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

Comparing Coro/Changes (file contents):
Revision 1.573 by root, Wed Jun 29 16:08:31 2011 UTC vs.
Revision 1.671 by root, Thu Aug 31 16:24:19 2017 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 9 - safe_cancel no longer croaks when called on already-destroyed
10 thread, but instead "just works".
11
126.513 Sat Jul 15 01:18:51 CEST 2017
13 - experimental 5.22 comaptibility changes (PadlistNAMES).
14 - removed attributes example and made sure the other examples
15 still work.
16 - fix some statevar types for newer perls, should be of no consequence.
17
186.512 Fri Jul 14 05:03:17 CEST 2017
19 - seems to pass test suite down till 5.8.9, so make this the
20 minimum requirement in Makefile.PL.
21 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
22 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
23 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
24 fortified longjmp variants in some libcs.
25 - re-try JIT allocation without PROT_EXEC, to hopefully improve
26 portability (still doesn't seem to work around broken selinux
27 mmap everywhere).
28 - speling fixes (Xavier Guimard).
29
306.511 Sun Jun 26 23:44:50 CEST 2016
31 - make it compile with threaded perls again.
32 - simplify and speed up __DIE__ and __WARN__ handling.
33
346.51 Sat Jun 25 20:57:20 CEST 2016
35 - 6.5 release didn't compile on most older perls due to
36 a logic error enabling experimental 5.24 code for all versions.
37 - since various broken or deficient patches float around
38 to work around the vtbl API breakage in perl, let's provide
39 our own workaround, which seems to be much less invasive
40 and more compatible than the approaches seen so far. YMMV.
41 - work around assertion failure in perl_destruct on 5.24
42 (analyzed by Dave Mitchell).
43
446.5 Wed Jun 22 22:23:50 CEST 2016
45 - swap_sv swapping order was not symmetrical, causing
46 wrong swaps when swap_sv was used multiple times on the
47 same sv in the same thread.
48 - swap_sv calls can now be undone by calling it again
49 with the same variables.
50 - swap_sv calls will now be undone in async_pool threads.
51 - split Coro::Semaphore::up/adjust into separate xs functions
52 for better error reporting, at slight codesize increase.
53 - (libcoro) arm assembly support, please test and report.
54 - adjust to PL_savestack changes in perl 5.24 (adapted from
55 the debian patch, which unfortunately gets it wrong).
56
576.49 Sat Oct 17 01:40:12 CEST 2015
58 - throwing an exception to a thread waiting in
59 Coro::Handle using Coro::EV did not stop the watchers,
60 causing the next call to fail (testcase by Martin Pritchard).
61 - bump minimum perl version to 5.10.
62
636.48 Sun Oct 4 19:03:51 CEST 2015
64 - fix memory leak when Coro::Handle uses Coro::EV internally
65 (testcase by Sten Sten).
66 - update code to libev 4 API (internally, Coro still used the
67 version 3 API calls).
68 - remove unused hv_sig.
69
706.47 Sat Jul 11 03:58:20 CEST 2015
71 - swap_svs - NVs can be in the SV head beginning with 5.20.
72
736.46 Tue Jun 30 14:36:30 CEST 2015
74 - restore portability to perls with windows fork emulation
75 (patch by Petr Písař).
76
776.45 Tue Jun 30 01:40:08 CEST 2015
78 - restore portability to perl 5.16 and below.
79
806.44 Tue Jun 30 00:41:54 CEST 2015
81 - avoid segfaulting (or worse) when tracing a canceled thread.
82 - the Event module silently broke it's public hook API, causing
83 failures on perls compiled with -Duselongdouble. Requiring
84 the latest version of Event because I don't know which release
85 changed this (it's not mentioned in the ChangeLog).
86 - allow xs level enterleave hooks via CoroAPI.
87
886.43 Thu Jun 4 15:38:14 CEST 2015
89 - use stability canary.
90 - port to stableperl-5.22.0-1.001.
91 - update libecb.
92
936.42 Wed Feb 11 20:29:52 CET 2015
94 - Coro::SemaphoreSet->try did not actually work (analyzed by
95 SATO Kentaro).
96 - upgrade libecb to be C11 compliant.
97
986.41 Sat Sep 6 22:08:46 CEST 2014
99 - restore portability to perl 5.8.x.
100 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
101 up some memory, but fixes "use feature" and similar modules
102 when used inside a Coro.
103 - allow zero as argument to Coro::Channel to mean the same thing
104 as no argument. this works with older versions as well,
105 but wasn't legal until now.
106 - slightly better c header file detection.
107
1086.39 Mon Jun 2 00:00:08 CEST 2014
109 - work around more incompatible changes in 5.20.
110
1116.38 Sun Jun 1 21:54:23 CEST 2014
112 - check that perl slots actually have enough space to hold
113 interpreter variables.
114 - untested port to perl 5.19 (64 bit tmps indices) (reported
115 by Andreas König).
116 - croak when cancel is called without a thread context.
117
1186.37 Tue Mar 4 13:27:33 CET 2014
119 - *sigh*, removed leftover debugging code from debugging a
120 perl bug, of all things.
121
1226.36 Tue Mar 4 07:11:59 CET 2014
123 - semaphores would not clear the destroy hook when interrupted
124 by ->throw, causing segfaults or worse.
125 - ->throw on a thread waiting for a semaphore did not acquire
126 the semaphore, but also didn't wake up other waiters,
127 possibly causing a deadlock.
128 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
129 to report a full stacktrace, which should help find out
130 where the actual call is.
131 - "atomically" destroy data in slf_destroy, because it is
132 the right thing to do, just in case.
133 - disable ecb.h libm dependency, to compile on ancient systems
134 or under adverse conditions.
135
1366.33 Mon Nov 18 11:26:27 CET 2013
137 - do not crash when freeing padlists with holes (in 5.18).
138 - tentative SVt_BIND 5.19 port/fix.
139
1406.32 Tue Nov 5 15:35:35 CET 2013
141 - use a new algorithm to derive padlists for perl 5.18. The old
142 one could lead to 0-pointer accesses inside perl (reported
143 by Darin McBride).
144
1456.31 Thu May 9 07:39:48 CEST 2013
146 - Coro::AIO requests would crash if the thread was ready'd
147 while the request was ongoing.
148
1496.29 Wed May 8 02:55:18 CEST 2013
150 - when an on_destroy handler destructs the coro currently being
151 destructed a perl scalar could be accessed after being freed,
152 likely causing a crash.
153
1546.28 Wed Mar 6 06:58:02 CET 2013
155 - clean remnants of existing __DIE__ and __WARN__ handlers so
156 they lose their magic and will not cause segfaults later
157 (testcase by Andrey Sagulin).
158 - improved Coro::State documentation a bit.
159 - Coro::Debug::command now flushes the output.
160 - add hack detection code for x32 abi, because the braindead slugs
161 who designed that made it look exactly like x86_64 without
162 providing proper compile time symbols to test for it. as a result,
163 this detection cannot work reliably.
164 - valgrind stack registering was broken.
165 - do not rely on Time::HiRes anymore in Coro::Debug.
166
1676.23 Fri Dec 7 23:36:37 CET 2012
168 - use experimental fiber implementation on native windows
169 perls.
170 - use sizeof (void *) as multiplication factor for stack sizes,
171 to accomodate the totally braindamaged microsoft 64 bit "os".
172 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
173 activeperl 5.16 crashes when PerlIO_define_layer is called due
174 to some bug in the perl dll, strawberry perl at least passes
175 the testsuite.
176 - implement Coro::Handle->peeraddr/host/port, for slightly
177 improved compatibility with LWP.
178 - implement 5.17 compatibility by almost blindly applying a
179 good-looking patch by Father Chrysostomos.
180 - move stack management functions into libcoro 3.
181 - libcoro version 3 "released".
182 - support magic values as timed_io_once args.
183 - recommend AnyEvent 7+ or EV 4+, also require EV
184 version 4 or newer for Coro::EV.
185
1866.10 Tue Oct 9 01:14:27 CEST 2012
187 - updated ecb.h, it had a typo that caused it to not compile on many
188 big endian systems (reported by many people).
189 - disable memory fences in ecb.h to improve portability.
190
1916.09 Sat Oct 6 23:25:02 CEST 2012
192 - Coro::EV I/O watchers were not interruptible by exceptions
193 (Coro::State::throw) (testcase by sten).
194 - ->throw now puts threads into the ready queue, as this seems to
195 be expected by existing code, and code that doesn't cope with spurious
196 wakeups needs fixing anyway.
197 - use fd -1 in mmap.
198 - cast I32 to int in error message printf.
199 - warn about broken so-called "hardened" kernels.
200
2016.08 Fri Apr 13 12:05:47 CEST 2012
202 - be more aggressive about exiting like perl does - formerly,
203 exiting from the non-main thread would not execute END blocks.
204
2056.07 Fri Nov 11 21:21:48 CET 2011
206 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
207 - update ecb.h.
208
2096.06 Mon Aug 8 23:59:48 CEST 2011
210 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
211 backend, which is an order of magnitude slower. unfortunately, cygwins
212 pthread implementation isn't very complete either, so allocate the stack
213 twice just to be sure.
214 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
215 in its ithreaded perl - best recompile cygwin and use the 'w'indows
216 backend for much better performance. also disable ithreads for
217 even better performance...).
218
2196.05 Thu Aug 4 21:36:36 CEST 2011
220 - blush, condvar values would not be propagated from send to recv anymore
221 (reported by Chip Salzenberg).
222 - use exponential increase for the readline buffer length in
223 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
224
2256.04 Wed Aug 3 17:15:45 CEST 2011
226 - use even more efficient and more compatible condvars for
227 compatibility to AnyEvent 6.x :)
228 - more inconsequential ecb.h updates.
229
2306.03 Wed Aug 3 11:41:30 CEST 2011
231 - change how Coro patches AnyEvent condvars for compatibility to
232 AnyEvent 6.x.
233 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
234 hasn't been included.
235
2366.02 Wed Jul 13 04:35:19 CEST 2011
237 - "improve portability to Gentoo" - gentoo manages to put perl variables
238 in memory areas that are farther than 2gb apart, which the jit couldn't
239 handle and barfed. now it's just a bit slower on gentoo and similar
240 systems.
241
2426.01 Sun Jul 3 12:31:14 CEST 2011
243 - workarounds are good, but the test for whether pthreads are used
244 was not good. this one should be better.
245 - check differently whether gcc generates cfi instructions itself.
246
2476.0 Wed Jun 29 19:43:35 CEST 2011
8 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be 248 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
9 destroyed and cleaned up automatically (e.g. async { schedule }). 249 destroyed and cleaned up automatically (e.g. async { schedule }).
10 - implement a JIT compiler for part of the thread switch code, 250 - implement a JIT compiler for part of the thread switch code,
11 which gives a 50% speed improvement on threaded perls, and 251 which gives a 50% speed improvement on threaded perls, and
12 about 4% on non-thraeded perls (so threaded perls now finally 252 about 4% on non-threaded perls (so threaded perls now finally
13 reach about half the speed of non-threaded perls). 253 reach about half the speed of non-threaded perls).
254 - slightly modernise Coro::Intro, add section about rouse functions.
14 - avoid DEFSV and ERRSV, giving another 10% improvement 255 - avoid DEFSV and ERRSV, giving another 10% improvement
15 in thread switching. 256 in thread switching.
16 - Coro::State->is_destroyed is now called is_zombie. 257 - Coro::State->is_destroyed is now called is_zombie.
17 - implement a Coro->safe_cancel method that might fail, but 258 - implement a Coro->safe_cancel method that might fail, but
18 cancels in a "safer" way if it succeeds. 259 cancels in a "safer" way if it succeeds.
… …
370 - ->throw is now supported on Coro::State objects. 611 - ->throw is now supported on Coro::State objects.
371 - clean up cctx creation code a bit. 612 - clean up cctx creation code a bit.
372 - entersub is actually a UNOP, not a LOGOP (not a bugfix). 613 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
373 614
3744.9 Sat Nov 8 17:45:27 CET 2008 6154.9 Sat Nov 8 17:45:27 CET 2008
375 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 616 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
376 - (libcoro) no longer leak threads in the experimental pthread backend, 617 - (libcoro) no longer leak threads in the experimental pthread backend,
377 also speed it up considerably. 618 also speed it up considerably.
378 - (libcoro) do not rely on makecontext passing void *'s unscathed. 619 - (libcoro) do not rely on makecontext passing void *'s unscathed.
379 - fix compiletime dependencies on libcoro in the Makefile. 620 - fix compiletime dependencies on libcoro in the Makefile.
380 - cctx_count wasn't always updated properly. 621 - cctx_count wasn't always updated properly.
… …
442 - remove debugging code related to MgPV_nolen_const, also try to 683 - remove debugging code related to MgPV_nolen_const, also try to
443 make it compile with perl 5.8.6 (yes, apple apparently loves 684 make it compile with perl 5.8.6 (yes, apple apparently loves
444 outdated software). Reported by John S. 685 outdated software). Reported by John S.
445 686
4464.744 Tue Jul 8 22:06:35 CEST 2008 6874.744 Tue Jul 8 22:06:35 CEST 2008
447 - correctly provide default DIE/WARN handlers as documented. 688 - correctly provide default DIE/WARN handlers as documented.
448 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 689 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
449 implementations inside perl work fine for us. 690 implementations inside perl work fine for us.
450 691
4514.743 Mon Jun 16 00:21:57 CEST 2008 6924.743 Mon Jun 16 00:21:57 CEST 2008
452 - when using Coro::EV without running EV::loop it could 693 - when using Coro::EV without running EV::loop it could
… …
474 - sprinkle "no warnings" freely over everything, also suppress 715 - sprinkle "no warnings" freely over everything, also suppress
475 warnings for some other modules. 716 warnings for some other modules.
476 - fix typo in WSAEWOULDBLOCK. 717 - fix typo in WSAEWOULDBLOCK.
477 718
4784.72 Sun May 25 05:14:36 CEST 2008 7194.72 Sun May 25 05:14:36 CEST 2008
479 - tweak META.yaml a bit, unfortunately, there is no documented way 720 - tweak META.yaml a bit, unfortunately, there is no documented way
480 to have optional dependencies with CPAN. doh :( 721 to have optional dependencies with CPAN. doh :(
481 - avoid running some tests on windows because they would fail due to 722 - avoid running some tests on windows because they would fail due to
482 perl bug (broken fork, broken pipes...). 723 perl bug (broken fork, broken pipes...).
483 - work around perl on windows bugs where perl returns undocumented 724 - work around perl on windows bugs where perl returns undocumented
484 error codes for sysread, syswrite etc. by taking advantage 725 error codes for sysread, syswrite etc. by taking advantage
… …
488 - use unix domain sockets in testsuite to work around 729 - use unix domain sockets in testsuite to work around
489 common perl implementation bugs on widows (they are emulated by 730 common perl implementation bugs on widows (they are emulated by
490 tcp sockets on windows. ugh.) 731 tcp sockets on windows. ugh.)
491 732
4924.71 Sat May 24 20:01:27 CEST 2008 7334.71 Sat May 24 20:01:27 CEST 2008
493 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 734 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
494 - take advantage of async name resolution of AnyEvent::Util. 735 - take advantage of async name resolution of AnyEvent::Util.
495 - work around brutal inet_aton override in Coro::LWP. 736 - work around brutal inet_aton override in Coro::LWP.
496 - take advantage of the readyhook in Coro::EV, for smoother 737 - take advantage of the readyhook in Coro::EV, for smoother
497 scheduling. 738 scheduling.
498 739
4994.7 Sun May 11 00:32:19 CEST 2008 7404.7 Sun May 11 00:32:19 CEST 2008
500 - completely reworked the Coro manpage. 741 - completely reworked the Coro manpage.
501 - added Coro::AnyEvent, generic event loop integration. 742 - added Coro::AnyEvent, generic event loop integration.
502 - implement cancel, ready and kill commands in Coro::Debug. 743 - implement cancel, ready and kill commands in Coro::Debug.
503 - document find_coro in Coro::Debug. 744 - document find_coro in Coro::Debug.
504 - incompatible change: rename has_stack to has_cctx. 745 - incompatible change: rename has_stack to has_cctx.
505 - Coro::AIO and Coro::BDB no longer force event model detection, 746 - Coro::AIO and Coro::BDB no longer force event model detection,
506 use AnyEvent::AIO and AnyEvent::BDB. 747 use AnyEvent::AIO and AnyEvent::BDB.
… …
630 871
6314.11 Thu Oct 11 02:40:24 CEST 2007 8724.11 Thu Oct 11 02:40:24 CEST 2007
632 - port to threaded perls. 873 - port to threaded perls.
633 874
6344.1 Thu Oct 11 02:38:16 CEST 2007 8754.1 Thu Oct 11 02:38:16 CEST 2007
635 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 876 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
636 be local to each coro (see Coro::State). 877 be local to each coro (see Coro::State).
637 - incompatible change: for very deep reasons, cede and cede_notself 878 - incompatible change: for very deep reasons, cede and cede_notself
638 cannot return anything, so nothing will be returned. 879 cannot return anything, so nothing will be returned.
639 - possibly bring back 5.10 compatibility (untested). 880 - possibly bring back 5.10 compatibility (untested).
640 - work around stupid (and wrong) warning on 5.10 :(. 881 - work around stupid (and wrong) warning on 5.10 :(.
… …
712 but stateful semantics. 953 but stateful semantics.
713 - fixed a lot of typos in Coro.pm (patch submitted by David 954 - fixed a lot of typos in Coro.pm (patch submitted by David
714 Steinbrunner, which applied flawlessly). 955 Steinbrunner, which applied flawlessly).
715 956
7163.6 Sat Apr 14 17:13:31 CEST 2007 9573.6 Sat Apr 14 17:13:31 CEST 2007
717 - added some bugfixes to get eg/myhttpd working again. 958 - added some bugfixes to get eg/myhttpd working again.
718 - added Coro::Storable for often-cede'ing freeze/thaw. 959 - added Coro::Storable for often-cede'ing freeze/thaw.
719 - try to do a clean exit when a coroutine calls exit 960 - try to do a clean exit when a coroutine calls exit
720 (EXPERIMENTAL). 961 (EXPERIMENTAL).
721 - got rid of indirect call through _coro_init. 962 - got rid of indirect call through _coro_init.
722 - updated the partly antique examples in eg/ to 963 - updated the partly antique examples in eg/ to

Diff Legend

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