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

Comparing Coro/Changes (file contents):
Revision 1.79 by root, Thu May 8 00:55:28 2003 UTC vs.
Revision 1.263 by root, Sat Oct 6 19:25:00 2007 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2
3TODO: localise SIG{__DIE__} and/or __WARN__
4TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO.
5TODO: maybe implement a default message channel, very much like Erlang
6 (which is cool in a lot of important aspects (failures!),
7 but very lacking in others (higher level ipc)).
8
94.03 Sat Oct 6 21:24:00 CEST 2007
10 - added Coro::throw method.
11 - minor code cleanups.
12
134.02 Sat Oct 6 02:36:47 CEST 2007
14 - fix a very minor per-coroutine memleak (a single codereference).
15 - fixed a bug where the currently in-use c context would be freed
16 prematurely (can happen only when programs change the stacksize
17 or use tracing).
18 - tracing can no longer keep a coro alive after it terminated.
19 - do static branch prediction in the common path for gcc. gives
20 about 2-5% speed improvement here.
21
224.01 Fri Oct 5 22:10:49 CEST 2007
23 - instead of recreating *a* standard output handle we simply
24 use STDOUT, which is faster and hopefully more robust.
25
264.0 Fri Oct 5 12:56:00 CEST 2007
27 - incompatibly changed Coro::Storable::freeze.
28 - major new feature: added Coro::Debug, for interactive coroutine
29 debugging, tracing and much more.
30 - major bug fix: unbelievable, but true: $_, $/ and many other
31 "saved" variables actually weren't being saved. This has been fixed,
32 of course, while increasing performance while losing all the save
33 flags.
34 - save flags are gone, and all the api functions dealing with them.
35 - added Coro::Semaphore::adjust.
36 - added Coro::Util::fork_eval.
37 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
38 - added Coro::killall.
39 - reduce initial stack sizes to allow for "micro-coroutines".
40 - better async_pool resource management, moved parts of async_pool
41 handling to XS (major speed improvement).
42 - actually croak before modifying important data structures.
43 - refuse to transfer while compiling.
44 - possibly support eval EXPR better now.
45 - enable assembly per default on linux+bsd x86+amd64.
46 - all internal members were renamed _something for easier subclassing.
47 - many minor tweaks.
48
493.63 Wed May 16 14:10:06 CEST 2007
50 - implement handcoded assembly for x86/amd64 SVR ABI.
51
523.62 Fri Apr 27 21:36:06 CEST 2007
53 - upgrade libcoro (which might set unwind info correctly).
54 - change default on linux to setjmp/longjmp.
55
563.61 Thu Apr 19 12:36:18 CEST 2007
57 - Coro::Storable caused an endless loop when thawing invalid
58 pst's sometimes.
59 - use a Semaphore in Coro::Storable, as Storable doesn't
60 seem to be reentrant (although it is documented to
61 be threadsafe...).
62 - fix Coro::Signal to bring back the original unreliable
63 but stateful semantics.
64 - fixed a lot of typos in Coro.pm (patch submitted by David
65 Steinbrunner, which applied flawlessly).
66
673.6 Sat Apr 14 17:13:31 CEST 2007
68 - added some bugfixes to get eg/myhttpd working again.
69 - added Coro::Storable for often-cede'ing freeze/thaw.
70 - try to do a clean exit when a coroutine calls exit
71 (EXPERIMENTAL).
72 - got rid of indirect call through _coro_init.
73 - updated the partly antique examples in eg/ to
74 work again and be a bit less magic, too.
75 - fixed Coro::Signal semantics to work as documented again.
76
773.55 Sun Mar 25 01:20:47 CET 2007
78 - add SAVE_DEFFH to save the default filehandle and enable
79 it by default.
80 - finally move socket-operations from Coro::Socket to Coro::Handle
81 to be able to unblock foreign sockets.
82 - add Coro::State::save_also and guarded_save.
83 - add count accessor to Coro::Semaphore.
84 - add Coro::State::cctx_stacksize.
85 - just for the fun of it, do not rely on implicit context,
86 which can dramatically improve performance, but people
87 using windows-process-emulation perls do not care much about
88 performance.
89
903.51 Sun Mar 4 14:18:04 CET 2007
91 - fixed a problem when you weakened references to Coro::State's
92 (patch by Yuval Kogman).
93
943.501 Wed Feb 28 12:44:07 CET 2007
95 - rename some global symbols as macosx from hell redefines
96 them without asking.
97
983.5 Tue Feb 13 20:22:53 CET 2007
99 - do AnyEvent model detection earlier, avoiding problems
100 caused by first using AnyEvent and later Coro::Event.
101 - implement and document Coro::Event event objects.
102 - fix a potential problem in Coro::Event causing crashes.
103 - initialise PL_comppad when creating a new coroutine,
104 avoids crashes on early coro destruction.
105
1063.41 Mon Jan 22 19:19:49 CET 2007
107 - readline on Coro::Handle did not support undefined $/,
108 nor did it deliver partial lines on EOF or error.
109 - implement malloc fallback for stack allocation because
110 stupid broken idiotic OSX has a stupid broken
111 idiotic fits-the-whole-os mmap "implementation" and
112 my dick feels longer if Coro is portable even to
113 obsolete platforms.
114
1153.4 Fri Jan 19 21:52:54 CET 2007
116 - remove t/09_timer.t, as it isn't really testing much
117 but was rather flaky in practise.
118 - async_pool coro would keep arguments and callback alive until
119 it was reused.
120 - cancellation of a coroutine could cause spurious idle calls
121 in cede_notself.
122
1233.3 Sat Jan 6 03:45:00 CET 2007
124 - implement $coro->on_destroy.
125 - Coro::Event blocking semantics have been changed,
126 documented and - hopefully - improved.
127 - fix nice adding, not subtracting, from priority.
128 - fix ->prio and api_is_ready (patch by Mark Hinds).
129 - fixed an assert ("... == prev__cctx->idle_te")
130 that could errronously trigger.
131 - fix various large and small memleaks.
132 - use a (hopefully) more stable cancel implementation
133 that immediately frees the coroutine data.
134 - cede/cede_notself return a status now.
135 - added Coro::guard function.
136 - added a global coroutine pool for jobs (on my machine,
137 I can create and execute 48k simple coros/s with async,
138 and 128k coros with async_pool).
139 - Coro::AIO now uses the coroutine priority as io priority.
140
1413.2 Fri Dec 22 05:07:09 CET 2006
142 - improve portability to slightly older perls.
143 - use cleaner coroutine destruction.
144 - simplify configuration for users.
145 - optionally (unrecommended) prefer perl functions over
146 their coro replacements.
147
1483.11 Tue Dec 5 13:11:24 CET 2006
149 - fixed some bogus assert's, but as perl.h disables assert even
150 without NDEBUG (thank you very much), not too many people should
151 notice (that did include myself). Andreas König noticed, though :)
152 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
153 it. Also noticed by Andreas König :)
154 - save/restore tainted status.
155 - verified to pass the testsuite on my 5.9.5.
156
1573.1 Mon Dec 4 23:03:40 CET 2006
158 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
159 - incompatible change: transfer flags are now per-state.
160 - give a better error message on deadlock.
161 - document Coro::nready.
162 - enhanced testsuite.
163
1643.01 Sun Dec 3 23:47:42 CET 2006
165 - forgot to include Coro::Timer.
166
1673.0 Sun Dec 3 22:57:25 CET 2006
168 - the "FINALLY COMPLETELY STABLE" release (coming soon:
169 the "FAMOUS LAST WORDS" release).
170 - implement a new stack sharing algorithm, which uses a stack
171 pool (size currently hardcoded).
172 - make stack sharing mandatory (it no longer uses a heuristic).
173 - eval/die no longer cause weird problems under heavy use.
174 - Coro::Event could cause livelocks if it was used but
175 no Coro::Event watchers were used.
176 - Coro::Event now uses asynccheck as prepare does not
177 check for changed watchers.
178 - Coro::Event allows multiple waiting coros and will wake up one
179 per event.
180 - Coro::Event should be cleaner and more efficient now.
181 - new utility function Coro::unblock_sub.
182 - document the sad fact that Event is no longer reentrant.
183 - putting a coroutine into the ready queue twice could under
184 some circumstances lead to stack corruption.
185 - minor incompatibility: subclassing Coro::State is supported
186 directly now without going through a _coro_state member.
187 - state/coro switching is much faster now.
188 - very minor optimisations and code/documentation cleanup.
189 - avoid problems due to compiler inlining.
190 - removed timers from Coro::Timer -> use AnyEvent instead.
191 - replaced Coro::idle coroutine by (cleaner) idle handler.
192 - updated to newest libcoro.
193 - implement enhanced support for valgrind.
194 - implement is_ready and return value for ready.
195 - removed Coro::Cont, it was a misnomer (it's generators),
196 and it was somewhat annoying to get it right. Will come back
197 if somebody asks for it :)
198 - many ->wait methods and Coro::Event could return
199 spuriously without the event having happened.
200
2012.5 Tue Nov 7 12:22:33 CET 2006
202 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
203 use AnyEvent, moved them to Coro/.
204 - added Coro::LWP which contains all the uglyness required to
205 make LWP non-blocking.
206 - should work with perl 5.9.x now (Andreas König made me do it).
207 - fixed another bug in Coro::Select when the timeout was undef.
208 - reuse PL_start_env for all coros, saves some memory per coroutine.
209 - manage PL_top_env differently, hopefully to avoid panic: top_env.
210 - timeout argument was not properly used in Coro::Socket.
211 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
212 - emulate undocumented(!) functionality of IO::Socket required
213 by LWP(!!).
214 - updated eg/lwp to work with newer lwp's.
215 - remove "FATAL: uncaught exception" prefix. Coroutines that die
216 kill the whole process, just as exceptions in the main "coroutine"
217 did already.
218
2192.1 Wed Nov 1 23:01:13 CET 2006
220 - fix a long-standing bug in Coro::Select where select with
221 zero timeout would instead change the current default filehandle.
222 - use a simpler and hopefully more robust way to clone padlists
223 (uses less memory and a perl function instead of our own).
224 - coro can now create a stack guard area on many architectures.
225 - Coro::AIO properly reexports additional functions from IO::AIO.
226 - updated libcoro with a workaround for OS X,
227 pach and testing by Michael Schwern.
228
2292.0 Tue Oct 24 05:47:17 CEST 2006
230 - support additional aio requests in Coro::AIO.
231
2321.9 (never properly released due to a glitch)
233
2341.8 Thu Feb 2 00:59:06 CET 2006
235 - applied suggested patch by SAMV to avoid problems during stupid
236 mark & sweep gc run.
237 - applied patch by Scott Walters for 5.9.3 compatibility.
238
2391.7 Tue Dec 27 01:41:58 CET 2005
240 - added Coro::AIO, a thin wrapper around IO::AIO.
241 - improved Makefile.PL explanations.
242
2431.6 Mon Dec 12 21:30:05 CET 2005
244 - additionally save PM_curpm might fix as-of-yet
245 unseen problems with regex matches being attributed to the
246 wrong package.
247 - add t/10_bugs.t, which currently checks against imho broken
248 perls that use 0.26MB of stack space per Perl_magic_get
249 invocation (newer linuxes) as opposed to the 0.0002MB perl
250 normally uses.
251 - make stacksize configurable for the ultimate debian experience.
252
2531.51 Mon Dec 12 18:48:36 CET 2005
254 - remove debugging warn() accidentally left in Coro::Select.
255
2561.5 Tue Nov 29 13:32:44 CET 2005
257 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
258 - fix Coro::Select 'select'.
259 - strict'ify some modules.
260 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
261 - use XSLoader in selected modules.
262 - remove some 5.6 compatibility cruft.
263
2641.4 Tue Sep 6 00:11:05 CEST 2005
265 - libcoro did not take into account the trampoline on amd64,
266 when 'l'inux method was used.
267
2681.31 Tue Aug 30 23:31:33 CEST 2005
269 - some portability fixes/workarounds.
270
2711.3 Sat Aug 20 03:08:56 CEST 2005
272 - no code changes. module seems to work fine.
273 - improve Coro::State docs, remove reference to nonexisting
274 Coro::State::flush method.
275 - no longer autodetect windows, present it as an option instead.
276
2771.2 Mon May 16 02:00:55 CEST 2005
278 - included libcoro.c earlier in State.xs, might improve portability.
279 - use a faster and possibly more stable Coro::Cont implementation.
280 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
281 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
282 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
283
2841.11 Thu Mar 3 18:00:52 CET 2005
285 - change of contact address.
286
2871.1 Tue Feb 22 20:51:16 CET 2005
288 - support [l]inux method on amd64.
289 - allow some context switches while compiling/eval'ing.
290
2911.0 Tue Aug 10 03:47:27 CEST 2004
292 - Event 0.89 fixes the bug again.
293 - fix the remaining(??) memleak. I didn't free the cache-AV for
294 padlists, so one AV per created anon-sub leaked.
295
2960.97 Fri May 14 15:23:32 CEST 2004
297 - remove stupid gcc extension that was an experiment only anyway.
298
2990.96 Thu May 13 18:09:29 CEST 2004
300 - tell people if they have yet another broken Event variant.
301 - made Coro to work _better_ with threads (it seems that, as long
302 as you use Coros in one thread only everything will work fine.
303 Looking at the perl source the problem might be missing locking
304 between Coro and some perl internal routines. That's not easily
305 fixable).
306 - (hopefully) got rid of the static xnext inside transfer.
307
3080.95 Sun Feb 15 03:20:28 CET 2004
309 - removed Coro::State::flush, implemented a new and much
310 faster caching mechanism by attaching magic to the CV.
311 - a memleak with (real) closures remains.
312 - dramatically reduced size of stacks to 65536 (linux/x86), after
313 observing that even myhttpd never had more than a kilobyte
314 of stack in use (stack is still only physically allocated in
315 pagesize increments for systems suporting mmap).
316 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
317 - fix for threaded perl.
318
3190.9 Sun Nov 23 04:32:39 CET 2003
320 - fix a few very hard to track down but (of course) stupid
321 memory leaks.
322 - stringify version number to avoid locale problems :(.
323 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
324 hopefully ignoring it "just works".
325
3260.8 Wed Nov 5 19:38:40 CET 2003
327 - port to perl5.005_03, but only 5.8.x+ is supported!
328 - honor LocalAddr even without LocalPort in Coro::Socket.
329
3300.7 Tue May 27 03:12:38 CEST 2003
331 - the version jump indicates some level of testing, not gobs
332 of new features.
333 - uh... I found the prompt function in ExtUtils::MakeMaker.
334 Highly correct stuff, that is...
335 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
336 the fix is not well understood by the author, so beware :(.
2 337
30.652 Thu May 8 02:54:46 CEST 2003 3380.652 Thu May 8 02:54:46 CEST 2003
4 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD, 339 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
5 cause version 4 doesn't have ucontext. 340 cause version 4 doesn't have ucontext.
6 - Benjamin Reed reported that setjmp works fine on darwin, 341 - Benjamin Reed reported that setjmp works fine on darwin,

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines