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

Comparing Coro/Changes (file contents):
Revision 1.567 by root, Sun May 22 20:13:33 2011 UTC vs.
Revision 1.645 by root, Tue Jun 30 12:42:44 2015 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, andallow to unswap by repeat?
7TODO: swap_sv not undone in pool
8TODO: croak when async_pool tries to run canceled thread?
6 9
7TODO: tutorial 106.46 Tue Jun 30 14:36:30 CEST 2015
8TODO: in_eval, PL_tained 11 - restore portability to perls with windows fork emulation
12 (reported by Petr Písař).
13
146.45 Tue Jun 30 01:40:08 CEST 2015
15 - restore portability to perl 5.16 and below.
16
176.44 Tue Jun 30 00:41:54 CEST 2015
18 - avoid segfaulting (or worse) when tracing a canceled thread.
19 - the Event module silently broke it's public hook API, causing
20 failures on perls compiled with -Duselongdouble. Requiring
21 the latest version of Event because I don't know which release
22 changed this (it's not mentioned in the ChangeLog).
23 - allow xs level enterleave hooks via CoroAPI.
24
256.43 Thu Jun 4 15:38:14 CEST 2015
26 - use stability canary.
27 - port to stableperl-5.22.0-1.001.
28 - update libecb.
29
306.42 Wed Feb 11 20:29:52 CET 2015
31 - Coro::SemaphoreSet->try did not actually work (analyzed by
32 SATO Kentaro).
33 - upgrade libecb to be C11 compliant.
34
356.41 Sat Sep 6 22:08:46 CEST 2014
36 - restore portability to perl 5.8.x.
37 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
38 up some memory, but fixes "use feature" and similar modules
39 when used inside a Coro.
40 - allow zero as argument to Coro::Channel to mean the same thing
41 as no argument. this works with older versions as well,
42 but wasn't legal until now.
43 - slightly better c header file detection.
44
456.39 Mon Jun 2 00:00:08 CEST 2014
46 - work around more incompatible changes in 5.20.
47
486.38 Sun Jun 1 21:54:23 CEST 2014
49 - check that perl slots actually have enough space to hold
50 interpreter variables.
51 - untested port to perl 5.19 (64 bit tmps indices) (reported
52 by Andreas König).
53 - croak when cancel is called without a thread context.
54
556.37 Tue Mar 4 13:27:33 CET 2014
56 - *sigh*, removed leftover debugging code from debugging a
57 perl bug, of all things.
58
596.36 Tue Mar 4 07:11:59 CET 2014
60 - semaphores would not clear the destroy hook when interrupted
61 by ->throw, causing segfaults or worse.
62 - ->throw on a thread waiting for a semaphore did not acquire
63 the semaphore, but also didn't wake up other waiters,
64 possibly causing a deadlock.
65 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
66 to report a full stacktrace, which should help find out
67 where the actual call is.
68 - "atomically" destroy data in slf_destroy, because it is
69 the right thing to do, just in case.
70 - disable ecb.h libm dependency, to compile on ancient systems
71 or under adverse conditions.
72
736.33 Mon Nov 18 11:26:27 CET 2013
74 - do not crash when freeing padlists with holes (in 5.18).
75 - tentative SVt_BIND 5.19 port/fix.
76
776.32 Tue Nov 5 15:35:35 CET 2013
78 - use a new algorithm to derive padlists for perl 5.18. The old
79 one could lead to 0-pointer accesses inside perl (reported
80 by Darin McBride).
81
826.31 Thu May 9 07:39:48 CEST 2013
83 - Coro::AIO requests would crash if the thread was ready'd
84 while the request was ongoing.
85
866.29 Wed May 8 02:55:18 CEST 2013
87 - when an on_destroy handler destructs the coro currently being
88 destructed a perl scalar could be accessed after being freed,
89 likely causing a crash.
90
916.28 Wed Mar 6 06:58:02 CET 2013
92 - clean remnants of existing __DIE__ and __WARN__ handlers so
93 they lose their magic and will not cause segfaults later
94 (testcase by Andrey Sagulin).
95 - improved Coro::State documentation a bit.
96 - Coro::Debug::command now flushes the output.
97 - add hack detection code for x32 abi, because the braindead slugs
98 who designed that made it look exactly like x86_64 without
99 providing proper compile time symbols to test for it. as a result,
100 this detection cannot work reliably.
101 - valgrind stack registering was broken.
102 - do not rely on Time::HiRes anymore in Coro::Debug.
103
1046.23 Fri Dec 7 23:36:37 CET 2012
105 - use experimental fiber implementation on native windows
106 perls.
107 - use sizeof (void *) as multiplication factor for stack sizes,
108 to accomodate the totally braindamaged microsoft 64 bit "os".
109 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
110 activeperl 5.16 crashes when PerlIO_define_layer is called due
111 to some bug in the perl dll, strawberry perl at least passes
112 the testsuite.
113 - implement Coro::Handle->peeraddr/host/port, for slightly
114 improved compatibility with LWP.
115 - implement 5.17 compatibility by almost blindly applying a
116 good-looking patch by Father Chrysostomos.
117 - move stack management functions into libcoro 3.
118 - libcoro version 3 "released".
119 - support magic values as timed_io_once args.
120 - recommend AnyEvent 7+ or EV 4+, also require EV
121 version 4 or newer for Coro::EV.
122
1236.10 Tue Oct 9 01:14:27 CEST 2012
124 - updated ecb.h, it had a typo that caused it to not compile on many
125 big endian systems (reported by many people).
126 - disable memory fences in ecb.h to improve portability.
127
1286.09 Sat Oct 6 23:25:02 CEST 2012
129 - Coro::EV I/O watchers were not interruptible by exceptions
130 (Coro::State::throw) (testcase by sten).
131 - ->throw now puts threads into the ready queue, as this seems to
132 be expected by existing code, and code that doesn't cope with spurious
133 wakeups needs fixing anyway.
134 - use fd -1 in mmap.
135 - cast I32 to int in error message printf.
136 - warn about broken so-called "hardened" kernels.
137
1386.08 Fri Apr 13 12:05:47 CEST 2012
139 - be more aggressive about exiting like perl does - formerly,
140 exiting from the non-main thread would not execute END blocks.
141
1426.07 Fri Nov 11 21:21:48 CET 2011
143 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
144 - update ecb.h.
145
1466.06 Mon Aug 8 23:59:48 CEST 2011
147 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
148 backend, which is an order of magnitude slower. unfortunately, cygwins
149 pthread implementation isn't very complete either, so allocate the stack
150 twice just to be sure.
151 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
152 in its ithreaded perl - best recompile cygwin and use the 'w'indows
153 backend for much better performance. also disable ithreads for
154 even better performance...).
155
1566.05 Thu Aug 4 21:36:36 CEST 2011
157 - blush, condvar values would not be propagated from send to recv anymore
158 (reported by Chip Salzenberg).
159 - use exponential increase for the readline buffer length in
160 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
161
1626.04 Wed Aug 3 17:15:45 CEST 2011
163 - use even more efficient and more compatible condvars for
164 compatibility to AnyEvent 6.x :)
165 - more inconsequential ecb.h updates.
166
1676.03 Wed Aug 3 11:41:30 CEST 2011
168 - change how Coro patches AnyEvent condvars for compatibility to
169 AnyEvent 6.x.
170 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
171 hasn't been included.
172
1736.02 Wed Jul 13 04:35:19 CEST 2011
174 - "improve portability to Gentoo" - gentoo manages to put perl variables
175 in memory areas that are farther than 2gb apart, which the jit couldn't
176 handle and barfed. now it's just a bit slower on gentoo and similar
177 systems.
178
1796.01 Sun Jul 3 12:31:14 CEST 2011
180 - workarounds are good, but the test for whether pthreads are used
181 was not good. this one should be better.
182 - check differently whether gcc generates cfi instructions itself.
183
1846.0 Wed Jun 29 19:43:35 CEST 2011
9 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be 185 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
10 destroyed and cleaned up automatically (e.g. async { schedule }). 186 destroyed and cleaned up automatically (e.g. async { schedule }).
187 - implement a JIT compiler for part of the thread switch code,
188 which gives a 50% speed improvement on threaded perls, and
189 about 4% on non-threaded perls (so threaded perls now finally
190 reach about half the speed of non-threaded perls).
191 - slightly modernise Coro::Intro, add section about rouse functions.
192 - avoid DEFSV and ERRSV, giving another 10% improvement
193 in thread switching.
11 - Coro::State->is_destroyed is now called is_zombie. 194 - Coro::State->is_destroyed is now called is_zombie.
12 - implement a Coro->safe_cancel method that might fail, but 195 - implement a Coro->safe_cancel method that might fail, but
13 cancels in a "safer" way if it succeeds. 196 cancels in a "safer" way if it succeeds.
197 - add preliminary support for DEBUGGING perls.
14 - get rid of two hash-accesses when initialising a new Coro - this 198 - get rid of two hash-accesses when initialising a new Coro - this
15 speeds up coro creation by almost a factor of two. 199 speeds up coro creation by almost a factor of two.
16 - croak when a coro that is being cancelled tries to block 200 - croak when a coro that is being cancelled tries to block
17 (e.g. while executing a guard block), instead of crashing or 201 (e.g. while executing a guard block), instead of crashing or
18 deadlocking. 202 deadlocking.
… …
32 functions to have destructors. 216 functions to have destructors.
33 - get rid if coro refcounting - simply crash in other interpreter 217 - get rid if coro refcounting - simply crash in other interpreter
34 threads by nulling the pointers on clone. 218 threads by nulling the pointers on clone.
35 - simplify warn/die hook handling when loading Coro - the convoluted 219 - simplify warn/die hook handling when loading Coro - the convoluted
36 logic seems to be no longer neccessary. 220 logic seems to be no longer neccessary.
221 - use libecb instead of our own home-grown gcc hacks.
222 - document alternatives to Coro::LWP. Please use them :)
223 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
224 sigaltstack. Really. wine suffers from it, erlang suffers from it,
225 and it's known since at least 2006.
37 226
385.372 Wed Feb 23 06:14:30 CET 2011 2275.372 Wed Feb 23 06:14:30 CET 2011
39 - apparently mingw doesn't provide a working gettimeofday, try to 228 - apparently mingw doesn't provide a working gettimeofday, try to
40 work around that by relying on Time::HiRes (indirectly brought to 229 work around that by relying on Time::HiRes (indirectly brought to
41 my attention by Max Maischein). 230 my attention by Max Maischein).
… …
359 - ->throw is now supported on Coro::State objects. 548 - ->throw is now supported on Coro::State objects.
360 - clean up cctx creation code a bit. 549 - clean up cctx creation code a bit.
361 - entersub is actually a UNOP, not a LOGOP (not a bugfix). 550 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
362 551
3634.9 Sat Nov 8 17:45:27 CET 2008 5524.9 Sat Nov 8 17:45:27 CET 2008
364 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 553 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
365 - (libcoro) no longer leak threads in the experimental pthread backend, 554 - (libcoro) no longer leak threads in the experimental pthread backend,
366 also speed it up considerably. 555 also speed it up considerably.
367 - (libcoro) do not rely on makecontext passing void *'s unscathed. 556 - (libcoro) do not rely on makecontext passing void *'s unscathed.
368 - fix compiletime dependencies on libcoro in the Makefile. 557 - fix compiletime dependencies on libcoro in the Makefile.
369 - cctx_count wasn't always updated properly. 558 - cctx_count wasn't always updated properly.
… …
431 - remove debugging code related to MgPV_nolen_const, also try to 620 - remove debugging code related to MgPV_nolen_const, also try to
432 make it compile with perl 5.8.6 (yes, apple apparently loves 621 make it compile with perl 5.8.6 (yes, apple apparently loves
433 outdated software). Reported by John S. 622 outdated software). Reported by John S.
434 623
4354.744 Tue Jul 8 22:06:35 CEST 2008 6244.744 Tue Jul 8 22:06:35 CEST 2008
436 - correctly provide default DIE/WARN handlers as documented. 625 - correctly provide default DIE/WARN handlers as documented.
437 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 626 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
438 implementations inside perl work fine for us. 627 implementations inside perl work fine for us.
439 628
4404.743 Mon Jun 16 00:21:57 CEST 2008 6294.743 Mon Jun 16 00:21:57 CEST 2008
441 - when using Coro::EV without running EV::loop it could 630 - when using Coro::EV without running EV::loop it could
… …
463 - sprinkle "no warnings" freely over everything, also suppress 652 - sprinkle "no warnings" freely over everything, also suppress
464 warnings for some other modules. 653 warnings for some other modules.
465 - fix typo in WSAEWOULDBLOCK. 654 - fix typo in WSAEWOULDBLOCK.
466 655
4674.72 Sun May 25 05:14:36 CEST 2008 6564.72 Sun May 25 05:14:36 CEST 2008
468 - tweak META.yaml a bit, unfortunately, there is no documented way 657 - tweak META.yaml a bit, unfortunately, there is no documented way
469 to have optional dependencies with CPAN. doh :( 658 to have optional dependencies with CPAN. doh :(
470 - avoid running some tests on windows because they would fail due to 659 - avoid running some tests on windows because they would fail due to
471 perl bug (broken fork, broken pipes...). 660 perl bug (broken fork, broken pipes...).
472 - work around perl on windows bugs where perl returns undocumented 661 - work around perl on windows bugs where perl returns undocumented
473 error codes for sysread, syswrite etc. by taking advantage 662 error codes for sysread, syswrite etc. by taking advantage
… …
477 - use unix domain sockets in testsuite to work around 666 - use unix domain sockets in testsuite to work around
478 common perl implementation bugs on widows (they are emulated by 667 common perl implementation bugs on widows (they are emulated by
479 tcp sockets on windows. ugh.) 668 tcp sockets on windows. ugh.)
480 669
4814.71 Sat May 24 20:01:27 CEST 2008 6704.71 Sat May 24 20:01:27 CEST 2008
482 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 671 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
483 - take advantage of async name resolution of AnyEvent::Util. 672 - take advantage of async name resolution of AnyEvent::Util.
484 - work around brutal inet_aton override in Coro::LWP. 673 - work around brutal inet_aton override in Coro::LWP.
485 - take advantage of the readyhook in Coro::EV, for smoother 674 - take advantage of the readyhook in Coro::EV, for smoother
486 scheduling. 675 scheduling.
487 676
4884.7 Sun May 11 00:32:19 CEST 2008 6774.7 Sun May 11 00:32:19 CEST 2008
489 - completely reworked the Coro manpage. 678 - completely reworked the Coro manpage.
490 - added Coro::AnyEvent, generic event loop integration. 679 - added Coro::AnyEvent, generic event loop integration.
491 - implement cancel, ready and kill commands in Coro::Debug. 680 - implement cancel, ready and kill commands in Coro::Debug.
492 - document find_coro in Coro::Debug. 681 - document find_coro in Coro::Debug.
493 - incompatible change: rename has_stack to has_cctx. 682 - incompatible change: rename has_stack to has_cctx.
494 - Coro::AIO and Coro::BDB no longer force event model detection, 683 - Coro::AIO and Coro::BDB no longer force event model detection,
495 use AnyEvent::AIO and AnyEvent::BDB. 684 use AnyEvent::AIO and AnyEvent::BDB.
… …
619 808
6204.11 Thu Oct 11 02:40:24 CEST 2007 8094.11 Thu Oct 11 02:40:24 CEST 2007
621 - port to threaded perls. 810 - port to threaded perls.
622 811
6234.1 Thu Oct 11 02:38:16 CEST 2007 8124.1 Thu Oct 11 02:38:16 CEST 2007
624 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 813 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
625 be local to each coro (see Coro::State). 814 be local to each coro (see Coro::State).
626 - incompatible change: for very deep reasons, cede and cede_notself 815 - incompatible change: for very deep reasons, cede and cede_notself
627 cannot return anything, so nothing will be returned. 816 cannot return anything, so nothing will be returned.
628 - possibly bring back 5.10 compatibility (untested). 817 - possibly bring back 5.10 compatibility (untested).
629 - work around stupid (and wrong) warning on 5.10 :(. 818 - work around stupid (and wrong) warning on 5.10 :(.
… …
701 but stateful semantics. 890 but stateful semantics.
702 - fixed a lot of typos in Coro.pm (patch submitted by David 891 - fixed a lot of typos in Coro.pm (patch submitted by David
703 Steinbrunner, which applied flawlessly). 892 Steinbrunner, which applied flawlessly).
704 893
7053.6 Sat Apr 14 17:13:31 CEST 2007 8943.6 Sat Apr 14 17:13:31 CEST 2007
706 - added some bugfixes to get eg/myhttpd working again. 895 - added some bugfixes to get eg/myhttpd working again.
707 - added Coro::Storable for often-cede'ing freeze/thaw. 896 - added Coro::Storable for often-cede'ing freeze/thaw.
708 - try to do a clean exit when a coroutine calls exit 897 - try to do a clean exit when a coroutine calls exit
709 (EXPERIMENTAL). 898 (EXPERIMENTAL).
710 - got rid of indirect call through _coro_init. 899 - got rid of indirect call through _coro_init.
711 - updated the partly antique examples in eg/ to 900 - updated the partly antique examples in eg/ to

Diff Legend

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