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

Comparing Coro/Changes (file contents):
Revision 1.559 by root, Fri May 6 21:15:17 2011 UTC vs.
Revision 1.643 by root, Mon Jun 29 22:42:33 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
7 - croak when a coro being cancelled tries to block (e.g. while 106.44 Tue Jun 30 00:41:54 CEST 2015
8 executing a guard block). 11 - avoid segfaulting (or worse) when tracing a canceled thread.
12 - the Event module silently broke it's public hook API, causing
13 failures on perls compiled with -Duselongdouble. Requiring
14 the latest version of Event because I don't know which release
15 changed this (it's not mentioned in the ChangeLog).
16 - allow xs level enterleave hooks via CoroAPI.
17
186.43 Thu Jun 4 15:38:14 CEST 2015
19 - use stability canary.
20 - port to stableperl-5.22.0-1.001.
21 - update libecb.
22
236.42 Wed Feb 11 20:29:52 CET 2015
24 - Coro::SemaphoreSet->try did not actually work (analyzed by
25 SATO Kentaro).
26 - upgrade libecb to be C11 compliant.
27
286.41 Sat Sep 6 22:08:46 CEST 2014
29 - restore portability to perl 5.8.x.
30 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
31 up some memory, but fixes "use feature" and similar modules
32 when used inside a Coro.
33 - allow zero as argument to Coro::Channel to mean the same thing
34 as no argument. this works with older versions as well,
35 but wasn't legal until now.
36 - slightly better c header file detection.
37
386.39 Mon Jun 2 00:00:08 CEST 2014
39 - work around more incompatible changes in 5.20.
40
416.38 Sun Jun 1 21:54:23 CEST 2014
42 - check that perl slots actually have enough space to hold
43 interpreter variables.
44 - untested port to perl 5.19 (64 bit tmps indices) (reported
45 by Andreas König).
46 - croak when cancel is called without a thread context.
47
486.37 Tue Mar 4 13:27:33 CET 2014
49 - *sigh*, removed leftover debugging code from debugging a
50 perl bug, of all things.
51
526.36 Tue Mar 4 07:11:59 CET 2014
53 - semaphores would not clear the destroy hook when interrupted
54 by ->throw, causing segfaults or worse.
55 - ->throw on a thread waiting for a semaphore did not acquire
56 the semaphore, but also didn't wake up other waiters,
57 possibly causing a deadlock.
58 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
59 to report a full stacktrace, which should help find out
60 where the actual call is.
61 - "atomically" destroy data in slf_destroy, because it is
62 the right thing to do, just in case.
63 - disable ecb.h libm dependency, to compile on ancient systems
64 or under adverse conditions.
65
666.33 Mon Nov 18 11:26:27 CET 2013
67 - do not crash when freeing padlists with holes (in 5.18).
68 - tentative SVt_BIND 5.19 port/fix.
69
706.32 Tue Nov 5 15:35:35 CET 2013
71 - use a new algorithm to derive padlists for perl 5.18. The old
72 one could lead to 0-pointer accesses inside perl (reported
73 by Darin McBride).
74
756.31 Thu May 9 07:39:48 CEST 2013
76 - Coro::AIO requests would crash if the thread was ready'd
77 while the request was ongoing.
78
796.29 Wed May 8 02:55:18 CEST 2013
80 - when an on_destroy handler destructs the coro currently being
81 destructed a perl scalar could be accessed after being freed,
82 likely causing a crash.
83
846.28 Wed Mar 6 06:58:02 CET 2013
85 - clean remnants of existing __DIE__ and __WARN__ handlers so
86 they lose their magic and will not cause segfaults later
87 (testcase by Andrey Sagulin).
88 - improved Coro::State documentation a bit.
89 - Coro::Debug::command now flushes the output.
90 - add hack detection code for x32 abi, because the braindead slugs
91 who designed that made it look exactly like x86_64 without
92 providing proper compile time symbols to test for it. as a result,
93 this detection cannot work reliably.
94 - valgrind stack registering was broken.
95 - do not rely on Time::HiRes anymore in Coro::Debug.
96
976.23 Fri Dec 7 23:36:37 CET 2012
98 - use experimental fiber implementation on native windows
99 perls.
100 - use sizeof (void *) as multiplication factor for stack sizes,
101 to accomodate the totally braindamaged microsoft 64 bit "os".
102 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
103 activeperl 5.16 crashes when PerlIO_define_layer is called due
104 to some bug in the perl dll, strawberry perl at least passes
105 the testsuite.
106 - implement Coro::Handle->peeraddr/host/port, for slightly
107 improved compatibility with LWP.
108 - implement 5.17 compatibility by almost blindly applying a
109 good-looking patch by Father Chrysostomos.
110 - move stack management functions into libcoro 3.
111 - libcoro version 3 "released".
112 - support magic values as timed_io_once args.
113 - recommend AnyEvent 7+ or EV 4+, also require EV
114 version 4 or newer for Coro::EV.
115
1166.10 Tue Oct 9 01:14:27 CEST 2012
117 - updated ecb.h, it had a typo that caused it to not compile on many
118 big endian systems (reported by many people).
119 - disable memory fences in ecb.h to improve portability.
120
1216.09 Sat Oct 6 23:25:02 CEST 2012
122 - Coro::EV I/O watchers were not interruptible by exceptions
123 (Coro::State::throw) (testcase by sten).
124 - ->throw now puts threads into the ready queue, as this seems to
125 be expected by existing code, and code that doesn't cope with spurious
126 wakeups needs fixing anyway.
127 - use fd -1 in mmap.
128 - cast I32 to int in error message printf.
129 - warn about broken so-called "hardened" kernels.
130
1316.08 Fri Apr 13 12:05:47 CEST 2012
132 - be more aggressive about exiting like perl does - formerly,
133 exiting from the non-main thread would not execute END blocks.
134
1356.07 Fri Nov 11 21:21:48 CET 2011
136 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
137 - update ecb.h.
138
1396.06 Mon Aug 8 23:59:48 CEST 2011
140 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
141 backend, which is an order of magnitude slower. unfortunately, cygwins
142 pthread implementation isn't very complete either, so allocate the stack
143 twice just to be sure.
144 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
145 in its ithreaded perl - best recompile cygwin and use the 'w'indows
146 backend for much better performance. also disable ithreads for
147 even better performance...).
148
1496.05 Thu Aug 4 21:36:36 CEST 2011
150 - blush, condvar values would not be propagated from send to recv anymore
151 (reported by Chip Salzenberg).
152 - use exponential increase for the readline buffer length in
153 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
154
1556.04 Wed Aug 3 17:15:45 CEST 2011
156 - use even more efficient and more compatible condvars for
157 compatibility to AnyEvent 6.x :)
158 - more inconsequential ecb.h updates.
159
1606.03 Wed Aug 3 11:41:30 CEST 2011
161 - change how Coro patches AnyEvent condvars for compatibility to
162 AnyEvent 6.x.
163 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
164 hasn't been included.
165
1666.02 Wed Jul 13 04:35:19 CEST 2011
167 - "improve portability to Gentoo" - gentoo manages to put perl variables
168 in memory areas that are farther than 2gb apart, which the jit couldn't
169 handle and barfed. now it's just a bit slower on gentoo and similar
170 systems.
171
1726.01 Sun Jul 3 12:31:14 CEST 2011
173 - workarounds are good, but the test for whether pthreads are used
174 was not good. this one should be better.
175 - check differently whether gcc generates cfi instructions itself.
176
1776.0 Wed Jun 29 19:43:35 CEST 2011
178 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
179 destroyed and cleaned up automatically (e.g. async { schedule }).
180 - implement a JIT compiler for part of the thread switch code,
181 which gives a 50% speed improvement on threaded perls, and
182 about 4% on non-threaded perls (so threaded perls now finally
183 reach about half the speed of non-threaded perls).
184 - slightly modernise Coro::Intro, add section about rouse functions.
185 - avoid DEFSV and ERRSV, giving another 10% improvement
186 in thread switching.
187 - Coro::State->is_destroyed is now called is_zombie.
9 - implement a Coro::State->safe_cancel method that might fail, but 188 - implement a Coro->safe_cancel method that might fail, but
10 cancels in a "safer" way. 189 cancels in a "safer" way if it succeeds.
190 - add preliminary support for DEBUGGING perls.
191 - get rid of two hash-accesses when initialising a new Coro - this
192 speeds up coro creation by almost a factor of two.
193 - croak when a coro that is being cancelled tries to block
194 (e.g. while executing a guard block), instead of crashing or
195 deadlocking.
196 - use a more robust and also faster method to identify Coro::State
197 objects - speeds up everything a bit.
11 - implement Coro->cancel in XS for a 20% speed improvement, and to 198 - implement Coro->cancel in XS for a 20% speed improvement, and to
12 be able to implement mutual cancellation. 199 be able to implement mutual cancellation.
200 - speed up context switches by a percent or two by more efficiently
201 allocating context stack entries.
202 - implement Coro->join and Coro->on_destroy in XS for a speedup and
203 a reduction in memory use.
13 - cancelling a coro while it itself is cancelling another coro is 204 - cancelling a coro while it itself is cancelling another coro is
14 now supported and working, instead of triggering an assertion. 205 now supported and working, instead of triggering an assertion.
15 - be a bit more crash-resistant when calling (buggy) on_destroy 206 - be a bit more crash-resistant when calling (buggy) on_destroy
16 callbacks (best effort). 207 callbacks (best effort).
17 - move on_destroy into the slf_frame, to allow extension slf 208 - move on_destroy into the slf_frame, to allow extension slf
18 functions to have destructors. 209 functions to have destructors.
210 - get rid if coro refcounting - simply crash in other interpreter
211 threads by nulling the pointers on clone.
212 - simplify warn/die hook handling when loading Coro - the convoluted
213 logic seems to be no longer neccessary.
214 - use libecb instead of our own home-grown gcc hacks.
215 - document alternatives to Coro::LWP. Please use them :)
216 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
217 sigaltstack. Really. wine suffers from it, erlang suffers from it,
218 and it's known since at least 2006.
19 219
205.372 Wed Feb 23 06:14:30 CET 2011 2205.372 Wed Feb 23 06:14:30 CET 2011
21 - apparently mingw doesn't provide a working gettimeofday, try to 221 - apparently mingw doesn't provide a working gettimeofday, try to
22 work around that by relying on Time::HiRes (indirectly brought to 222 work around that by relying on Time::HiRes (indirectly brought to
23 my attention by Max Maischein). 223 my attention by Max Maischein).
… …
341 - ->throw is now supported on Coro::State objects. 541 - ->throw is now supported on Coro::State objects.
342 - clean up cctx creation code a bit. 542 - clean up cctx creation code a bit.
343 - entersub is actually a UNOP, not a LOGOP (not a bugfix). 543 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
344 544
3454.9 Sat Nov 8 17:45:27 CET 2008 5454.9 Sat Nov 8 17:45:27 CET 2008
346 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 546 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
347 - (libcoro) no longer leak threads in the experimental pthread backend, 547 - (libcoro) no longer leak threads in the experimental pthread backend,
348 also speed it up considerably. 548 also speed it up considerably.
349 - (libcoro) do not rely on makecontext passing void *'s unscathed. 549 - (libcoro) do not rely on makecontext passing void *'s unscathed.
350 - fix compiletime dependencies on libcoro in the Makefile. 550 - fix compiletime dependencies on libcoro in the Makefile.
351 - cctx_count wasn't always updated properly. 551 - cctx_count wasn't always updated properly.
… …
413 - remove debugging code related to MgPV_nolen_const, also try to 613 - remove debugging code related to MgPV_nolen_const, also try to
414 make it compile with perl 5.8.6 (yes, apple apparently loves 614 make it compile with perl 5.8.6 (yes, apple apparently loves
415 outdated software). Reported by John S. 615 outdated software). Reported by John S.
416 616
4174.744 Tue Jul 8 22:06:35 CEST 2008 6174.744 Tue Jul 8 22:06:35 CEST 2008
418 - correctly provide default DIE/WARN handlers as documented. 618 - correctly provide default DIE/WARN handlers as documented.
419 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 619 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
420 implementations inside perl work fine for us. 620 implementations inside perl work fine for us.
421 621
4224.743 Mon Jun 16 00:21:57 CEST 2008 6224.743 Mon Jun 16 00:21:57 CEST 2008
423 - when using Coro::EV without running EV::loop it could 623 - when using Coro::EV without running EV::loop it could
… …
445 - sprinkle "no warnings" freely over everything, also suppress 645 - sprinkle "no warnings" freely over everything, also suppress
446 warnings for some other modules. 646 warnings for some other modules.
447 - fix typo in WSAEWOULDBLOCK. 647 - fix typo in WSAEWOULDBLOCK.
448 648
4494.72 Sun May 25 05:14:36 CEST 2008 6494.72 Sun May 25 05:14:36 CEST 2008
450 - tweak META.yaml a bit, unfortunately, there is no documented way 650 - tweak META.yaml a bit, unfortunately, there is no documented way
451 to have optional dependencies with CPAN. doh :( 651 to have optional dependencies with CPAN. doh :(
452 - avoid running some tests on windows because they would fail due to 652 - avoid running some tests on windows because they would fail due to
453 perl bug (broken fork, broken pipes...). 653 perl bug (broken fork, broken pipes...).
454 - work around perl on windows bugs where perl returns undocumented 654 - work around perl on windows bugs where perl returns undocumented
455 error codes for sysread, syswrite etc. by taking advantage 655 error codes for sysread, syswrite etc. by taking advantage
… …
459 - use unix domain sockets in testsuite to work around 659 - use unix domain sockets in testsuite to work around
460 common perl implementation bugs on widows (they are emulated by 660 common perl implementation bugs on widows (they are emulated by
461 tcp sockets on windows. ugh.) 661 tcp sockets on windows. ugh.)
462 662
4634.71 Sat May 24 20:01:27 CEST 2008 6634.71 Sat May 24 20:01:27 CEST 2008
464 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 664 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
465 - take advantage of async name resolution of AnyEvent::Util. 665 - take advantage of async name resolution of AnyEvent::Util.
466 - work around brutal inet_aton override in Coro::LWP. 666 - work around brutal inet_aton override in Coro::LWP.
467 - take advantage of the readyhook in Coro::EV, for smoother 667 - take advantage of the readyhook in Coro::EV, for smoother
468 scheduling. 668 scheduling.
469 669
4704.7 Sun May 11 00:32:19 CEST 2008 6704.7 Sun May 11 00:32:19 CEST 2008
471 - completely reworked the Coro manpage. 671 - completely reworked the Coro manpage.
472 - added Coro::AnyEvent, generic event loop integration. 672 - added Coro::AnyEvent, generic event loop integration.
473 - implement cancel, ready and kill commands in Coro::Debug. 673 - implement cancel, ready and kill commands in Coro::Debug.
474 - document find_coro in Coro::Debug. 674 - document find_coro in Coro::Debug.
475 - incompatible change: rename has_stack to has_cctx. 675 - incompatible change: rename has_stack to has_cctx.
476 - Coro::AIO and Coro::BDB no longer force event model detection, 676 - Coro::AIO and Coro::BDB no longer force event model detection,
477 use AnyEvent::AIO and AnyEvent::BDB. 677 use AnyEvent::AIO and AnyEvent::BDB.
… …
601 801
6024.11 Thu Oct 11 02:40:24 CEST 2007 8024.11 Thu Oct 11 02:40:24 CEST 2007
603 - port to threaded perls. 803 - port to threaded perls.
604 804
6054.1 Thu Oct 11 02:38:16 CEST 2007 8054.1 Thu Oct 11 02:38:16 CEST 2007
606 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 806 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
607 be local to each coro (see Coro::State). 807 be local to each coro (see Coro::State).
608 - incompatible change: for very deep reasons, cede and cede_notself 808 - incompatible change: for very deep reasons, cede and cede_notself
609 cannot return anything, so nothing will be returned. 809 cannot return anything, so nothing will be returned.
610 - possibly bring back 5.10 compatibility (untested). 810 - possibly bring back 5.10 compatibility (untested).
611 - work around stupid (and wrong) warning on 5.10 :(. 811 - work around stupid (and wrong) warning on 5.10 :(.
… …
683 but stateful semantics. 883 but stateful semantics.
684 - fixed a lot of typos in Coro.pm (patch submitted by David 884 - fixed a lot of typos in Coro.pm (patch submitted by David
685 Steinbrunner, which applied flawlessly). 885 Steinbrunner, which applied flawlessly).
686 886
6873.6 Sat Apr 14 17:13:31 CEST 2007 8873.6 Sat Apr 14 17:13:31 CEST 2007
688 - added some bugfixes to get eg/myhttpd working again. 888 - added some bugfixes to get eg/myhttpd working again.
689 - added Coro::Storable for often-cede'ing freeze/thaw. 889 - added Coro::Storable for often-cede'ing freeze/thaw.
690 - try to do a clean exit when a coroutine calls exit 890 - try to do a clean exit when a coroutine calls exit
691 (EXPERIMENTAL). 891 (EXPERIMENTAL).
692 - got rid of indirect call through _coro_init. 892 - got rid of indirect call through _coro_init.
693 - updated the partly antique examples in eg/ to 893 - updated the partly antique examples in eg/ to

Diff Legend

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