signal to bring back the original unreliable
78 but stateful semantics.
79 - fixed a lot of typos in Coro.pm (patch submitted by David
80 Steinbrunner, which applied flawlessly).
81
823.6 Sat Apr 14 17:13:31 CEST 2007
83 - added some bugfixes to get eg/myhttpd working again.
84 - added Coro::Storable for often-cede'ing freeze/thaw.
85 - try to do a clean exit when a coroutine calls exit
86 (EXPERIMENTAL).
87 - got rid of indirect call through _coro_init.
88 - updated the partly antique examples in eg/ to
89 work again and be a bit less magic, too.
90 - fixed Coro::Signal semantics to work as documented again.
91
923.55 Sun Mar 25 01:20:47 CET 2007
93 - add SAVE_DEFFH to save the default filehandle and enable
94 it by default.
95 - finally move socket-operations from Coro::Socket to Coro::Handle
96 to be able to unblock foreign sockets.
97 - add Coro::State::save_also and guarded_save.
98 - add count accessor to Coro::Semaphore.
99 - add Coro::State::cctx_stacksize.
100 - just for the fun of it, do not rely on implicit context,
101 which can dramatically improve performance, but people
102 using windows-process-emulation perls do not care much about
103 performance.
104
1053.51 Sun Mar 4 14:18:04 CET 2007
106 - fixed a problem when you weakened references to Coro::State's
107 (patch by Yuval Kogman).
108
1093.501 Wed Feb 28 12:44:07 CET 2007
110 - rename some global symbols as macosx from hell redefines
111 them without asking.
112
1133.5 Tue Feb 13 20:22:53 CET 2007
114 - do AnyEvent model detection earlier, avoiding problems
115 caused by first using AnyEvent and later Coro::Event.
116 - implement and document Coro::Event event objects.
117 - fix a potential problem in Coro::Event causing crashes.
118 - initialise PL_comppad when creating a new coroutine,
119 avoids crashes on early coro destruction.
120
1213.41 Mon Jan 22 19:19:49 CET 2007
122 - readline on Coro::Handle did not support undefined $/,
123 nor did it deliver partial lines on EOF or error.
124 - implement malloc fallback for stack allocation because
125 stupid broken idiotic OSX has a stupid broken
126 idiotic fits-the-whole-os mmap "implementation" and
127 my dick feels longer if Coro is portable even to
128 obsolete platforms.
129
1303.4 Fri Jan 19 21:52:54 CET 2007
131 - remove t/09_timer.t, as it isn't really testing much
132 but was rather flaky in practise.
133 - async_pool coro would keep arguments and callback alive until
134 it was reused.
135 - cancellation of a coroutine could cause spurious idle calls
136 in cede_notself.
137
1383.3 Sat Jan 6 03:45:00 CET 2007
139 - implement $coro->on_destroy.
140 - Coro::Event blocking semantics have been changed,
141 documented and - hopefully - improved.
142 - fix nice adding, not subtracting, from priority.
143 - fix ->prio and api_is_ready (patch by Mark Hinds).
144 - fixed an assert ("... == prev__cctx->idle_te")
145 that could errronously trigger.
146 - fix various large and small memleaks.
147 - use a (hopefully) more stable cancel implementation
148 that immediately frees the coroutine data.
149 - cede/cede_notself return a status now.
150 - added Coro::guard function.
151 - added a global coroutine pool for jobs (on my machine,
152 I can create and execute 48k simple coros/s with async,
153 and 128k coros with async_pool).
154 - Coro::AIO now uses the coroutine priority as io priority.
155
1563.2 Fri Dec 22 05:07:09 CET 2006
157 - improve portability to slightly older perls.
158 - use cleaner coroutine destruction.
159 - simplify configuration for users.
160 - optionally (unrecommended) prefer perl functions over
161 their coro replacements.
162
1633.11 Tue Dec 5 13:11:24 CET 2006
164 - fixed some bogus assert's, but as perl.h disables assert even
165 without NDEBUG (thank you very much), not too many people should
166 notice (that did include myself). Andreas König noticed, though :)
167 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
168 it. Also noticed by Andreas König :)
169 - save/restore tainted status.
170 - verified to pass the testsuite on my 5.9.5.
171
1723.1 Mon Dec 4 23:03:40 CET 2006
173 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
174 - incompatible change: transfer flags are now per-state.
175 - give a better error message on deadlock.
176 - document Coro::nready.
177 - enhanced testsuite.
178
1793.01 Sun Dec 3 23:47:42 CET 2006
180 - forgot to include Coro::Timer.
181
1823.0 Sun Dec 3 22:57:25 CET 2006
183 - the "FINALLY COMPLETELY STABLE" release (coming soon:
184 the "FAMOUS LAST WORDS" release).
185 - implement a new stack sharing algorithm, which uses a stack
186 pool (size currently hardcoded).
187 - make stack sharing mandatory (it no longer uses a heuristic).
188 - eval/die no longer cause weird problems under heavy use.
189 - Coro::Event could cause livelocks if it was used but
190 no Coro::Event watchers were used.
191 - Coro::Event now uses asynccheck as prepare does not
192 check for changed watchers.
193 - Coro::Event allows multiple waiting coros and will wake up one
194 per event.
195 - Coro::Event should be cleaner and more efficient now.
196 - new utility function Coro::unblock_sub.
197 - document the sad fact that Event is no longer reentrant.
198 - putting a coroutine into the ready queue twice could under
199 some circumstances lead to stack corruption.
200 - minor incompatibility: subclassing Coro::State is supported
201 directly now without going through a _coro_state member.
202 - state/coro switching is much faster now.
203 - very minor optimisations and code/documentation cleanup.
204 - avoid problems due to compiler inlining.
205 - removed timers from Coro::Timer -> use AnyEvent instead.
206 - replaced Coro::idle coroutine by (cleaner) idle handler.
207 - updated to newest libcoro.
208 - implement enhanced support for valgrind.
209 - implement is_ready and return value for ready.
210 - removed Coro::Cont, it was a misnomer (it's generators),
211 and it was somewhat annoying to get it right. Will come back
212 if somebody asks for it :)
213 - many ->wait methods and Coro::Event could return
214 spuriously without the event having happened.
215
2162.5 Tue Nov 7 12:22:33 CET 2006
217 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
218 use AnyEvent, moved them to Coro/.
219 - added Coro::LWP which contains all the uglyness required to
220 make LWP non-blocking.
221 - should work with perl 5.9.x now (Andreas König made me do it).
222 - fixed another bug in Coro::Select when the timeout was undef.
223 - reuse PL_start_env for all coros, saves some memory per coroutine.
224 - manage PL_top_env differently, hopefully to avoid panic: top_env.
225 - timeout argument was not properly used in Coro::Socket.
226 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
227 - emulate undocumented(!) functionality of IO::Socket required
228 by LWP(!!).
229 - updated eg/lwp to work with newer lwp's.
230 - remove "FATAL: uncaught exception" prefix. Coroutines that die
231 kill the whole process, just as exceptions in the main "coroutine"
232 did already.
233
2342.1 Wed Nov 1 23:01:13 CET 2006
235 - fix a long-standing bug in Coro::Select where select with
236 zero timeout would instead change the current default filehandle.
237 - use a simpler and hopefully more robust way to clone padlists
238 (uses less memory and a perl function instead of our own).
239 - coro can now create a stack guard area on many architectures.
240 - Coro::AIO properly reexports additional functions from IO::AIO.
241 - updated libcoro with a workaround for OS X,
242 pach and testing by Michael Schwern.
243
2442.0 Tue Oct 24 05:47:17 CEST 2006
245 - support additional aio requests in Coro::AIO.
246
2471.9 (never properly released due to a glitch)
248
2491.8 Thu Feb 2 00:59:06 CET 2006
250 - applied suggested patch by SAMV to avoid problems during stupid
251 mark & sweep gc run.
252 - applied patch by Scott Walters for 5.9.3 compatibility.
253
2541.7 Tue Dec 27 01:41:58 CET 2005
255 - added Coro::AIO, a thin wrapper around IO::AIO.
256 - improved Makefile.PL explanations.
257
2581.6 Mon Dec 12 21:30:05 CET 2005
259 - additionally save PM_curpm might fix as-of-yet
260 unseen problems with regex matches being attributed to the
261 wrong package.
262 - add t/10_bugs.t, which currently checks against imho broken
263 perls that use 0.26MB of stack space per Perl_magic_get
264 invocation (newer linuxes) as opposed to the 0.0002MB perl
265 normally uses.
266 - make stacksize configurable for the ultimate debian experience.
267
2681.51 Mon Dec 12 18:48:36 CET 2005
269 - remove debugging warn() accidentally left in Coro::Select.
270
2711.5 Tue Nov 29 13:32:44 CET 2005
272 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
273 - fix Coro::Select 'select'.
274 - strict'ify some modules.
275 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
276 - use XSLoader in selected modules.
277 - remove some 5.6 compatibility cruft.
278
2791.4 Tue Sep 6 00:11:05 CEST 2005
280 - libcoro did not take into account the trampoline on amd64,
281 when 'l'inux method was used.
282
2831.31 Tue Aug 30 23:31:33 CEST 2005
284 - some portability fixes/workarounds.
285
2861.3 Sat Aug 20 03:08:56 CEST 2005
287 - no code changes. module seems to work fine.
288 - improve Coro::State docs, remove reference to nonexisting
289 Coro::State::flush method.
290 - no longer autodetect windows, present it as an option instead.
291
2921.2 Mon May 16 02:00:55 CEST 2005
293 - included libcoro.c earlier in State.xs, might improve portability.
294 - use a faster and possibly more stable Coro::Cont implementation.
295 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
296 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
297 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
298
2991.11 Thu Mar 3 18:00:52 CET 2005
300 - change of contact address.
301
3021.1 Tue Feb 22 20:51:16 CET 2005
303 - support [l]inux method on amd64.
304 - allow some context switches while compiling/eval'ing.
305
3061.0 Tue Aug 10 03:47:27 CEST 2004
307 - Event 0.89 fixes the bug again.
308 - fix the remaining(??) memleak. I didn't free the cache-AV for
309 padlists, so one AV per created anon-sub leaked.
2 310
30.97 Fri May 14 15:23:32 CEST 2004 3110.97 Fri May 14 15:23:32 CEST 2004
4 - remove stupid gcc extension that was an experiment only anyway. 312 - remove stupid gcc extension that was an experiment only anyway.
5 313
60.96 Thu May 13 18:09:29 CEST 2004 3140.96 Thu May 13 18:09:29 CEST 2004
7 - tell people if they have yet anothe rbroken Event variant. 315 - tell people if they have yet another broken Event variant.
8 - made Coro to work _better_ with threads (it seems that, as long 316 - made Coro to work _better_ with threads (it seems that, as long
9 as you use Coros in one thread only everything will work fine. 317 as you use Coros in one thread only everything will work fine.
10 Looking at the perl source the problem might be missing locking 318 Looking at the perl source the problem might be missing locking
11 between Coro and some perl internal routines. That's not easily 319 between Coro and some perl internal routines. That's not easily
12 fixable). 320 fixable).

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines