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

Comparing Coro/Changes (file contents):
Revision 1.484 by root, Mon Dec 15 20:52:04 2008 UTC vs.
Revision 1.609 by root, Thu Dec 13 00:21:11 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)). 6
7 - improved Coro::State documentation a bit.
8 - Coro::Debug::command now flushes the output.
9
106.23 Fri Dec 7 23:36:37 CET 2012
11 - use experimental fiber implementation on native windows
12 perls.
13 - use sizeof (void *) as multiplication factor for stack sizes,
14 to accomodate the totally braindamaged microsoft 64 bit "os".
15 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
16 activeperl 5.16 crashes when PerlIO_define_layer is called due
17 to some bug in the perl dll, strawberry perl at least passes
18 the testsuite.
19 - implement Coro::Handle->peeraddr/host/port, for slightly
20 improved compatibility with LWP.
21 - implement 5.17 compatibility by almost blindly applying a
22 good-looking patch by Father Chrysostomos.
23 - move stack management functions into libcoro 3.
24 - libcoro version 3 "released".
25 - support magic values as timed_io_once args.
26 - recommend AnyEvent 7+ or EV 4+, also require EV
27 version 4 or newer for Coro::EV.
28
296.10 Tue Oct 9 01:14:27 CEST 2012
30 - updated ecb.h, it had a typo that caused it to not compile on many
31 big endian systems (reported by many people).
32 - disable memory fences in ecb.h to improve portability.
33
346.09 Sat Oct 6 23:25:02 CEST 2012
35 - Coro::EV I/O watchers were not interruptible by exceptions
36 (Coro::State::throw) (testcase by sten).
37 - ->throw now puts threads into the ready queue, as this seems to
38 be expected by existing code, and code that doesn't cope with spurious
39 wakeups needs fixing anyway.
40 - use fd -1 in mmap.
41 - cast I32 to int in error message printf.
42 - warn about broken so-called "hardened" kernels.
43
446.08 Fri Apr 13 12:05:47 CEST 2012
45 - be more aggressive about exiting like perl does - formerly,
46 exiting from the non-main thread would not execute END blocks.
47
486.07 Fri Nov 11 21:21:48 CET 2011
49 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
50 - update ecb.h.
51
526.06 Mon Aug 8 23:59:48 CEST 2011
53 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
54 backend, which is an order of magnitude slower. unfortunately, cygwins
55 pthread implementation isn't very complete either, so allocate the stack
56 twice just to be sure.
57 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
58 in its ithreaded perl - best recompile cygwin and use the 'w'indows
59 backend for much better performance. also disable ithreads for
60 even better performance...).
61
626.05 Thu Aug 4 21:36:36 CEST 2011
63 - blush, condvar values would not be propagated from send to recv anymore
64 (reported by Chip Salzenberg).
65 - use exponential increase for the readline buffer length in
66 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
67
686.04 Wed Aug 3 17:15:45 CEST 2011
69 - use even more efficient and more compatible condvars for
70 compatibility to AnyEvent 6.x :)
71 - more inconsequential ecb.h updates.
72
736.03 Wed Aug 3 11:41:30 CEST 2011
74 - change how Coro patches AnyEvent condvars for compatibility to
75 AnyEvent 6.x.
76 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
77 hasn't been included.
78
796.02 Wed Jul 13 04:35:19 CEST 2011
80 - "improve portability to Gentoo" - gentoo manages to put perl variables
81 in memory areas that are farther than 2gb apart, which the jit couldn't
82 handle and barfed. now it's just a bit slower on gentoo and similar
83 systems.
84
856.01 Sun Jul 3 12:31:14 CEST 2011
86 - workarounds are good, but the test for whether pthreads are used
87 was not good. thisone should be better.
88 - check differently whether gcc generates cfi instructions itself.
89
906.0 Wed Jun 29 19:43:35 CEST 2011
91 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
92 destroyed and cleaned up automatically (e.g. async { schedule }).
93 - implement a JIT compiler for part of the thread switch code,
94 which gives a 50% speed improvement on threaded perls, and
95 about 4% on non-thraeded perls (so threaded perls now finally
96 reach about half the speed of non-threaded perls).
97 - slightly modernise Coro::Intro, add section about rouse functions.
98 - avoid DEFSV and ERRSV, giving another 10% improvement
99 in thread switching.
100 - Coro::State->is_destroyed is now called is_zombie.
101 - implement a Coro->safe_cancel method that might fail, but
102 cancels in a "safer" way if it succeeds.
103 - add preliminary support for DEBUGGING perls.
104 - get rid of two hash-accesses when initialising a new Coro - this
105 speeds up coro creation by almost a factor of two.
106 - croak when a coro that is being cancelled tries to block
107 (e.g. while executing a guard block), instead of crashing or
108 deadlocking.
109 - use a more robust and also faster method to identify Coro::State
110 objects - speeds up everything a bit.
111 - implement Coro->cancel in XS for a 20% speed improvement, and to
112 be able to implement mutual cancellation.
113 - speed up context switches by a percent or two by more efficiently
114 allocating context stack entries.
115 - implement Coro->join and Coro->on_destroy in XS for a speedup and
116 a reduction in memory use.
117 - cancelling a coro while it itself is cancelling another coro is
118 now supported and working, instead of triggering an assertion.
119 - be a bit more crash-resistant when calling (buggy) on_destroy
120 callbacks (best effort).
121 - move on_destroy into the slf_frame, to allow extension slf
122 functions to have destructors.
123 - get rid if coro refcounting - simply crash in other interpreter
124 threads by nulling the pointers on clone.
125 - simplify warn/die hook handling when loading Coro - the convoluted
126 logic seems to be no longer neccessary.
127 - use libecb instead of our own home-grown gcc hacks.
128 - document alternatives to Coro::LWP. Please use them :)
129 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
130 sigaltstack. Really. wine suffers from it, erlang suffers from it,
131 and it's known since at least 2006.
132
1335.372 Wed Feb 23 06:14:30 CET 2011
134 - apparently mingw doesn't provide a working gettimeofday, try to
135 work around that by relying on Time::HiRes (indirectly brought to
136 my attention by Max Maischein).
137 - fix some portability issues when Time::HiRes was used.
138
1395.371 Mon Feb 21 14:36:08 CET 2011
140 - backport to windows process emulation code again.
141
1425.37 Sat Feb 19 07:49:44 CET 2011
143 - add a big "Coro thread life cycle" section to "man Coro".
144 - try a tentative workaround against the breakage that 5.13 has
145 introduced without depreciation period. sigh.
146 - no longer use Time::HiRes if gettimeofday is available, which
147 saves quite a lot of memory.
148
1495.36 Sun Feb 13 05:33:41 CET 2011
150 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
151 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
152 method call.
153 - undocument Coro::Timer::sleep and obsolete whole module.
154 - optimise Coro::Timer::timeout memory and cpu usage.
155 - slightly updated Coro::Intro for recent changes.
156 - do not initialise PL_dirty anymore.
157
1585.25 Thu Nov 11 01:08:39 CET 2010
159 - try a different approach on netbsd - netbsd 5 finally has marginally
160 working pthreads, but still broken ucontext/sigaltstack.
161 - openbsd 4.8 finally got their act together, Coro works out of the box
162 with asm, setjmp and pthreads (no change, just informational).
163
1645.24 Sat Oct 23 11:27:12 CEST 2010
165 - port to the EV 4.0 API.
166 - work around bugs in mingw32, making strawberry perl work
167 out of the box.
168 - correctly modify Coro::AIO function prototypes
169 so that they reflect the "no optional parameters" rule.
170 - "ported" libcoro to C++.
171
1725.23 Mon May 17 18:50:42 CEST 2010
173 - be more resistant to ordering changes when initialising
174 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
175 Waldorf).
176 - document that perl 5.12 deliberately removed support for cloning.
177
1785.22 Wed Apr 14 03:55:35 CEST 2010
179 - correctly return udnef on errors in Coro::Handle::read/write
180 (testcase by Marc Mims).
181 - convert Coro::Util into a "perl compatibility wrapper" - the functions
182 are less useful now, but are drop-in replacements for existing
183 functions, listing better alternatives in the documentation. This also
184 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
185 with Coro::Util::inet_aton.
186 - do not override $Coro::idle unconditionally in Coro.pm, as other
187 modules could have provided their own idle coro already
188 (for exmaple, Coro::AnyEvent).
189 - fix Coro::Util::gethost* functions.
190 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
191
1925.21 Wed Dec 16 07:19:51 CET 2009
193 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
194 together.
195 - add some examples on how to combine other event loops with Coro in
196 Coro::AnyEvent, and how to run it (and not to block). Seems to be
197 the most common source of confusion.
198 - try to catch people naively blocking in an event callback.
199 - work around the perl filehandle bug issue in conjunction with
200 older common::sense (as indirectly pointed out by ZSystem).
201 - clarify the "not from signal handlers" section.
202
2035.2 Sun Oct 4 14:54:24 CEST 2009
204 - Coro::Storable destroyed the prototypes of the functions it wrapped.
205 - export rouse_cb and rouse_wait by default now.
206 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
207 - new method $state->swap_sv.
208 - added section on "windows process emulation" to the manpage,
209 after a not-so-fruitful (nor-friendly) "discussion" with chip
210 salzenberg (discussion implies arguments, but his only arguments
211 were ad-hominems, one wonders why he started it in the first
212 place). I hope this explains it well enough for him to understand,
213 and maybe well enough for others to understand.
214 - use common::sense everywhere now.
215 - idle callbacks are no longer supported, use idle coros instead.
216 - print a thread listing when a deadlock is detected.
217
2185.17 Sat Aug 22 23:09:31 CEST 2009
219 - work around a bug in the perl debugger causing crashes
220 when running under the debugger by marking _pool_handler as nodebug.
221 - speed up Coro::async considerably.
222 - try some hacks to get netbsd to work "more often" - their broken
223 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
224 - convert Coro to AE 5.0 API.
225
2265.162 Tue Jul 28 04:04:03 CEST 2009
227 - perl 5.8.2 is now minimum requirement.
228 - skip t/19_handle.t on broken windows perls.
229
2305.161 Wed Jul 22 04:47:38 CEST 2009
231 - Coro::AnyEvent::poll could have a different prototype when EV was
232 used as backend (analyzed by Tatsuhiko Miyagawa).
233 - Coro::AnyEvent errornously initialised the event loop when loaded,
234 not on demand.
235 - try to workaround rare */t/01_unblock.t failures.
236
2375.16 Tue Jul 21 01:44:37 CEST 2009
238 - Coro::AnyEvent failed to hook into the event loop
239 when no threads had been readied between detecting
240 the event loop and actually running it.
241 - considerably speed up Coro::Select by taking avdantage
242 of AnyEvent > 4.8 and some other optimisations.
243 - implement paragraph readline mode in Coro::Handle
244 (based on patches by Zsbán Ambrus).
245 - replace WSAEINPROGRESS by WSAEWOULDBLOCK (reported
246 and analyzed by Yasuhiro MATSUMOTO).
247 - clarified libcoro license and copyright.
248 - someone stole my EXTRA_META!!!
249 - implement Coro::Select::patch_pp_sselect and it's brother,
250 for hardcode select overriding.
251
2525.151 Mon Jul 6 05:41:57 CEST 2009
253 - backport to windows process emulation code again (patch by
254 Yasuhiro MATSUMOTO).
255 - slightly update Coro::MakeMaker.
256
2575.15 Tue Jun 30 10:28:06 CEST 2009
258 - deprecate Coro::Socket, document how to get ipv6 support via
259 AnyEvent::Socket instead.
260 - implement signal->wait ($cb) interface, similar to semaphores.
261 - work around SvOK not supporting getmagic, so we have to getmagic
262 to test for undef :/ (reported by Matthias Waldorf).
263 - load Coro::AnyEvent in all modules using AnyEvent.
264 - work around perl corrupting our internal data structures,
265 reported by Tokuhiro Matsuno.
266 - enable per-coroutine real and cpu time gathering
267 (Coro::State::enable_times).
268
2695.14 Wed Jun 24 01:37:48 CEST 2009
270 - provide explicit functions to "cede" to the event loop in Coro::AnyEvent,
271 as this seems to have been a difficult concept (poll, sleep, idle,
272 idle_upto).
273 - add Coro::AnyEvent::readable/writable functions.
274 - clarify Coro::EV/Event/AnyEvent manpages.
275 - free per-thread global scalars in the thread calling ->cancel, to
276 avoid crashes when $_, $@ etc., are magical but some of those
277 had already been freed.
278 - "unexperimentalise" the callback interface for Coro::Semaphore.
279 - speed up ready queue management/context switching by using a linked
280 list instead of an array (~5-10%).
281 - implement "watch" command in Coro::Debug shells.
282 - for fun, implement time-slicing as an example in the manpage.
283 - if AnyEvent detects EV or Event, but we don't have Coro::EV or
284 Coro::Event, use the normal AnyEvent handling instead of dieing
285 (the same is true for Coro::Handle).
286 - properly document Coro::EV::timed_io_once.
287 - avoid unneccessary ->cancel calls in Coro::Handle.
288 - maybe make it work on mingw32 with win32 backend
289 (based on patch by Yasuhiro Matsumoto).
290
2915.132 Fri May 29 09:00:39 CEST 2009
292 - do not keep a reference to the argument itself in
293 Coro::Semaphore::guard, as it could change later.
294 - support SO_RCVBUF/SO_SNDBUF nonstandard Coro::Socket options,
295 should support a prepare callback.
296
2975.131 Mon Mar 16 23:20:37 CET 2009
298 - implement and document Coro->suspend, Coro->resume.
299 - fix Coro::Select implementation to not (often) close
300 the passed file descriptors (testcase provided by pippijn).
7 301
85.13 Mon Dec 15 21:51:42 CET 2008 3025.13 Mon Dec 15 21:51:42 CET 2008
9 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave). 303 - EXPERIMENTAL: implement dynamic winds (on_enter/on_leave).
10 - don't set diehook to C<undef> but instead to NULL, to avoid 304 - don't set diehook to C<undef> but instead to NULL, to avoid
11 spurious warnings. 305 spurious warnings.

Diff Legend

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