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

Comparing Coro/Changes (file contents):
Revision 1.54 by root, Wed Oct 3 01:09:56 2001 UTC vs.
Revision 1.295 by root, Sat Dec 1 22:39:39 2007 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
30.51 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
8 - do no longer rely on the presence of EV::DNS if EV is used
9 (because it is gone), but instead take avdantage of EV::ADNS
10 if available.
11
124.22 Fri Nov 30 16:04:04 CET 2007
13 - really use optimised versions for Event and EV in Coro::Util
14 and Coro::Handle.
15
164.21 Sun Nov 25 10:48:59 CET 2007
17 - fix a spurious memory read.
18 - Coro::EV no longer keeps the eventloop "alive".
19
204.2 Fri Nov 9 20:47:05 CET 2007
21 - enable/disable tracing from a new coroutine, not a pooled one.
22 - fix a memleak in Coro::Event.
23 - removed killall call from fork_eval.
24 - made sure store_fd is already loaded so that fork_eval
25 does not have to parse autoload in each subprocess.
26 - only use assembly method if -O switch is in $Config{optimize}.
27 - add (optional) Coro::EV module, so far the best event loop module
28 directly supported by Coro.
29 - if the event model is EV, use EV::DNS to resolve
30 stuff in Coro::Util.
31 - don't get confused by multiple event notifications in Coro::Handle.
32 - initial support for EV (libevent interface).
33 - require Event and EV using configure_requires, to force their existance.
34
354.13 Wed Oct 24 07:26:45 CEST 2007
36 - add Coro::Storable::blocking_thaw.
37 - use a vastly more complicated technique to localise
38 $SIG{__WARN/DIE__} that also works on perls <= 5.8.8.
39 - use a coroutine for the idle callback Coro::Event,
40 instead of running Event in the current coroutine context.
41 This also catches recursive invocations.
42 - actually report fork errors in gethostbyname and inet_aton.
43
444.11 Thu Oct 11 02:40:24 CEST 2007
45 - port to threaded perls.
46
474.1 Thu Oct 11 02:38:16 CEST 2007
48 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
49 be local to each coro (see Coro::State).
50 - incompatible change: for very deep reasons, cede and cede_notself
51 cannot return anything, so nothing will be returned.
52 - possibly bring back 5.10 compatibility (untested).
53 - work around stupid (and wrong) warning on 5.10 :(.
54 - overlay the saved state over the context stack. This saves
55 a few hundred bytes per coroutine on average and also
56 speeds up context switching a bit.
57 - further tune default stack sizes.
58 - (more) correctly calculate stack usage in coro_rss.
59 - Coro::Storable::blocking_* did not properly lock
60 resulting in races between coroutines.
61 - added Coro::Storable::guard.
62 - stopping to trace a coroutine could destroy the cctx of
63 an unrelated coroutine.
64 - explain the relationship between Perl and C coroutines in
65 more detail in Coro::State.
66 - Coro::Util::inet_aton did not short-circuit dotted quad forms,
67 causing a fork per resolve. This also affected Coro::Socket.
68 - switch to a separate stack in $coro->call/eval to avoid
69 invalidating pointers.
70
714.03 Sat Oct 6 21:24:00 CEST 2007
72 - added Coro::throw method.
73 - minor code cleanups.
74
754.02 Sat Oct 6 02:36:47 CEST 2007
76 - fix a very minor per-coroutine memleak (a single codereference).
77 - fixed a bug where the currently in-use c context would be freed
78 prematurely (can happen only when programs change the stacksize
79 or use tracing).
80 - tracing can no longer keep a coro alive after it terminated.
81 - do static branch prediction in the common path for gcc. gives
82 about 2-5% speed improvement here.
83
844.01 Fri Oct 5 22:10:49 CEST 2007
85 - instead of recreating *a* standard output handle we simply
86 use STDOUT, which is faster and hopefully more robust.
87
884.0 Fri Oct 5 12:56:00 CEST 2007
89 - incompatibly changed Coro::Storable::freeze.
90 - major new feature: added Coro::Debug, for interactive coroutine
91 debugging, tracing and much more.
92 - major bug fix: unbelievable, but true: $_, $/ and many other
93 "saved" variables actually weren't being saved. This has been fixed,
94 of course, while increasing performance while losing all the save
95 flags.
96 - save flags are gone, and all the api functions dealing with them.
97 - added Coro::Semaphore::adjust.
98 - added Coro::Util::fork_eval.
99 - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}.
100 - added Coro::killall.
101 - reduce initial stack sizes to allow for "micro-coroutines".
102 - better async_pool resource management, moved parts of async_pool
103 handling to XS (major speed improvement).
104 - actually croak before modifying important data structures.
105 - refuse to transfer while compiling.
106 - possibly support eval EXPR better now.
107 - enable assembly per default on linux+bsd x86+amd64.
108 - all internal members were renamed _something for easier subclassing.
109 - many minor tweaks.
110
1113.63 Wed May 16 14:10:06 CEST 2007
112 - implement handcoded assembly for x86/amd64 SVR ABI.
113
1143.62 Fri Apr 27 21:36:06 CEST 2007
115 - upgrade libcoro (which might set unwind info correctly).
116 - change default on linux to setjmp/longjmp.
117
1183.61 Thu Apr 19 12:36:18 CEST 2007
119 - Coro::Storable caused an endless loop when thawing invalid
120 pst's sometimes.
121 - use a Semaphore in Coro::Storable, as Storable doesn't
122 seem to be reentrant (although it is documented to
123 be threadsafe...).
124 - fix Coro::Signal to bring back the original unreliable
125 but stateful semantics.
126 - fixed a lot of typos in Coro.pm (patch submitted by David
127 Steinbrunner, which applied flawlessly).
128
1293.6 Sat Apr 14 17:13:31 CEST 2007
130 - added some bugfixes to get eg/myhttpd working again.
131 - added Coro::Storable for often-cede'ing freeze/thaw.
132 - try to do a clean exit when a coroutine calls exit
133 (EXPERIMENTAL).
134 - got rid of indirect call through _coro_init.
135 - updated the partly antique examples in eg/ to
136 work again and be a bit less magic, too.
137 - fixed Coro::Signal semantics to work as documented again.
138
1393.55 Sun Mar 25 01:20:47 CET 2007
140 - add SAVE_DEFFH to save the default filehandle and enable
141 it by default.
142 - finally move socket-operations from Coro::Socket to Coro::Handle
143 to be able to unblock foreign sockets.
144 - add Coro::State::save_also and guarded_save.
145 - add count accessor to Coro::Semaphore.
146 - add Coro::State::cctx_stacksize.
147 - just for the fun of it, do not rely on implicit context,
148 which can dramatically improve performance, but people
149 using windows-process-emulation perls do not care much about
150 performance.
151
1523.51 Sun Mar 4 14:18:04 CET 2007
153 - fixed a problem when you weakened references to Coro::State's
154 (patch by Yuval Kogman).
155
1563.501 Wed Feb 28 12:44:07 CET 2007
157 - rename some global symbols as macosx from hell redefines
158 them without asking.
159
1603.5 Tue Feb 13 20:22:53 CET 2007
161 - do AnyEvent model detection earlier, avoiding problems
162 caused by first using AnyEvent and later Coro::Event.
163 - implement and document Coro::Event event objects.
164 - fix a potential problem in Coro::Event causing crashes.
165 - initialise PL_comppad when creating a new coroutine,
166 avoids crashes on early coro destruction.
167
1683.41 Mon Jan 22 19:19:49 CET 2007
169 - readline on Coro::Handle did not support undefined $/,
170 nor did it deliver partial lines on EOF or error.
171 - implement malloc fallback for stack allocation because
172 stupid broken idiotic OSX has a stupid broken
173 idiotic fits-the-whole-os mmap "implementation" and
174 my dick feels longer if Coro is portable even to
175 obsolete platforms.
176
1773.4 Fri Jan 19 21:52:54 CET 2007
178 - remove t/09_timer.t, as it isn't really testing much
179 but was rather flaky in practise.
180 - async_pool coro would keep arguments and callback alive until
181 it was reused.
182 - cancellation of a coroutine could cause spurious idle calls
183 in cede_notself.
184
1853.3 Sat Jan 6 03:45:00 CET 2007
186 - implement $coro->on_destroy.
187 - Coro::Event blocking semantics have been changed,
188 documented and - hopefully - improved.
189 - fix nice adding, not subtracting, from priority.
190 - fix ->prio and api_is_ready (patch by Mark Hinds).
191 - fixed an assert ("... == prev__cctx->idle_te")
192 that could errronously trigger.
193 - fix various large and small memleaks.
194 - use a (hopefully) more stable cancel implementation
195 that immediately frees the coroutine data.
196 - cede/cede_notself return a status now.
197 - added Coro::guard function.
198 - added a global coroutine pool for jobs (on my machine,
199 I can create and execute 48k simple coros/s with async,
200 and 128k coros with async_pool).
201 - Coro::AIO now uses the coroutine priority as io priority.
202
2033.2 Fri Dec 22 05:07:09 CET 2006
204 - improve portability to slightly older perls.
205 - use cleaner coroutine destruction.
206 - simplify configuration for users.
207 - optionally (unrecommended) prefer perl functions over
208 their coro replacements.
209
2103.11 Tue Dec 5 13:11:24 CET 2006
211 - fixed some bogus assert's, but as perl.h disables assert even
212 without NDEBUG (thank you very much), not too many people should
213 notice (that did include myself). Andreas König noticed, though :)
214 - do not save/restore PL_sortcxix on >= 5.9.x, it doesn't seem to have
215 it. Also noticed by Andreas König :)
216 - save/restore tainted status.
217 - verified to pass the testsuite on my 5.9.5.
218
2193.1 Mon Dec 4 23:03:40 CET 2006
220 - INCOMPATIBLE CHANGE: $/ is now per-coroutine (but slow).
221 - incompatible change: transfer flags are now per-state.
222 - give a better error message on deadlock.
223 - document Coro::nready.
224 - enhanced testsuite.
225
2263.01 Sun Dec 3 23:47:42 CET 2006
227 - forgot to include Coro::Timer.
228
2293.0 Sun Dec 3 22:57:25 CET 2006
230 - the "FINALLY COMPLETELY STABLE" release (coming soon:
231 the "FAMOUS LAST WORDS" release).
232 - implement a new stack sharing algorithm, which uses a stack
233 pool (size currently hardcoded).
234 - make stack sharing mandatory (it no longer uses a heuristic).
235 - eval/die no longer cause weird problems under heavy use.
236 - Coro::Event could cause livelocks if it was used but
237 no Coro::Event watchers were used.
238 - Coro::Event now uses asynccheck as prepare does not
239 check for changed watchers.
240 - Coro::Event allows multiple waiting coros and will wake up one
241 per event.
242 - Coro::Event should be cleaner and more efficient now.
243 - new utility function Coro::unblock_sub.
244 - document the sad fact that Event is no longer reentrant.
245 - putting a coroutine into the ready queue twice could under
246 some circumstances lead to stack corruption.
247 - minor incompatibility: subclassing Coro::State is supported
248 directly now without going through a _coro_state member.
249 - state/coro switching is much faster now.
250 - very minor optimisations and code/documentation cleanup.
251 - avoid problems due to compiler inlining.
252 - removed timers from Coro::Timer -> use AnyEvent instead.
253 - replaced Coro::idle coroutine by (cleaner) idle handler.
254 - updated to newest libcoro.
255 - implement enhanced support for valgrind.
256 - implement is_ready and return value for ready.
257 - removed Coro::Cont, it was a misnomer (it's generators),
258 and it was somewhat annoying to get it right. Will come back
259 if somebody asks for it :)
260 - many ->wait methods and Coro::Event could return
261 spuriously without the event having happened.
262
2632.5 Tue Nov 7 12:22:33 CET 2006
264 - made Coro::Util, Coro::Select, Coro::Handle and Coro::Socket
265 use AnyEvent, moved them to Coro/.
266 - added Coro::LWP which contains all the uglyness required to
267 make LWP non-blocking.
268 - should work with perl 5.9.x now (Andreas König made me do it).
269 - fixed another bug in Coro::Select when the timeout was undef.
270 - reuse PL_start_env for all coros, saves some memory per coroutine.
271 - manage PL_top_env differently, hopefully to avoid panic: top_env.
272 - timeout argument was not properly used in Coro::Socket.
273 - allow limited forms of subclassing in Coro::Handle/Coro::Socket.
274 - emulate undocumented(!) functionality of IO::Socket required
275 by LWP(!!).
276 - updated eg/lwp to work with newer lwp's.
277 - remove "FATAL: uncaught exception" prefix. Coroutines that die
278 kill the whole process, just as exceptions in the main "coroutine"
279 did already.
280
2812.1 Wed Nov 1 23:01:13 CET 2006
282 - fix a long-standing bug in Coro::Select where select with
283 zero timeout would instead change the current default filehandle.
284 - use a simpler and hopefully more robust way to clone padlists
285 (uses less memory and a perl function instead of our own).
286 - coro can now create a stack guard area on many architectures.
287 - Coro::AIO properly reexports additional functions from IO::AIO.
288 - updated libcoro with a workaround for OS X,
289 pach and testing by Michael Schwern.
290
2912.0 Tue Oct 24 05:47:17 CEST 2006
292 - support additional aio requests in Coro::AIO.
293
2941.9 (never properly released due to a glitch)
295
2961.8 Thu Feb 2 00:59:06 CET 2006
297 - applied suggested patch by SAMV to avoid problems during stupid
298 mark & sweep gc run.
299 - applied patch by Scott Walters for 5.9.3 compatibility.
300
3011.7 Tue Dec 27 01:41:58 CET 2005
302 - added Coro::AIO, a thin wrapper around IO::AIO.
303 - improved Makefile.PL explanations.
304
3051.6 Mon Dec 12 21:30:05 CET 2005
306 - additionally save PM_curpm might fix as-of-yet
307 unseen problems with regex matches being attributed to the
308 wrong package.
309 - add t/10_bugs.t, which currently checks against imho broken
310 perls that use 0.26MB of stack space per Perl_magic_get
311 invocation (newer linuxes) as opposed to the 0.0002MB perl
312 normally uses.
313 - make stacksize configurable for the ultimate debian experience.
314
3151.51 Mon Dec 12 18:48:36 CET 2005
316 - remove debugging warn() accidentally left in Coro::Select.
317
3181.5 Tue Nov 29 13:32:44 CET 2005
319 - use Coro::Event inside Coro::Select to avoid spurious deadlocks.
320 - fix Coro::Select 'select'.
321 - strict'ify some modules.
322 - libcoro errornously restored the SIGUSR2 handler to SIGUSR1.
323 - use XSLoader in selected modules.
324 - remove some 5.6 compatibility cruft.
325
3261.4 Tue Sep 6 00:11:05 CEST 2005
327 - libcoro did not take into account the trampoline on amd64,
328 when 'l'inux method was used.
329
3301.31 Tue Aug 30 23:31:33 CEST 2005
331 - some portability fixes/workarounds.
332
3331.3 Sat Aug 20 03:08:56 CEST 2005
334 - no code changes. module seems to work fine.
335 - improve Coro::State docs, remove reference to nonexisting
336 Coro::State::flush method.
337 - no longer autodetect windows, present it as an option instead.
338
3391.2 Mon May 16 02:00:55 CEST 2005
340 - included libcoro.c earlier in State.xs, might improve portability.
341 - use a faster and possibly more stable Coro::Cont implementation.
342 - accept x86_64 in addition to amd64 for optimized linux-amd64 support.
343 - fix bugs Coro::SemaphoreSet that could cause locks never to be freed.
344 - fix bugs in CoroAPI.h and document it in Coro::MakeMaker.
345
3461.11 Thu Mar 3 18:00:52 CET 2005
347 - change of contact address.
348
3491.1 Tue Feb 22 20:51:16 CET 2005
350 - support [l]inux method on amd64.
351 - allow some context switches while compiling/eval'ing.
352
3531.0 Tue Aug 10 03:47:27 CEST 2004
354 - Event 0.89 fixes the bug again.
355 - fix the remaining(??) memleak. I didn't free the cache-AV for
356 padlists, so one AV per created anon-sub leaked.
357
3580.97 Fri May 14 15:23:32 CEST 2004
359 - remove stupid gcc extension that was an experiment only anyway.
360
3610.96 Thu May 13 18:09:29 CEST 2004
362 - tell people if they have yet another broken Event variant.
363 - made Coro to work _better_ with threads (it seems that, as long
364 as you use Coros in one thread only everything will work fine.
365 Looking at the perl source the problem might be missing locking
366 between Coro and some perl internal routines. That's not easily
367 fixable).
368 - (hopefully) got rid of the static xnext inside transfer.
369
3700.95 Sun Feb 15 03:20:28 CET 2004
371 - removed Coro::State::flush, implemented a new and much
372 faster caching mechanism by attaching magic to the CV.
373 - a memleak with (real) closures remains.
374 - dramatically reduced size of stacks to 65536 (linux/x86), after
375 observing that even myhttpd never had more than a kilobyte
376 of stack in use (stack is still only physically allocated in
377 pagesize increments for systems suporting mmap).
378 - fix for a "cygwin-64" architecture added, proposed by Stefan Traby.
379 - fix for threaded perl.
380
3810.9 Sun Nov 23 04:32:39 CET 2003
382 - fix a few very hard to track down but (of course) stupid
383 memory leaks.
384 - stringify version number to avoid locale problems :(.
385 - perl 5.9.0 does not have SvPADBUSY, reported by Scott Walters,
386 hopefully ignoring it "just works".
387
3880.8 Wed Nov 5 19:38:40 CET 2003
389 - port to perl5.005_03, but only 5.8.x+ is supported!
390 - honor LocalAddr even without LocalPort in Coro::Socket.
391
3920.7 Tue May 27 03:12:38 CEST 2003
393 - the version jump indicates some level of testing, not gobs
394 of new features.
395 - uh... I found the prompt function in ExtUtils::MakeMaker.
396 Highly correct stuff, that is...
397 - fixed(?) a bug with die's in coroutines causing "panic: top_env".
398 the fix is not well understood by the author, so beware :(.
399
4000.652 Thu May 8 02:54:46 CEST 2003
401 - Applied patch by Slaven Rezic to set default to "s" on FreeBSD,
402 cause version 4 doesn't have ucontext.
403 - Benjamin Reed reported that setjmp works fine on darwin,
404 so preselect it.
405
4060.651 Sat Mar 29 15:00:23 CET 2003
407 - fix a bug in Coro::Handle where some bytes could get lost
408 on reading, reported by jason@nichego.net.
409
4100.65 Sun Mar 23 00:08:26 CET 2003
411 - added a README.linux-glibc.
412 - new module Coro::Select.
413 - also save/restore PL_comppad, fixes segfaults with 5.8.1.
414 (I never claimed that I know what I am doing ;).
415 - set default to setjmp/longjmp on non-x86-linux.
416
4170.6 Thu Nov 21 11:09:06 CET 2002
418 - made lazy context switching the default.
419 - fixed the bug where SemaphoreSet::guard
420 was the same as timed_guard and timed_guard was missing.
421 - fix a memleak in Coro::Timer.
422 - reclassified context sharing as not experimental.
423
4240.534 Sun Apr 14 03:05:12 CEST 2002
425 - fixed backspace => tab in header parsing (myhttpd).
426 - added eg/readline, for elmex.
427 - Coro::Event now calls ->start on first call instead of
428 ->again, which created wrong timeouts for the first ->next.
429 - fixed a bug where Coro::Socket returned a closed socket instead
430 of nothing on ECONNREFUSED and similar error conditions.
431
4320.533 Mon Feb 18 18:49:40 CET 2002
433 - adapt to finally working Event-0.85.
434 - fixed buggy event initialization (trapped under windows, but how
435 could it possibly have worked anywhere?)
436
4370.532 Wed Jan 16 02:45:32 CET 2002
438 - removed the reference to pp_entersub, might work on
439 (some) win32 perls now (testsuite works under cygwin,
440 without Event, which freezes).
441
4420.531 Mon Dec 10 22:18:44 CET 2001
443 - Fixed a bug in SemaphoreSet::timed_down.
444
4450.53 Tue Nov 27 21:11:13 CET 2001
446 - some tests for Event.
447 - slightly different internal architecture to get refcnt into the object
448 and not the reference. might fix a few bugs, certainly did introduce
449 new ones.
450 - Coro::Timer (independent of Event).
451 - new timed_wait functions for Coro::Signal, Semaphore, SemaphoreSet.
452
4530.52 Tue Nov 6 21:36:18 CET 2001
454 - ported to cygwin (trivial).
455 patch by Gerrit P. Haase <gp@familiehaase.de>.
456 - small setjmp code fix by Sullivan.DanielJ@epamail.epa.gov.
457
4580.51 Thu Nov 1 20:39:01 CET 2001
4 - terminate/cancel now work properly (otherwise termination 459 - terminate/cancel now work properly (otherwise termination
5 could cause a "next coroutine is not and contains not..." 460 could cause a "next coroutine is not and contains not..."
6 error. 461 error.
462 - added Coro::Socket::shutdown.
463 - Coro::Event::loop no is the same as Event::loop.
464 - implemented terminate with args + join.
7 465
80.5 Fri Sep 28 16:15:35 CEST 2001 4660.5 Fri Sep 28 16:15:35 CEST 2001
9 - fixed "print" on a Coro::Handle. The print method worked. 467 - fixed "print" on a Coro::Handle. The print method worked.
10 - small tweaks (seem to reduce memory consumption a lot) 468 - small tweaks (seem to reduce memory consumption a lot)
11 in various modules. 469 in various modules.
12 - splendid use of "no warnings" scattered throughout. 470 - splendid use of "no warnings" scattered throughout.
13 - added Coro::Handle::rbuf, fixed Coro::Handle::fh. 471 - added Coro::Handle::rbuf, fixed Coro::Handle::fh.

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines