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

Comparing Coro/Changes (file contents):
Revision 1.571 by root, Tue Jun 28 22:18:57 2011 UTC vs.
Revision 1.685 by root, Wed Mar 11 09:47:16 2020 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?
8TODO: how to debug safe_cancel failure (vebrose mode? show how to make backtrace?)
9TODO: scalar context for Coro::AIO wrappers?
6 10
7TODO: tutorial 11TODO: State.xs could short-circuit some macros such as CORO_EXECUTE_SLF
8TODO: in_eval, PL_tained 12TODO: use frame->destroy to improve Coro::EV
13
146.56 Thu Feb 20 19:40:07 CET 2020
15 - Coro::EV::time*once functions used ev_once, which is not cancelable.
16 This could cause problems due to I/O watchers not being stopped up
17 on exceptions/cancellation.
18
196.55 Fri Jul 19 17:28:09 CEST 2019
20 - Coro::RWLock woke up only one reader instead of all after a wrlock
21 (analyzed and patch by Maxim Skorobogatov).
22 - slightly improved documentation of Coro::RWLock and RWLock testcase.
23 - quickly (re)-ported to minix 3.3 before minix crashed again.
24
256.54 Sun Dec 16 11:06:07 CET 2018
26 - reinstate Coro::Select::select prototype - it breaks more than it fixes.
27
286.53 Sun Dec 16 10:33:15 CET 2018
29 - the autogenerated prototypes of Coro::AIO now add ";" at the end
30 to avoid generating named unary operators and messing up parsing
31 (mostly affects aio_mlockall 4 | 8 which was parsed as
32 (aio_mlockall 4) | 8).
33 - do not use a prototype for select in Coro::Select, as CORE::select also
34 doesn't have one.
35
366.52 Tue Aug 14 18:49:43 CEST 2018
37 - libcoro update: allow sharing of coro threads using the pthreads
38 backend among different pthreads, as is required by Coro::Multicore.
39 - always allow ->call and ->eval when the coro is the current coroutine. this
40 makes it possible, among other things, to call these on $Coro::main.
41 - minor code cleanups.
42
436.514 Thu Aug 31 18:25:31 CEST 2017
44 - safe_cancel no longer croaks when called on already-destroyed
45 thread, but instead "just works".
46
476.513 Sat Jul 15 01:18:51 CEST 2017
48 - experimental 5.22 comaptibility changes (PadlistNAMES).
49 - removed attributes example and made sure the other examples
50 still work.
51 - fix some statevar types for newer perls, should be of no consequence.
52
536.512 Fri Jul 14 05:03:17 CEST 2017
54 - seems to pass test suite down till 5.8.9, so make this the
55 minimum requirement in Makefile.PL.
56 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
57 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
58 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
59 fortified longjmp variants in some libcs.
60 - re-try JIT allocation without PROT_EXEC, to hopefully improve
61 portability (still doesn't seem to work around broken selinux
62 mmap everywhere).
63 - speling fixes (Xavier Guimard).
64
656.511 Sun Jun 26 23:44:50 CEST 2016
66 - make it compile with threaded perls again.
67 - simplify and speed up __DIE__ and __WARN__ handling.
68
696.51 Sat Jun 25 20:57:20 CEST 2016
70 - 6.5 release didn't compile on most older perls due to
71 a logic error enabling experimental 5.24 code for all versions.
72 - since various broken or deficient patches float around
73 to work around the vtbl API breakage in perl, let's provide
74 our own workaround, which seems to be much less invasive
75 and more compatible than the approaches seen so far. YMMV.
76 - work around assertion failure in perl_destruct on 5.24
77 (analyzed by Dave Mitchell).
78
796.5 Wed Jun 22 22:23:50 CEST 2016
80 - swap_sv swapping order was not symmetrical, causing
81 wrong swaps when swap_sv was used multiple times on the
82 same sv in the same thread.
83 - swap_sv calls can now be undone by calling it again
84 with the same variables.
85 - swap_sv calls will now be undone in async_pool threads.
86 - split Coro::Semaphore::up/adjust into separate xs functions
87 for better error reporting, at slight codesize increase.
88 - (libcoro) arm assembly support, please test and report.
89 - adjust to PL_savestack changes in perl 5.24 (adapted from
90 the debian patch, which unfortunately gets it wrong).
91
926.49 Sat Oct 17 01:40:12 CEST 2015
93 - throwing an exception to a thread waiting in
94 Coro::Handle using Coro::EV did not stop the watchers,
95 causing the next call to fail (testcase by Martin Pritchard).
96 - bump minimum perl version to 5.10.
97
986.48 Sun Oct 4 19:03:51 CEST 2015
99 - fix memory leak when Coro::Handle uses Coro::EV internally
100 (testcase by Sten Sten).
101 - update code to libev 4 API (internally, Coro still used the
102 version 3 API calls).
103 - remove unused hv_sig.
104
1056.47 Sat Jul 11 03:58:20 CEST 2015
106 - swap_svs - NVs can be in the SV head beginning with 5.20.
107
1086.46 Tue Jun 30 14:36:30 CEST 2015
109 - restore portability to perls with windows fork emulation
110 (patch by Petr Písař).
111
1126.45 Tue Jun 30 01:40:08 CEST 2015
113 - restore portability to perl 5.16 and below.
114
1156.44 Tue Jun 30 00:41:54 CEST 2015
116 - avoid segfaulting (or worse) when tracing a canceled thread.
117 - the Event module silently broke it's public hook API, causing
118 failures on perls compiled with -Duselongdouble. Requiring
119 the latest version of Event because I don't know which release
120 changed this (it's not mentioned in the ChangeLog).
121 - allow xs level enterleave hooks via CoroAPI.
122
1236.43 Thu Jun 4 15:38:14 CEST 2015
124 - use stability canary.
125 - port to stableperl-5.22.0-1.001.
126 - update libecb.
127
1286.42 Wed Feb 11 20:29:52 CET 2015
129 - Coro::SemaphoreSet->try did not actually work (analyzed by
130 SATO Kentaro).
131 - upgrade libecb to be C11 compliant.
132
1336.41 Sat Sep 6 22:08:46 CEST 2014
134 - restore portability to perl 5.8.x.
135 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
136 up some memory, but fixes "use feature" and similar modules
137 when used inside a Coro.
138 - allow zero as argument to Coro::Channel to mean the same thing
139 as no argument. this works with older versions as well,
140 but wasn't legal until now.
141 - slightly better c header file detection.
142
1436.39 Mon Jun 2 00:00:08 CEST 2014
144 - work around more incompatible changes in 5.20.
145
1466.38 Sun Jun 1 21:54:23 CEST 2014
147 - check that perl slots actually have enough space to hold
148 interpreter variables.
149 - untested port to perl 5.19 (64 bit tmps indices) (reported
150 by Andreas König).
151 - croak when cancel is called without a thread context.
152
1536.37 Tue Mar 4 13:27:33 CET 2014
154 - *sigh*, removed leftover debugging code from debugging a
155 perl bug, of all things.
156
1576.36 Tue Mar 4 07:11:59 CET 2014
158 - semaphores would not clear the destroy hook when interrupted
159 by ->throw, causing segfaults or worse.
160 - ->throw on a thread waiting for a semaphore did not acquire
161 the semaphore, but also didn't wake up other waiters,
162 possibly causing a deadlock.
163 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
164 to report a full stacktrace, which should help find out
165 where the actual call is.
166 - "atomically" destroy data in slf_destroy, because it is
167 the right thing to do, just in case.
168 - disable ecb.h libm dependency, to compile on ancient systems
169 or under adverse conditions.
170
1716.33 Mon Nov 18 11:26:27 CET 2013
172 - do not crash when freeing padlists with holes (in 5.18).
173 - tentative SVt_BIND 5.19 port/fix.
174
1756.32 Tue Nov 5 15:35:35 CET 2013
176 - use a new algorithm to derive padlists for perl 5.18. The old
177 one could lead to 0-pointer accesses inside perl (reported
178 by Darin McBride).
179
1806.31 Thu May 9 07:39:48 CEST 2013
181 - Coro::AIO requests would crash if the thread was ready'd
182 while the request was ongoing.
183
1846.29 Wed May 8 02:55:18 CEST 2013
185 - when an on_destroy handler destructs the coro currently being
186 destructed a perl scalar could be accessed after being freed,
187 likely causing a crash.
188
1896.28 Wed Mar 6 06:58:02 CET 2013
190 - clean remnants of existing __DIE__ and __WARN__ handlers so
191 they lose their magic and will not cause segfaults later
192 (testcase by Andrey Sagulin).
193 - improved Coro::State documentation a bit.
194 - Coro::Debug::command now flushes the output.
195 - add hack detection code for x32 abi, because the braindead slugs
196 who designed that made it look exactly like x86_64 without
197 providing proper compile time symbols to test for it. as a result,
198 this detection cannot work reliably.
199 - valgrind stack registering was broken.
200 - do not rely on Time::HiRes anymore in Coro::Debug.
201
2026.23 Fri Dec 7 23:36:37 CET 2012
203 - use experimental fiber implementation on native windows
204 perls.
205 - use sizeof (void *) as multiplication factor for stack sizes,
206 to accomodate the totally braindamaged microsoft 64 bit "os".
207 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
208 activeperl 5.16 crashes when PerlIO_define_layer is called due
209 to some bug in the perl dll, strawberry perl at least passes
210 the testsuite.
211 - implement Coro::Handle->peeraddr/host/port, for slightly
212 improved compatibility with LWP.
213 - implement 5.17 compatibility by almost blindly applying a
214 good-looking patch by Father Chrysostomos.
215 - move stack management functions into libcoro 3.
216 - libcoro version 3 "released".
217 - support magic values as timed_io_once args.
218 - recommend AnyEvent 7+ or EV 4+, also require EV
219 version 4 or newer for Coro::EV.
220
2216.10 Tue Oct 9 01:14:27 CEST 2012
222 - updated ecb.h, it had a typo that caused it to not compile on many
223 big endian systems (reported by many people).
224 - disable memory fences in ecb.h to improve portability.
225
2266.09 Sat Oct 6 23:25:02 CEST 2012
227 - Coro::EV I/O watchers were not interruptible by exceptions
228 (Coro::State::throw) (testcase by sten).
229 - ->throw now puts threads into the ready queue, as this seems to
230 be expected by existing code, and code that doesn't cope with spurious
231 wakeups needs fixing anyway.
232 - use fd -1 in mmap.
233 - cast I32 to int in error message printf.
234 - warn about broken so-called "hardened" kernels.
235
2366.08 Fri Apr 13 12:05:47 CEST 2012
237 - be more aggressive about exiting like perl does - formerly,
238 exiting from the non-main thread would not execute END blocks.
239
2406.07 Fri Nov 11 21:21:48 CET 2011
241 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
242 - update ecb.h.
243
2446.06 Mon Aug 8 23:59:48 CEST 2011
245 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
246 backend, which is an order of magnitude slower. unfortunately, cygwins
247 pthread implementation isn't very complete either, so allocate the stack
248 twice just to be sure.
249 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
250 in its ithreaded perl - best recompile cygwin and use the 'w'indows
251 backend for much better performance. also disable ithreads for
252 even better performance...).
253
2546.05 Thu Aug 4 21:36:36 CEST 2011
255 - blush, condvar values would not be propagated from send to recv anymore
256 (reported by Chip Salzenberg).
257 - use exponential increase for the readline buffer length in
258 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
259
2606.04 Wed Aug 3 17:15:45 CEST 2011
261 - use even more efficient and more compatible condvars for
262 compatibility to AnyEvent 6.x :)
263 - more inconsequential ecb.h updates.
264
2656.03 Wed Aug 3 11:41:30 CEST 2011
266 - change how Coro patches AnyEvent condvars for compatibility to
267 AnyEvent 6.x.
268 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
269 hasn't been included.
270
2716.02 Wed Jul 13 04:35:19 CEST 2011
272 - "improve portability to Gentoo" - gentoo manages to put perl variables
273 in memory areas that are farther than 2gb apart, which the jit couldn't
274 handle and barfed. now it's just a bit slower on gentoo and similar
275 systems.
276
2776.01 Sun Jul 3 12:31:14 CEST 2011
278 - workarounds are good, but the test for whether pthreads are used
279 was not good. this one should be better.
280 - check differently whether gcc generates cfi instructions itself.
281
2826.0 Wed Jun 29 19:43:35 CEST 2011
9 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be 283 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
10 destroyed and cleaned up automatically (e.g. async { schedule }). 284 destroyed and cleaned up automatically (e.g. async { schedule }).
11 - implement a JIT compiler for part of the thread switch code, 285 - implement a JIT compiler for part of the thread switch code,
12 which gives a 50% speed improvement on threaded perls, and 286 which gives a 50% speed improvement on threaded perls, and
13 about 4% on non-thraeded perls (so threaded perls now finally 287 about 4% on non-threaded perls (so threaded perls now finally
14 reach about half the speed of non-threaded perls). 288 reach about half the speed of non-threaded perls).
289 - slightly modernise Coro::Intro, add section about rouse functions.
15 - avoid DEFSV and ERRSV, giving another 10% improvement 290 - avoid DEFSV and ERRSV, giving another 10% improvement
16 in thread switching. 291 in thread switching.
17 - Coro::State->is_destroyed is now called is_zombie. 292 - Coro::State->is_destroyed is now called is_zombie.
18 - implement a Coro->safe_cancel method that might fail, but 293 - implement a Coro->safe_cancel method that might fail, but
19 cancels in a "safer" way if it succeeds. 294 cancels in a "safer" way if it succeeds.
295 - add preliminary support for DEBUGGING perls.
20 - get rid of two hash-accesses when initialising a new Coro - this 296 - get rid of two hash-accesses when initialising a new Coro - this
21 speeds up coro creation by almost a factor of two. 297 speeds up coro creation by almost a factor of two.
22 - croak when a coro that is being cancelled tries to block 298 - croak when a coro that is being cancelled tries to block
23 (e.g. while executing a guard block), instead of crashing or 299 (e.g. while executing a guard block), instead of crashing or
24 deadlocking. 300 deadlocking.
… …
41 - simplify warn/die hook handling when loading Coro - the convoluted 317 - simplify warn/die hook handling when loading Coro - the convoluted
42 logic seems to be no longer neccessary. 318 logic seems to be no longer neccessary.
43 - use libecb instead of our own home-grown gcc hacks. 319 - use libecb instead of our own home-grown gcc hacks.
44 - document alternatives to Coro::LWP. Please use them :) 320 - document alternatives to Coro::LWP. Please use them :)
45 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds 321 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
46 sigaltstack. 322 sigaltstack. Really. wine suffers from it, erlang suffers from it,
323 and it's known since at least 2006.
47 324
485.372 Wed Feb 23 06:14:30 CET 2011 3255.372 Wed Feb 23 06:14:30 CET 2011
49 - apparently mingw doesn't provide a working gettimeofday, try to 326 - apparently mingw doesn't provide a working gettimeofday, try to
50 work around that by relying on Time::HiRes (indirectly brought to 327 work around that by relying on Time::HiRes (indirectly brought to
51 my attention by Max Maischein). 328 my attention by Max Maischein).
… …
369 - ->throw is now supported on Coro::State objects. 646 - ->throw is now supported on Coro::State objects.
370 - clean up cctx creation code a bit. 647 - clean up cctx creation code a bit.
371 - entersub is actually a UNOP, not a LOGOP (not a bugfix). 648 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
372 649
3734.9 Sat Nov 8 17:45:27 CET 2008 6504.9 Sat Nov 8 17:45:27 CET 2008
374 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 651 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
375 - (libcoro) no longer leak threads in the experimental pthread backend, 652 - (libcoro) no longer leak threads in the experimental pthread backend,
376 also speed it up considerably. 653 also speed it up considerably.
377 - (libcoro) do not rely on makecontext passing void *'s unscathed. 654 - (libcoro) do not rely on makecontext passing void *'s unscathed.
378 - fix compiletime dependencies on libcoro in the Makefile. 655 - fix compiletime dependencies on libcoro in the Makefile.
379 - cctx_count wasn't always updated properly. 656 - cctx_count wasn't always updated properly.
… …
441 - remove debugging code related to MgPV_nolen_const, also try to 718 - remove debugging code related to MgPV_nolen_const, also try to
442 make it compile with perl 5.8.6 (yes, apple apparently loves 719 make it compile with perl 5.8.6 (yes, apple apparently loves
443 outdated software). Reported by John S. 720 outdated software). Reported by John S.
444 721
4454.744 Tue Jul 8 22:06:35 CEST 2008 7224.744 Tue Jul 8 22:06:35 CEST 2008
446 - correctly provide default DIE/WARN handlers as documented. 723 - correctly provide default DIE/WARN handlers as documented.
447 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 724 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
448 implementations inside perl work fine for us. 725 implementations inside perl work fine for us.
449 726
4504.743 Mon Jun 16 00:21:57 CEST 2008 7274.743 Mon Jun 16 00:21:57 CEST 2008
451 - when using Coro::EV without running EV::loop it could 728 - when using Coro::EV without running EV::loop it could
… …
473 - sprinkle "no warnings" freely over everything, also suppress 750 - sprinkle "no warnings" freely over everything, also suppress
474 warnings for some other modules. 751 warnings for some other modules.
475 - fix typo in WSAEWOULDBLOCK. 752 - fix typo in WSAEWOULDBLOCK.
476 753
4774.72 Sun May 25 05:14:36 CEST 2008 7544.72 Sun May 25 05:14:36 CEST 2008
478 - tweak META.yaml a bit, unfortunately, there is no documented way 755 - tweak META.yaml a bit, unfortunately, there is no documented way
479 to have optional dependencies with CPAN. doh :( 756 to have optional dependencies with CPAN. doh :(
480 - avoid running some tests on windows because they would fail due to 757 - avoid running some tests on windows because they would fail due to
481 perl bug (broken fork, broken pipes...). 758 perl bug (broken fork, broken pipes...).
482 - work around perl on windows bugs where perl returns undocumented 759 - work around perl on windows bugs where perl returns undocumented
483 error codes for sysread, syswrite etc. by taking advantage 760 error codes for sysread, syswrite etc. by taking advantage
… …
487 - use unix domain sockets in testsuite to work around 764 - use unix domain sockets in testsuite to work around
488 common perl implementation bugs on widows (they are emulated by 765 common perl implementation bugs on widows (they are emulated by
489 tcp sockets on windows. ugh.) 766 tcp sockets on windows. ugh.)
490 767
4914.71 Sat May 24 20:01:27 CEST 2008 7684.71 Sat May 24 20:01:27 CEST 2008
492 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 769 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
493 - take advantage of async name resolution of AnyEvent::Util. 770 - take advantage of async name resolution of AnyEvent::Util.
494 - work around brutal inet_aton override in Coro::LWP. 771 - work around brutal inet_aton override in Coro::LWP.
495 - take advantage of the readyhook in Coro::EV, for smoother 772 - take advantage of the readyhook in Coro::EV, for smoother
496 scheduling. 773 scheduling.
497 774
4984.7 Sun May 11 00:32:19 CEST 2008 7754.7 Sun May 11 00:32:19 CEST 2008
499 - completely reworked the Coro manpage. 776 - completely reworked the Coro manpage.
500 - added Coro::AnyEvent, generic event loop integration. 777 - added Coro::AnyEvent, generic event loop integration.
501 - implement cancel, ready and kill commands in Coro::Debug. 778 - implement cancel, ready and kill commands in Coro::Debug.
502 - document find_coro in Coro::Debug. 779 - document find_coro in Coro::Debug.
503 - incompatible change: rename has_stack to has_cctx. 780 - incompatible change: rename has_stack to has_cctx.
504 - Coro::AIO and Coro::BDB no longer force event model detection, 781 - Coro::AIO and Coro::BDB no longer force event model detection,
505 use AnyEvent::AIO and AnyEvent::BDB. 782 use AnyEvent::AIO and AnyEvent::BDB.
… …
629 906
6304.11 Thu Oct 11 02:40:24 CEST 2007 9074.11 Thu Oct 11 02:40:24 CEST 2007
631 - port to threaded perls. 908 - port to threaded perls.
632 909
6334.1 Thu Oct 11 02:38:16 CEST 2007 9104.1 Thu Oct 11 02:38:16 CEST 2007
634 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 911 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
635 be local to each coro (see Coro::State). 912 be local to each coro (see Coro::State).
636 - incompatible change: for very deep reasons, cede and cede_notself 913 - incompatible change: for very deep reasons, cede and cede_notself
637 cannot return anything, so nothing will be returned. 914 cannot return anything, so nothing will be returned.
638 - possibly bring back 5.10 compatibility (untested). 915 - possibly bring back 5.10 compatibility (untested).
639 - work around stupid (and wrong) warning on 5.10 :(. 916 - work around stupid (and wrong) warning on 5.10 :(.
… …
711 but stateful semantics. 988 but stateful semantics.
712 - fixed a lot of typos in Coro.pm (patch submitted by David 989 - fixed a lot of typos in Coro.pm (patch submitted by David
713 Steinbrunner, which applied flawlessly). 990 Steinbrunner, which applied flawlessly).
714 991
7153.6 Sat Apr 14 17:13:31 CEST 2007 9923.6 Sat Apr 14 17:13:31 CEST 2007
716 - added some bugfixes to get eg/myhttpd working again. 993 - added some bugfixes to get eg/myhttpd working again.
717 - added Coro::Storable for often-cede'ing freeze/thaw. 994 - added Coro::Storable for often-cede'ing freeze/thaw.
718 - try to do a clean exit when a coroutine calls exit 995 - try to do a clean exit when a coroutine calls exit
719 (EXPERIMENTAL). 996 (EXPERIMENTAL).
720 - got rid of indirect call through _coro_init. 997 - got rid of indirect call through _coro_init.
721 - updated the partly antique examples in eg/ to 998 - updated the partly antique examples in eg/ to

Diff Legend

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