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

Comparing Coro/Changes (file contents):
Revision 1.478 by root, Sat Dec 13 22:08:12 2008 UTC vs.
Revision 1.608 by root, Sat Dec 8 00:43:39 2012 UTC

1Revision history for Perl extension Coro. 1Revision history for Perl extension Coro.
2 2
3TODO: should explore PerlIO::coroaio (perl leaks like hell). 3TODO: should explore PerlIO::coroaio (perl leaks like hell).
4TODO: maybe implement a default message channel, very much like Erlang's 4TODO: channel->maxsize(newsize)?
5 actor model (which is cool in a lot of important aspects (failures!), 5TODO: __GCC_HAVE_DWARF2_CFI_ASM
6 but very lacking in others (higher level ipc)).
7 6
8new -> ready 7 - improved Coro::State documentation a bit.
9new -> canceled
10ready -> running
11ready -> canceled
12running -> ready
13running -> blocked
14running -> canceled
15blocked -> ready
16blocked -> canceled
17 8
185.13 96.23 Fri Dec 7 23:36:37 CET 2012
10 - use experimental fiber implementation on native windows
11 perls.
12 - use sizeof (void *) as multiplication factor for stack sizes,
13 to accomodate the totally braindamaged microsoft 64 bit "os".
14 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
15 activeperl 5.16 crashes when PerlIO_define_layer is called due
16 to some bug in the perl dll, strawberry perl at least passes
17 the testsuite.
18 - implement Coro::Handle->peeraddr/host/port, for slightly
19 improved compatibility with LWP.
20 - implement 5.17 compatibility by almost blindly applying a
21 good-looking patch by Father Chrysostomos.
22 - move stack management functions into libcoro 3.
23 - libcoro version 3 "released".
24 - support magic values as timed_io_once args.
25 - recommend AnyEvent 7+ or EV 4+, also require EV
26 version 4 or newer for Coro::EV.
27
286.10 Tue Oct 9 01:14:27 CEST 2012
29 - updated ecb.h, it had a typo that caused it to not compile on many
30 big endian systems (reported by many people).
31 - disable memory fences in ecb.h to improve portability.
32
336.09 Sat Oct 6 23:25:02 CEST 2012
34 - Coro::EV I/O watchers were not interruptible by exceptions
35 (Coro::State::throw) (testcase by sten).
36 - ->throw now puts threads into the ready queue, as this seems to
37 be expected by existing code, and code that doesn't cope with spurious
38 wakeups needs fixing anyway.
39 - use fd -1 in mmap.
40 - cast I32 to int in error message printf.
41 - warn about broken so-called "hardened" kernels.
42
436.08 Fri Apr 13 12:05:47 CEST 2012
44 - be more aggressive about exiting like perl does - formerly,
45 exiting from the non-main thread would not execute END blocks.
46
476.07 Fri Nov 11 21:21:48 CET 2011
48 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
49 - update ecb.h.
50
516.06 Mon Aug 8 23:59:48 CEST 2011
52 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
53 backend, which is an order of magnitude slower. unfortunately, cygwins
54 pthread implementation isn't very complete either, so allocate the stack
55 twice just to be sure.
56 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
57 in its ithreaded perl - best recompile cygwin and use the 'w'indows
58 backend for much better performance. also disable ithreads for
59 even better performance...).
60
616.05 Thu Aug 4 21:36:36 CEST 2011
62 - blush, condvar values would not be propagated from send to recv anymore
63 (reported by Chip Salzenberg).
64 - use exponential increase for the readline buffer length in
65 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
66
676.04 Wed Aug 3 17:15:45 CEST 2011
68 - use even more efficient and more compatible condvars for
69 compatibility to AnyEvent 6.x :)
70 - more inconsequential ecb.h updates.
71
726.03 Wed Aug 3 11:41:30 CEST 2011
73 - change how Coro patches AnyEvent condvars for compatibility to
74 AnyEvent 6.x.
75 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
76 hasn't been included.
77
786.02 Wed Jul 13 04:35:19 CEST 2011
79 - "improve portability to Gentoo" - gentoo manages to put perl variables
80 in memory areas that are farther than 2gb apart, which the jit couldn't
81 handle and barfed. now it's just a bit slower on gentoo and similar
82 systems.
83
846.01 Sun Jul 3 12:31:14 CEST 2011
85 - workarounds are good, but the test for whether pthreads are used
86 was not good. thisone should be better.
87 - check differently whether gcc generates cfi instructions itself.
88
896.0 Wed Jun 29 19:43:35 CEST 2011
90 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
91 destroyed and cleaned up automatically (e.g. async { schedule }).
92 - implement a JIT compiler for part of the thread switch code,
93 which gives a 50% speed improvement on threaded perls, and
94 about 4% on non-thraeded perls (so threaded perls now finally
95 reach about half the speed of non-threaded perls).
96 - slightly modernise Coro::Intro, add section about rouse functions.
97 - avoid DEFSV and ERRSV, giving another 10% improvement
98 in thread switching.
99 - Coro::State->is_destroyed is now called is_zombie.
100 - implement a Coro->safe_cancel method that might fail, but
101 cancels in a "safer" way if it succeeds.
102 - add preliminary support for DEBUGGING perls.
103 - get rid of two hash-accesses when initialising a new Coro - this
104 speeds up coro creation by almost a factor of two.
105 - croak when a coro that is being cancelled tries to block
106 (e.g. while executing a guard block), instead of crashing or
107 deadlocking.
108 - use a more robust and also faster method to identify Coro::State
109 objects - speeds up everything a bit.
110 - implement Coro->cancel in XS for a 20% speed improvement, and to
111 be able to implement mutual cancellation.
112 - speed up context switches by a percent or two by more efficiently
113 allocating context stack entries.
114 - implement Coro->join and Coro->on_destroy in XS for a speedup and
115 a reduction in memory use.
116 - cancelling a coro while it itself is cancelling another coro is
117 now supported and working, instead of triggering an assertion.
118 - be a bit more crash-resistant when calling (buggy) on_destroy
119 callbacks (best effort).
120 - move on_destroy into the slf_frame, to allow extension slf
121 functions to have destructors.
122 - get rid if coro refcounting - simply crash in other interpreter
123 threads by nulling the pointers on clone.
124 - simplify warn/die hook handling when loading Coro - the convoluted
125 logic seems to be no longer neccessary.
126 - use libecb instead of our own home-grown gcc hacks.
127 - document alternatives to Coro::LWP. Please use them :)
128 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
129 sigaltstack. Really. wine suffers from it, erlang suffers from it,
130 and it's known since at least 2006.
131
1325.372 Wed Feb 23 06:14:30 CET 2011
133 - apparently mingw doesn't provide a working gettimeofday, try to
134 work around that by relying on Time::HiRes (indirectly brought to
135 my attention by Max Maischein).
136 - fix some portability issues when Time::HiRes was used.
137
1385.371 Mon Feb 21 14:36:08 CET 2011
139 - backport to windows process emulation code again.
140
1415.37 Sat Feb 19 07:49:44 CET 2011
142 - add a big "Coro thread life cycle" section to "man Coro".
143 - try a tentative workaround against the breakage that 5.13 has
144 introduced without depreciation period. sigh.
145 - no longer use Time::HiRes if gettimeofday is available, which
146 saves quite a lot of memory.
147
1485.36 Sun Feb 13 05:33:41 CET 2011
149 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
150 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
151 method call.
152 - undocument Coro::Timer::sleep and obsolete whole module.
153 - optimise Coro::Timer::timeout memory and cpu usage.
154 - slightly updated Coro::Intro for recent changes.
155 - do not initialise PL_dirty anymore.
156
1575.25 Thu Nov 11 01:08:39 CET 2010
158 - try a different approach on netbsd - netbsd 5 finally has marginally
159 working pthreads, but still broken ucontext/sigaltstack.
160 - openbsd 4.8 finally got their act together, Coro works out of the box
161 with asm, setjmp and pthreads (no change, just informational).
162
1635.24 Sat Oct 23 11:27:12 CEST 2010
164 - port to the EV 4.0 API.
165 - work around bugs in mingw32, making strawberry perl work
166 out of the box.
167 - correctly modify Coro::AIO function prototypes
168 so that they reflect the "no optional parameters" rule.
169 - "ported" libcoro to C++.
170
1715.23 Mon May 17 18:50:42 CEST 2010
172 - be more resistant to ordering changes when initialising
173 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
174 Waldorf).
175 - document that perl 5.12 deliberately removed support for cloning.
176
1775.22 Wed Apr 14 03:55:35 CEST 2010
178 - correctly return udnef on errors in Coro::Handle::read/write
179 (testcase by Marc Mims).
180 - convert Coro::Util into a "perl compatibility wrapper" - the functions
181 are less useful now, but are drop-in replacements for existing
182 functions, listing better alternatives in the documentation. This also
183 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
184 with Coro::Util::inet_aton.
185 - do not override $Coro::idle unconditionally in Coro.pm, as other
186 modules could have provided their own idle coro already
187 (for exmaple, Coro::AnyEvent).
188 - fix Coro::Util::gethost* functions.
189 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
190
1915.21 Wed Dec 16 07:19:51 CET 2009
192 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
193 together.
194 - add some examples on how to combine other event loops with Coro in
195 Coro::AnyEvent, and how to run it (and not to block). Seems to be
196 the most common source of confusion.
197 - try to catch people naively blocking in an event callback.
198 - work around the perl filehandle bug issue in conjunction with
199 older common::sense (as indirectly pointed out by ZSystem).
200 - clarify the "not from signal handlers" section.
201
2025.2 Sun Oct 4 14:54:24 CEST 2009
203 - Coro::Storable destroyed the prototypes of the functions it wrapped.
204 - export rouse_cb and rouse_wait by default now.
205 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
206 - new method $state->swap_sv.
207 - added section on "windows process emulation" to the manpage,
208 after a not-so-fruitful (nor-friendly) "discussion" with chip
209 salzenberg (discussion implies arguments, but his only arguments
210 were ad-hominems, one wonders why he started it in the first
211 place). I hope this explains it well enough for him to understand,
212 and maybe well enough for others to understand.
213 - use common::sense everywhere now.
214 - idle callbacks are no longer supported, use idle coros instead.
215 - print a thread listing when a deadlock is detected.
216
2175.17 Sat Aug 22 23:09:31 CEST 2009
218 - work around a bug in the perl debugger causing crashes
219 when running under the debugger by marking _pool_handler as nodebug.
220 - speed up Coro::async considerably.
221 - try some hacks to get netbsd to work "more often" - their broken
222 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
223 - convert Coro to AE 5.0 API.
224
2255.162 Tue Jul 28 04:04:03 CEST 2009
226 - perl 5.8.2 is now minimum requirement.
227 - skip t/19_handle.t on broken windows perls.
228
2295.161 Wed Jul 22 04:47:38 CEST 2009
230 - Coro::AnyEvent::poll could have a different prototype when EV was
231 used as backend (analyzed by Tatsuhiko Miyagawa).
232 - Coro::AnyEvent errornously initialised the event loop when loaded,
233 not on demand.
234 - try to workaround rare */t/01_unblock.t failures.
235
2365.16 Tue Jul 21 01:44:37 CEST 2009
237 - Coro::AnyEvent failed to hook into the event loop
238 when no threads had been readied between detecting
239 the event loop and actually running it.
240 - considerably speed up Coro::Select by taking avdantage
241 of AnyEvent > 4.8 and some other optimisations.
242 - implement paragraph readline mode in Coro::Handle
243 (based on patches by Zsbán Ambrus).
244 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
245 and analyzed by Yasuhiro MATSUMOTO).
246 - clarified libcoro license and copyright.
247 - someone stole my EXTRA_META!!!
248 - implement Coro::Select::patch_pp_sselect and it's brother,
249 for hardcode select overriding.
250
2515.151 Mon Jul 6 05:41:57 CEST 2009
252 - backport to windows process emulation code again (patch by
253 Yasuhiro MATSUMOTO).
254 - slightly update Coro::MakeMaker.
255
2565.15 Tue Jun 30 10:28:06 CEST 2009
257 - deprecate Coro::Socket, document how to get ipv6 support via
258 AnyEvent::Socket instead.
259 - implement signal->wait ($cb) interface, similar to semaphores.
260 - work around SvOK not supporting getmagic, so we have to getmagic
261 to test for undef :/ (reported by Matthias Waldorf).
262 - load Coro::AnyEvent in all modules using AnyEvent.
263 - work around perl corrupting our internal data structures,
264 reported by Tokuhiro Matsuno.
265 - enable per-coroutine real and cpu time gathering
266 (Coro::State::enable_times).
267
2685.14 Wed Jun 24 01:37:48 CEST 2009
269 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
270 as this seems to have been a difficult concept (poll, sleep, idle,
271 idle_upto).
272 - add Coro::AnyEvent::readable/writable functions.
273 - clarify Coro::EV/Event/AnyEvent manpages.
274 - free per-thread global scalars in the thread calling ->cancel, to
275 avoid crashes when $_, $@ etc., are magical but some of those
276 had already been freed.
277 - "unexperimentalise" the callback interface for Coro::Semaphore.
278 - speed up ready queue management/context switching by using a linked
279 list instead of an array (~5-10%).
280 - implement "watch" command in Coro::Debug shells.
281 - for fun, implement time-slicing as an example in the manpage.
282 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
283 Coro::Event, use the normal AnyEvent handling instead of dieing
284 (the same is true for Coro::Handle).
285 - properly document Coro::EV::timed_io_once.
286 - avoid unneccessary ->cancel calls in Coro::Handle.
287 - maybe make it work on mingw32 with win32 backend
288 (based on patch by Yasuhiro Matsumoto).
289
2905.132 Fri May 29 09:00:39 CEST 2009
291 - do not keep a reference to the argument itself in
292 Coro::Semaphore::guard, as it could change later.
293 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
294 should support a prepare callback.
295
2965.131 Mon Mar 16 23:20:37 CET 2009
297 - implement and document Coro->suspend, Coro->resume.
298 - fix Coro::Select implementation to not (often) close
299 the passed file descriptors (testcase provided by pippijn).
300
3015.13 Mon Dec 15 21:51:42 CET 2008
302 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
19 - don't set diehook to C<undef> but instead to NULL, to avoid 303 - don't set diehook to C<undef> but instead to NULL, to avoid
20 spurious warnings. 304 spurious warnings.
21 - fix a lot of bugs in Coro::SemaphoreSet. 305 - fix a lot of bugs in Coro::SemaphoreSet.
22 - Coro::SemaphoreSet will less often create a semaphore needlessly. 306 - Coro::SemaphoreSet will less often create a semaphore needlessly.
23 - add Coro::SemaphoreSet::count and wait methods. 307 - add Coro::SemaphoreSet::count and wait methods.
24 - take advantage of the new Guard module. 308 - take advantage of the new Guard module.
25 - deprecate Coro::guard. 309 - deprecate Coro::guard.
310 - try to fix the dreaded 01_unblock tests once more. I hate it when
311 testsuites need more fixing than the code they are supposed to test.
312 - croak in more cases when a required callback isn't resolvable.
313 - fix some minor issues in Coro::State->call/eval.
314 - use current coroutine context instead of a temporary one
315 when temporarily switching to another coroutine.
316 - do not call C-level on_destroy handlers during global destruction,
317 to avoid needless segfaults.
26 318
275.12 Sun Dec 7 13:30:38 CET 2008 3195.12 Sun Dec 7 13:30:38 CET 2008
28 - add default config for MirOS, which seems to be a bug-to-bug 320 - add default config for MirOS, which seems to be a bug-to-bug
29 compatible fork of openbsd ("world domination by releasing 321 compatible fork of openbsd ("world domination by releasing
30 openbsd cvs before the openbsd folks do it" or so :). 322 openbsd cvs before the openbsd folks do it" or so :).

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines