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

Comparing Coro/Changes (file contents):
Revision 1.565 by root, Fri May 13 19:39:26 2011 UTC vs.
Revision 1.662 by root, Sun Jun 26 21:46:03 2016 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: channel->maxsize(newsize)? 4TODO: channel->maxsize(newsize)?
5TODO: http://www.microsoft.com/msj/archive/s2ce.aspx 5TODO: __GCC_HAVE_DWARF2_CFI_ASM
6TODO: swap_sv, maybe add scope_swap_sv?
7TODO: croak when async_pool tries to run canceled thread?
6 8
7TODO: tutorial 96.511 Sun Jun 26 23:44:50 CEST 2016
10 - make it compile with threaded perls again.
11 - simplify and speed up __DIE__ and __WARN__ handling.
12
136.51 Sat Jun 25 20:57:20 CEST 2016
14 - 6.5 release didn't compile on most older perls due to
15 a logic error enabling experimental 5.24 code for all versions.
16 - since various broken or deficient patches float around
17 to work around the vtbl API breakage in perl, let's provide
18 our own workaround, which seems to be much less invasive
19 and more compatible than the approaches seen so far. YMMV.
20 - work around assertion failure in perl_destruct on 5.24
21 (analyzed by Dave Mitchell).
22
236.5 Wed Jun 22 22:23:50 CEST 2016
24 - swap_sv swapping order was not symmetrical, causing
25 wrong swaps when swap_sv was used multiple times on the
26 same sv in the same thread.
27 - swap_sv calls can now be undone by calling it again
28 with the same variables.
29 - swap_sv calls will now be undone in async_pool threads.
30 - split Coro::Semaphore::up/adjust into separate xs functions
31 for better error reporting, at slight codesize increase.
32 - (libcoro) arm assembly support, please test and report.
33 - adjust to PL_savestack changes in perl 5.24 (adapted from
34 the debian patch, which unfortunately gets it wrong).
35
366.49 Sat Oct 17 01:40:12 CEST 2015
37 - throwing an exception to a thread waiting in
38 Coro::Handle using Coro::EV did not stop the watchers,
39 causing the next call to fail (testcase by Martin Pritchard).
40 - bump minimum perl version to 5.10.
41
426.48 Sun Oct 4 19:03:51 CEST 2015
43 - fix memory leak when Coro::Handle uses Coro::EV internally
44 (testcase by Sten Sten).
45 - update code to libev 4 API (internally, Coro still used the
46 version 3 API calls).
47 - remove unused hv_sig.
48
496.47 Sat Jul 11 03:58:20 CEST 2015
50 - swap_svs - NVs can be in the SV head beginning with 5.20.
51
526.46 Tue Jun 30 14:36:30 CEST 2015
53 - restore portability to perls with windows fork emulation
54 (patch by Petr Písař).
55
566.45 Tue Jun 30 01:40:08 CEST 2015
57 - restore portability to perl 5.16 and below.
58
596.44 Tue Jun 30 00:41:54 CEST 2015
60 - avoid segfaulting (or worse) when tracing a canceled thread.
61 - the Event module silently broke it's public hook API, causing
62 failures on perls compiled with -Duselongdouble. Requiring
63 the latest version of Event because I don't know which release
64 changed this (it's not mentioned in the ChangeLog).
65 - allow xs level enterleave hooks via CoroAPI.
66
676.43 Thu Jun 4 15:38:14 CEST 2015
68 - use stability canary.
69 - port to stableperl-5.22.0-1.001.
70 - update libecb.
71
726.42 Wed Feb 11 20:29:52 CET 2015
73 - Coro::SemaphoreSet->try did not actually work (analyzed by
74 SATO Kentaro).
75 - upgrade libecb to be C11 compliant.
76
776.41 Sat Sep 6 22:08:46 CEST 2014
78 - restore portability to perl 5.8.x.
79 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
80 up some memory, but fixes "use feature" and similar modules
81 when used inside a Coro.
82 - allow zero as argument to Coro::Channel to mean the same thing
83 as no argument. this works with older versions as well,
84 but wasn't legal until now.
85 - slightly better c header file detection.
86
876.39 Mon Jun 2 00:00:08 CEST 2014
88 - work around more incompatible changes in 5.20.
89
906.38 Sun Jun 1 21:54:23 CEST 2014
91 - check that perl slots actually have enough space to hold
92 interpreter variables.
93 - untested port to perl 5.19 (64 bit tmps indices) (reported
94 by Andreas König).
95 - croak when cancel is called without a thread context.
96
976.37 Tue Mar 4 13:27:33 CET 2014
98 - *sigh*, removed leftover debugging code from debugging a
99 perl bug, of all things.
100
1016.36 Tue Mar 4 07:11:59 CET 2014
102 - semaphores would not clear the destroy hook when interrupted
103 by ->throw, causing segfaults or worse.
104 - ->throw on a thread waiting for a semaphore did not acquire
105 the semaphore, but also didn't wake up other waiters,
106 possibly causing a deadlock.
107 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
108 to report a full stacktrace, which should help find out
109 where the actual call is.
110 - "atomically" destroy data in slf_destroy, because it is
111 the right thing to do, just in case.
112 - disable ecb.h libm dependency, to compile on ancient systems
113 or under adverse conditions.
114
1156.33 Mon Nov 18 11:26:27 CET 2013
116 - do not crash when freeing padlists with holes (in 5.18).
117 - tentative SVt_BIND 5.19 port/fix.
118
1196.32 Tue Nov 5 15:35:35 CET 2013
120 - use a new algorithm to derive padlists for perl 5.18. The old
121 one could lead to 0-pointer accesses inside perl (reported
122 by Darin McBride).
123
1246.31 Thu May 9 07:39:48 CEST 2013
125 - Coro::AIO requests would crash if the thread was ready'd
126 while the request was ongoing.
127
1286.29 Wed May 8 02:55:18 CEST 2013
129 - when an on_destroy handler destructs the coro currently being
130 destructed a perl scalar could be accessed after being freed,
131 likely causing a crash.
132
1336.28 Wed Mar 6 06:58:02 CET 2013
134 - clean remnants of existing __DIE__ and __WARN__ handlers so
135 they lose their magic and will not cause segfaults later
136 (testcase by Andrey Sagulin).
137 - improved Coro::State documentation a bit.
138 - Coro::Debug::command now flushes the output.
139 - add hack detection code for x32 abi, because the braindead slugs
140 who designed that made it look exactly like x86_64 without
141 providing proper compile time symbols to test for it. as a result,
142 this detection cannot work reliably.
143 - valgrind stack registering was broken.
144 - do not rely on Time::HiRes anymore in Coro::Debug.
145
1466.23 Fri Dec 7 23:36:37 CET 2012
147 - use experimental fiber implementation on native windows
148 perls.
149 - use sizeof (void *) as multiplication factor for stack sizes,
150 to accomodate the totally braindamaged microsoft 64 bit "os".
151 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
152 activeperl 5.16 crashes when PerlIO_define_layer is called due
153 to some bug in the perl dll, strawberry perl at least passes
154 the testsuite.
155 - implement Coro::Handle->peeraddr/host/port, for slightly
156 improved compatibility with LWP.
157 - implement 5.17 compatibility by almost blindly applying a
158 good-looking patch by Father Chrysostomos.
159 - move stack management functions into libcoro 3.
160 - libcoro version 3 "released".
161 - support magic values as timed_io_once args.
162 - recommend AnyEvent 7+ or EV 4+, also require EV
163 version 4 or newer for Coro::EV.
164
1656.10 Tue Oct 9 01:14:27 CEST 2012
166 - updated ecb.h, it had a typo that caused it to not compile on many
167 big endian systems (reported by many people).
168 - disable memory fences in ecb.h to improve portability.
169
1706.09 Sat Oct 6 23:25:02 CEST 2012
171 - Coro::EV I/O watchers were not interruptible by exceptions
172 (Coro::State::throw) (testcase by sten).
173 - ->throw now puts threads into the ready queue, as this seems to
174 be expected by existing code, and code that doesn't cope with spurious
175 wakeups needs fixing anyway.
176 - use fd -1 in mmap.
177 - cast I32 to int in error message printf.
178 - warn about broken so-called "hardened" kernels.
179
1806.08 Fri Apr 13 12:05:47 CEST 2012
181 - be more aggressive about exiting like perl does - formerly,
182 exiting from the non-main thread would not execute END blocks.
183
1846.07 Fri Nov 11 21:21:48 CET 2011
185 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
186 - update ecb.h.
187
1886.06 Mon Aug 8 23:59:48 CEST 2011
189 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
190 backend, which is an order of magnitude slower. unfortunately, cygwins
191 pthread implementation isn't very complete either, so allocate the stack
192 twice just to be sure.
193 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
194 in its ithreaded perl - best recompile cygwin and use the 'w'indows
195 backend for much better performance. also disable ithreads for
196 even better performance...).
197
1986.05 Thu Aug 4 21:36:36 CEST 2011
199 - blush, condvar values would not be propagated from send to recv anymore
200 (reported by Chip Salzenberg).
201 - use exponential increase for the readline buffer length in
202 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
203
2046.04 Wed Aug 3 17:15:45 CEST 2011
205 - use even more efficient and more compatible condvars for
206 compatibility to AnyEvent 6.x :)
207 - more inconsequential ecb.h updates.
208
2096.03 Wed Aug 3 11:41:30 CEST 2011
210 - change how Coro patches AnyEvent condvars for compatibility to
211 AnyEvent 6.x.
212 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
213 hasn't been included.
214
2156.02 Wed Jul 13 04:35:19 CEST 2011
216 - "improve portability to Gentoo" - gentoo manages to put perl variables
217 in memory areas that are farther than 2gb apart, which the jit couldn't
218 handle and barfed. now it's just a bit slower on gentoo and similar
219 systems.
220
2216.01 Sun Jul 3 12:31:14 CEST 2011
222 - workarounds are good, but the test for whether pthreads are used
223 was not good. this one should be better.
224 - check differently whether gcc generates cfi instructions itself.
225
2266.0 Wed Jun 29 19:43:35 CEST 2011
8 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be 227 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
9 destroyed and cleaned up automatically (e.g. async { schedule }). 228 destroyed and cleaned up automatically (e.g. async { schedule }).
229 - implement a JIT compiler for part of the thread switch code,
230 which gives a 50% speed improvement on threaded perls, and
231 about 4% on non-threaded perls (so threaded perls now finally
232 reach about half the speed of non-threaded perls).
233 - slightly modernise Coro::Intro, add section about rouse functions.
234 - avoid DEFSV and ERRSV, giving another 10% improvement
235 in thread switching.
10 - Coro::State->is_destroyed is now called is_zombie. 236 - Coro::State->is_destroyed is now called is_zombie.
11 - implement a Coro->safe_cancel method that might fail, but 237 - implement a Coro->safe_cancel method that might fail, but
12 cancels in a "safer" way if it succeeds. 238 cancels in a "safer" way if it succeeds.
239 - add preliminary support for DEBUGGING perls.
240 - get rid of two hash-accesses when initialising a new Coro - this
241 speeds up coro creation by almost a factor of two.
13 - croak when a coro that is being cancelled tries to block 242 - croak when a coro that is being cancelled tries to block
14 (e.g. while executing a guard block), instead of crashing or 243 (e.g. while executing a guard block), instead of crashing or
15 deadlocking. 244 deadlocking.
16 - use a more robust and also faster method to identify Coro::State 245 - use a more robust and also faster method to identify Coro::State
17 objects - speeds up everything a bit. 246 objects - speeds up everything a bit.
… …
27 callbacks (best effort). 256 callbacks (best effort).
28 - move on_destroy into the slf_frame, to allow extension slf 257 - move on_destroy into the slf_frame, to allow extension slf
29 functions to have destructors. 258 functions to have destructors.
30 - get rid if coro refcounting - simply crash in other interpreter 259 - get rid if coro refcounting - simply crash in other interpreter
31 threads by nulling the pointers on clone. 260 threads by nulling the pointers on clone.
261 - simplify warn/die hook handling when loading Coro - the convoluted
262 logic seems to be no longer neccessary.
263 - use libecb instead of our own home-grown gcc hacks.
264 - document alternatives to Coro::LWP. Please use them :)
265 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
266 sigaltstack. Really. wine suffers from it, erlang suffers from it,
267 and it's known since at least 2006.
32 268
335.372 Wed Feb 23 06:14:30 CET 2011 2695.372 Wed Feb 23 06:14:30 CET 2011
34 - apparently mingw doesn't provide a working gettimeofday, try to 270 - apparently mingw doesn't provide a working gettimeofday, try to
35 work around that by relying on Time::HiRes (indirectly brought to 271 work around that by relying on Time::HiRes (indirectly brought to
36 my attention by Max Maischein). 272 my attention by Max Maischein).
… …
354 - ->throw is now supported on Coro::State objects. 590 - ->throw is now supported on Coro::State objects.
355 - clean up cctx creation code a bit. 591 - clean up cctx creation code a bit.
356 - entersub is actually a UNOP, not a LOGOP (not a bugfix). 592 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
357 593
3584.9 Sat Nov 8 17:45:27 CET 2008 5944.9 Sat Nov 8 17:45:27 CET 2008
359 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 595 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
360 - (libcoro) no longer leak threads in the experimental pthread backend, 596 - (libcoro) no longer leak threads in the experimental pthread backend,
361 also speed it up considerably. 597 also speed it up considerably.
362 - (libcoro) do not rely on makecontext passing void *'s unscathed. 598 - (libcoro) do not rely on makecontext passing void *'s unscathed.
363 - fix compiletime dependencies on libcoro in the Makefile. 599 - fix compiletime dependencies on libcoro in the Makefile.
364 - cctx_count wasn't always updated properly. 600 - cctx_count wasn't always updated properly.
… …
426 - remove debugging code related to MgPV_nolen_const, also try to 662 - remove debugging code related to MgPV_nolen_const, also try to
427 make it compile with perl 5.8.6 (yes, apple apparently loves 663 make it compile with perl 5.8.6 (yes, apple apparently loves
428 outdated software). Reported by John S. 664 outdated software). Reported by John S.
429 665
4304.744 Tue Jul 8 22:06:35 CEST 2008 6664.744 Tue Jul 8 22:06:35 CEST 2008
431 - correctly provide default DIE/WARN handlers as documented. 667 - correctly provide default DIE/WARN handlers as documented.
432 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 668 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
433 implementations inside perl work fine for us. 669 implementations inside perl work fine for us.
434 670
4354.743 Mon Jun 16 00:21:57 CEST 2008 6714.743 Mon Jun 16 00:21:57 CEST 2008
436 - when using Coro::EV without running EV::loop it could 672 - when using Coro::EV without running EV::loop it could
… …
458 - sprinkle "no warnings" freely over everything, also suppress 694 - sprinkle "no warnings" freely over everything, also suppress
459 warnings for some other modules. 695 warnings for some other modules.
460 - fix typo in WSAEWOULDBLOCK. 696 - fix typo in WSAEWOULDBLOCK.
461 697
4624.72 Sun May 25 05:14:36 CEST 2008 6984.72 Sun May 25 05:14:36 CEST 2008
463 - tweak META.yaml a bit, unfortunately, there is no documented way 699 - tweak META.yaml a bit, unfortunately, there is no documented way
464 to have optional dependencies with CPAN. doh :( 700 to have optional dependencies with CPAN. doh :(
465 - avoid running some tests on windows because they would fail due to 701 - avoid running some tests on windows because they would fail due to
466 perl bug (broken fork, broken pipes...). 702 perl bug (broken fork, broken pipes...).
467 - work around perl on windows bugs where perl returns undocumented 703 - work around perl on windows bugs where perl returns undocumented
468 error codes for sysread, syswrite etc. by taking advantage 704 error codes for sysread, syswrite etc. by taking advantage
… …
472 - use unix domain sockets in testsuite to work around 708 - use unix domain sockets in testsuite to work around
473 common perl implementation bugs on widows (they are emulated by 709 common perl implementation bugs on widows (they are emulated by
474 tcp sockets on windows. ugh.) 710 tcp sockets on windows. ugh.)
475 711
4764.71 Sat May 24 20:01:27 CEST 2008 7124.71 Sat May 24 20:01:27 CEST 2008
477 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 713 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
478 - take advantage of async name resolution of AnyEvent::Util. 714 - take advantage of async name resolution of AnyEvent::Util.
479 - work around brutal inet_aton override in Coro::LWP. 715 - work around brutal inet_aton override in Coro::LWP.
480 - take advantage of the readyhook in Coro::EV, for smoother 716 - take advantage of the readyhook in Coro::EV, for smoother
481 scheduling. 717 scheduling.
482 718
4834.7 Sun May 11 00:32:19 CEST 2008 7194.7 Sun May 11 00:32:19 CEST 2008
484 - completely reworked the Coro manpage. 720 - completely reworked the Coro manpage.
485 - added Coro::AnyEvent, generic event loop integration. 721 - added Coro::AnyEvent, generic event loop integration.
486 - implement cancel, ready and kill commands in Coro::Debug. 722 - implement cancel, ready and kill commands in Coro::Debug.
487 - document find_coro in Coro::Debug. 723 - document find_coro in Coro::Debug.
488 - incompatible change: rename has_stack to has_cctx. 724 - incompatible change: rename has_stack to has_cctx.
489 - Coro::AIO and Coro::BDB no longer force event model detection, 725 - Coro::AIO and Coro::BDB no longer force event model detection,
490 use AnyEvent::AIO and AnyEvent::BDB. 726 use AnyEvent::AIO and AnyEvent::BDB.
… …
614 850
6154.11 Thu Oct 11 02:40:24 CEST 2007 8514.11 Thu Oct 11 02:40:24 CEST 2007
616 - port to threaded perls. 852 - port to threaded perls.
617 853
6184.1 Thu Oct 11 02:38:16 CEST 2007 8544.1 Thu Oct 11 02:38:16 CEST 2007
619 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 855 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
620 be local to each coro (see Coro::State). 856 be local to each coro (see Coro::State).
621 - incompatible change: for very deep reasons, cede and cede_notself 857 - incompatible change: for very deep reasons, cede and cede_notself
622 cannot return anything, so nothing will be returned. 858 cannot return anything, so nothing will be returned.
623 - possibly bring back 5.10 compatibility (untested). 859 - possibly bring back 5.10 compatibility (untested).
624 - work around stupid (and wrong) warning on 5.10 :(. 860 - work around stupid (and wrong) warning on 5.10 :(.
… …
696 but stateful semantics. 932 but stateful semantics.
697 - fixed a lot of typos in Coro.pm (patch submitted by David 933 - fixed a lot of typos in Coro.pm (patch submitted by David
698 Steinbrunner, which applied flawlessly). 934 Steinbrunner, which applied flawlessly).
699 935
7003.6 Sat Apr 14 17:13:31 CEST 2007 9363.6 Sat Apr 14 17:13:31 CEST 2007
701 - added some bugfixes to get eg/myhttpd working again. 937 - added some bugfixes to get eg/myhttpd working again.
702 - added Coro::Storable for often-cede'ing freeze/thaw. 938 - added Coro::Storable for often-cede'ing freeze/thaw.
703 - try to do a clean exit when a coroutine calls exit 939 - try to do a clean exit when a coroutine calls exit
704 (EXPERIMENTAL). 940 (EXPERIMENTAL).
705 - got rid of indirect call through _coro_init. 941 - got rid of indirect call through _coro_init.
706 - updated the partly antique examples in eg/ to 942 - updated the partly antique examples in eg/ to

Diff Legend

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