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

Comparing Coro/Changes (file contents):
Revision 1.551 by root, Sat Feb 19 06:51:22 2011 UTC vs.
Revision 1.687 by root, Wed Apr 22 19:21:20 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?
10
11TODO: State.xs could short-circuit some macros such as CORO_EXECUTE_SLF
12TODO: use frame->destroy to improve Coro::EV
136.57
14 - allow rouse_wait to be called in a different thread.
15 - slightly clarify rouse_cb/rouse_wait documentation.
16 - slightly improve rouse callback performance and memory usage.
17 - try to report correct calller when throwing exception in
18 on $condvar->recv, not a useless line in Coro::AnyEvent.
19
206.56 Thu Feb 20 19:40:07 CET 2020
21 - Coro::EV::time*once functions used ev_once, which is not cancelable.
22 This could cause problems due to I/O watchers not being stopped up
23 on exceptions/cancellation.
24
256.55 Fri Jul 19 17:28:09 CEST 2019
26 - Coro::RWLock woke up only one reader instead of all after a wrlock
27 (analyzed and patch by Maxim Skorobogatov).
28 - slightly improved documentation of Coro::RWLock and RWLock testcase.
29 - quickly (re)-ported to minix 3.3 before minix crashed again.
30
316.54 Sun Dec 16 11:06:07 CET 2018
32 - reinstate Coro::Select::select prototype - it breaks more than it fixes.
33
346.53 Sun Dec 16 10:33:15 CET 2018
35 - the autogenerated prototypes of Coro::AIO now add ";" at the end
36 to avoid generating named unary operators and messing up parsing
37 (mostly affects aio_mlockall 4 | 8 which was parsed as
38 (aio_mlockall 4) | 8).
39 - do not use a prototype for select in Coro::Select, as CORE::select also
40 doesn't have one.
41
426.52 Tue Aug 14 18:49:43 CEST 2018
43 - libcoro update: allow sharing of coro threads using the pthreads
44 backend among different pthreads, as is required by Coro::Multicore.
45 - always allow ->call and ->eval when the coro is the current coroutine. this
46 makes it possible, among other things, to call these on $Coro::main.
47 - minor code cleanups.
48
496.514 Thu Aug 31 18:25:31 CEST 2017
50 - safe_cancel no longer croaks when called on already-destroyed
51 thread, but instead "just works".
52
536.513 Sat Jul 15 01:18:51 CEST 2017
54 - experimental 5.22 comaptibility changes (PadlistNAMES).
55 - removed attributes example and made sure the other examples
56 still work.
57 - fix some statevar types for newer perls, should be of no consequence.
58
596.512 Fri Jul 14 05:03:17 CEST 2017
60 - seems to pass test suite down till 5.8.9, so make this the
61 minimum requirement in Makefile.PL.
62 - experimental 5.24 compatibility changes (SUB_ARGARRAY).
63 - experimental 5.26 compatibility changes (PL_comppad -> PADOFFSET).
64 - disable FORTIFY_SOURCE to avoid broken (but well-intentioned)
65 fortified longjmp variants in some libcs.
66 - re-try JIT allocation without PROT_EXEC, to hopefully improve
67 portability (still doesn't seem to work around broken selinux
68 mmap everywhere).
69 - speling fixes (Xavier Guimard).
70
716.511 Sun Jun 26 23:44:50 CEST 2016
72 - make it compile with threaded perls again.
73 - simplify and speed up __DIE__ and __WARN__ handling.
74
756.51 Sat Jun 25 20:57:20 CEST 2016
76 - 6.5 release didn't compile on most older perls due to
77 a logic error enabling experimental 5.24 code for all versions.
78 - since various broken or deficient patches float around
79 to work around the vtbl API breakage in perl, let's provide
80 our own workaround, which seems to be much less invasive
81 and more compatible than the approaches seen so far. YMMV.
82 - work around assertion failure in perl_destruct on 5.24
83 (analyzed by Dave Mitchell).
84
856.5 Wed Jun 22 22:23:50 CEST 2016
86 - swap_sv swapping order was not symmetrical, causing
87 wrong swaps when swap_sv was used multiple times on the
88 same sv in the same thread.
89 - swap_sv calls can now be undone by calling it again
90 with the same variables.
91 - swap_sv calls will now be undone in async_pool threads.
92 - split Coro::Semaphore::up/adjust into separate xs functions
93 for better error reporting, at slight codesize increase.
94 - (libcoro) arm assembly support, please test and report.
95 - adjust to PL_savestack changes in perl 5.24 (adapted from
96 the debian patch, which unfortunately gets it wrong).
97
986.49 Sat Oct 17 01:40:12 CEST 2015
99 - throwing an exception to a thread waiting in
100 Coro::Handle using Coro::EV did not stop the watchers,
101 causing the next call to fail (testcase by Martin Pritchard).
102 - bump minimum perl version to 5.10.
103
1046.48 Sun Oct 4 19:03:51 CEST 2015
105 - fix memory leak when Coro::Handle uses Coro::EV internally
106 (testcase by Sten Sten).
107 - update code to libev 4 API (internally, Coro still used the
108 version 3 API calls).
109 - remove unused hv_sig.
110
1116.47 Sat Jul 11 03:58:20 CEST 2015
112 - swap_svs - NVs can be in the SV head beginning with 5.20.
113
1146.46 Tue Jun 30 14:36:30 CEST 2015
115 - restore portability to perls with windows fork emulation
116 (patch by Petr Písař).
117
1186.45 Tue Jun 30 01:40:08 CEST 2015
119 - restore portability to perl 5.16 and below.
120
1216.44 Tue Jun 30 00:41:54 CEST 2015
122 - avoid segfaulting (or worse) when tracing a canceled thread.
123 - the Event module silently broke it's public hook API, causing
124 failures on perls compiled with -Duselongdouble. Requiring
125 the latest version of Event because I don't know which release
126 changed this (it's not mentioned in the ChangeLog).
127 - allow xs level enterleave hooks via CoroAPI.
128
1296.43 Thu Jun 4 15:38:14 CEST 2015
130 - use stability canary.
131 - port to stableperl-5.22.0-1.001.
132 - update libecb.
133
1346.42 Wed Feb 11 20:29:52 CET 2015
135 - Coro::SemaphoreSet->try did not actually work (analyzed by
136 SATO Kentaro).
137 - upgrade libecb to be C11 compliant.
138
1396.41 Sat Sep 6 22:08:46 CEST 2014
140 - restore portability to perl 5.8.x.
141 - give new Coro's a valid GvHV(PL_hintgv) - this is slow and takes
142 up some memory, but fixes "use feature" and similar modules
143 when used inside a Coro.
144 - allow zero as argument to Coro::Channel to mean the same thing
145 as no argument. this works with older versions as well,
146 but wasn't legal until now.
147 - slightly better c header file detection.
148
1496.39 Mon Jun 2 00:00:08 CEST 2014
150 - work around more incompatible changes in 5.20.
151
1526.38 Sun Jun 1 21:54:23 CEST 2014
153 - check that perl slots actually have enough space to hold
154 interpreter variables.
155 - untested port to perl 5.19 (64 bit tmps indices) (reported
156 by Andreas König).
157 - croak when cancel is called without a thread context.
158
1596.37 Tue Mar 4 13:27:33 CET 2014
160 - *sigh*, removed leftover debugging code from debugging a
161 perl bug, of all things.
162
1636.36 Tue Mar 4 07:11:59 CET 2014
164 - semaphores would not clear the destroy hook when interrupted
165 by ->throw, causing segfaults or worse.
166 - ->throw on a thread waiting for a semaphore did not acquire
167 the semaphore, but also didn't wake up other waiters,
168 possibly causing a deadlock.
169 - "FATAL: $Coro::IDLE blocked itself" will now use Carp::confess
170 to report a full stacktrace, which should help find out
171 where the actual call is.
172 - "atomically" destroy data in slf_destroy, because it is
173 the right thing to do, just in case.
174 - disable ecb.h libm dependency, to compile on ancient systems
175 or under adverse conditions.
176
1776.33 Mon Nov 18 11:26:27 CET 2013
178 - do not crash when freeing padlists with holes (in 5.18).
179 - tentative SVt_BIND 5.19 port/fix.
180
1816.32 Tue Nov 5 15:35:35 CET 2013
182 - use a new algorithm to derive padlists for perl 5.18. The old
183 one could lead to 0-pointer accesses inside perl (reported
184 by Darin McBride).
185
1866.31 Thu May 9 07:39:48 CEST 2013
187 - Coro::AIO requests would crash if the thread was ready'd
188 while the request was ongoing.
189
1906.29 Wed May 8 02:55:18 CEST 2013
191 - when an on_destroy handler destructs the coro currently being
192 destructed a perl scalar could be accessed after being freed,
193 likely causing a crash.
194
1956.28 Wed Mar 6 06:58:02 CET 2013
196 - clean remnants of existing __DIE__ and __WARN__ handlers so
197 they lose their magic and will not cause segfaults later
198 (testcase by Andrey Sagulin).
199 - improved Coro::State documentation a bit.
200 - Coro::Debug::command now flushes the output.
201 - add hack detection code for x32 abi, because the braindead slugs
202 who designed that made it look exactly like x86_64 without
203 providing proper compile time symbols to test for it. as a result,
204 this detection cannot work reliably.
205 - valgrind stack registering was broken.
206 - do not rely on Time::HiRes anymore in Coro::Debug.
207
2086.23 Fri Dec 7 23:36:37 CET 2012
209 - use experimental fiber implementation on native windows
210 perls.
211 - use sizeof (void *) as multiplication factor for stack sizes,
212 to accomodate the totally braindamaged microsoft 64 bit "os".
213 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
214 activeperl 5.16 crashes when PerlIO_define_layer is called due
215 to some bug in the perl dll, strawberry perl at least passes
216 the testsuite.
217 - implement Coro::Handle->peeraddr/host/port, for slightly
218 improved compatibility with LWP.
219 - implement 5.17 compatibility by almost blindly applying a
220 good-looking patch by Father Chrysostomos.
221 - move stack management functions into libcoro 3.
222 - libcoro version 3 "released".
223 - support magic values as timed_io_once args.
224 - recommend AnyEvent 7+ or EV 4+, also require EV
225 version 4 or newer for Coro::EV.
226
2276.10 Tue Oct 9 01:14:27 CEST 2012
228 - updated ecb.h, it had a typo that caused it to not compile on many
229 big endian systems (reported by many people).
230 - disable memory fences in ecb.h to improve portability.
231
2326.09 Sat Oct 6 23:25:02 CEST 2012
233 - Coro::EV I/O watchers were not interruptible by exceptions
234 (Coro::State::throw) (testcase by sten).
235 - ->throw now puts threads into the ready queue, as this seems to
236 be expected by existing code, and code that doesn't cope with spurious
237 wakeups needs fixing anyway.
238 - use fd -1 in mmap.
239 - cast I32 to int in error message printf.
240 - warn about broken so-called "hardened" kernels.
241
2426.08 Fri Apr 13 12:05:47 CEST 2012
243 - be more aggressive about exiting like perl does - formerly,
244 exiting from the non-main thread would not execute END blocks.
245
2466.07 Fri Nov 11 21:21:48 CET 2011
247 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
248 - update ecb.h.
249
2506.06 Mon Aug 8 23:59:48 CEST 2011
251 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
252 backend, which is an order of magnitude slower. unfortunately, cygwins
253 pthread implementation isn't very complete either, so allocate the stack
254 twice just to be sure.
255 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
256 in its ithreaded perl - best recompile cygwin and use the 'w'indows
257 backend for much better performance. also disable ithreads for
258 even better performance...).
259
2606.05 Thu Aug 4 21:36:36 CEST 2011
261 - blush, condvar values would not be propagated from send to recv anymore
262 (reported by Chip Salzenberg).
263 - use exponential increase for the readline buffer length in
264 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
265
2666.04 Wed Aug 3 17:15:45 CEST 2011
267 - use even more efficient and more compatible condvars for
268 compatibility to AnyEvent 6.x :)
269 - more inconsequential ecb.h updates.
270
2716.03 Wed Aug 3 11:41:30 CEST 2011
272 - change how Coro patches AnyEvent condvars for compatibility to
273 AnyEvent 6.x.
274 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
275 hasn't been included.
276
2776.02 Wed Jul 13 04:35:19 CEST 2011
278 - "improve portability to Gentoo" - gentoo manages to put perl variables
279 in memory areas that are farther than 2gb apart, which the jit couldn't
280 handle and barfed. now it's just a bit slower on gentoo and similar
281 systems.
282
2836.01 Sun Jul 3 12:31:14 CEST 2011
284 - workarounds are good, but the test for whether pthreads are used
285 was not good. this one should be better.
286 - check differently whether gcc generates cfi instructions itself.
287
2886.0 Wed Jun 29 19:43:35 CEST 2011
289 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
290 destroyed and cleaned up automatically (e.g. async { schedule }).
291 - implement a JIT compiler for part of the thread switch code,
292 which gives a 50% speed improvement on threaded perls, and
293 about 4% on non-threaded perls (so threaded perls now finally
294 reach about half the speed of non-threaded perls).
295 - slightly modernise Coro::Intro, add section about rouse functions.
296 - avoid DEFSV and ERRSV, giving another 10% improvement
297 in thread switching.
298 - Coro::State->is_destroyed is now called is_zombie.
299 - implement a Coro->safe_cancel method that might fail, but
300 cancels in a "safer" way if it succeeds.
301 - add preliminary support for DEBUGGING perls.
302 - get rid of two hash-accesses when initialising a new Coro - this
303 speeds up coro creation by almost a factor of two.
304 - croak when a coro that is being cancelled tries to block
305 (e.g. while executing a guard block), instead of crashing or
306 deadlocking.
307 - use a more robust and also faster method to identify Coro::State
308 objects - speeds up everything a bit.
309 - implement Coro->cancel in XS for a 20% speed improvement, and to
310 be able to implement mutual cancellation.
311 - speed up context switches by a percent or two by more efficiently
312 allocating context stack entries.
313 - implement Coro->join and Coro->on_destroy in XS for a speedup and
314 a reduction in memory use.
315 - cancelling a coro while it itself is cancelling another coro is
316 now supported and working, instead of triggering an assertion.
317 - be a bit more crash-resistant when calling (buggy) on_destroy
318 callbacks (best effort).
319 - move on_destroy into the slf_frame, to allow extension slf
320 functions to have destructors.
321 - get rid if coro refcounting - simply crash in other interpreter
322 threads by nulling the pointers on clone.
323 - simplify warn/die hook handling when loading Coro - the convoluted
324 logic seems to be no longer neccessary.
325 - use libecb instead of our own home-grown gcc hacks.
326 - document alternatives to Coro::LWP. Please use them :)
327 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
328 sigaltstack. Really. wine suffers from it, erlang suffers from it,
329 and it's known since at least 2006.
330
3315.372 Wed Feb 23 06:14:30 CET 2011
332 - apparently mingw doesn't provide a working gettimeofday, try to
333 work around that by relying on Time::HiRes (indirectly brought to
334 my attention by Max Maischein).
335 - fix some portability issues when Time::HiRes was used.
336
3375.371 Mon Feb 21 14:36:08 CET 2011
338 - backport to windows process emulation code again.
6 339
75.37 Sat Feb 19 07:49:44 CET 2011 3405.37 Sat Feb 19 07:49:44 CET 2011
8 - add a big "Coro thread life cycle" section to "man Coro". 341 - add a big "Coro thread life cycle" section to "man Coro".
9 - try a tentative workaround against the breakage that 5.13 has 342 - try a tentative workaround against the breakage that 5.13 has
10 introduced without depreciation period. sigh. 343 introduced without depreciation period. sigh.
… …
319 - ->throw is now supported on Coro::State objects. 652 - ->throw is now supported on Coro::State objects.
320 - clean up cctx creation code a bit. 653 - clean up cctx creation code a bit.
321 - entersub is actually a UNOP, not a LOGOP (not a bugfix). 654 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
322 655
3234.9 Sat Nov 8 17:45:27 CET 2008 6564.9 Sat Nov 8 17:45:27 CET 2008
324 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 657 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
325 - (libcoro) no longer leak threads in the experimental pthread backend, 658 - (libcoro) no longer leak threads in the experimental pthread backend,
326 also speed it up considerably. 659 also speed it up considerably.
327 - (libcoro) do not rely on makecontext passing void *'s unscathed. 660 - (libcoro) do not rely on makecontext passing void *'s unscathed.
328 - fix compiletime dependencies on libcoro in the Makefile. 661 - fix compiletime dependencies on libcoro in the Makefile.
329 - cctx_count wasn't always updated properly. 662 - cctx_count wasn't always updated properly.
… …
391 - remove debugging code related to MgPV_nolen_const, also try to 724 - remove debugging code related to MgPV_nolen_const, also try to
392 make it compile with perl 5.8.6 (yes, apple apparently loves 725 make it compile with perl 5.8.6 (yes, apple apparently loves
393 outdated software). Reported by John S. 726 outdated software). Reported by John S.
394 727
3954.744 Tue Jul 8 22:06:35 CEST 2008 7284.744 Tue Jul 8 22:06:35 CEST 2008
396 - correctly provide default DIE/WARN handlers as documented. 729 - correctly provide default DIE/WARN handlers as documented.
397 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 730 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
398 implementations inside perl work fine for us. 731 implementations inside perl work fine for us.
399 732
4004.743 Mon Jun 16 00:21:57 CEST 2008 7334.743 Mon Jun 16 00:21:57 CEST 2008
401 - when using Coro::EV without running EV::loop it could 734 - when using Coro::EV without running EV::loop it could
… …
423 - sprinkle "no warnings" freely over everything, also suppress 756 - sprinkle "no warnings" freely over everything, also suppress
424 warnings for some other modules. 757 warnings for some other modules.
425 - fix typo in WSAEWOULDBLOCK. 758 - fix typo in WSAEWOULDBLOCK.
426 759
4274.72 Sun May 25 05:14:36 CEST 2008 7604.72 Sun May 25 05:14:36 CEST 2008
428 - tweak META.yaml a bit, unfortunately, there is no documented way 761 - tweak META.yaml a bit, unfortunately, there is no documented way
429 to have optional dependencies with CPAN. doh :( 762 to have optional dependencies with CPAN. doh :(
430 - avoid running some tests on windows because they would fail due to 763 - avoid running some tests on windows because they would fail due to
431 perl bug (broken fork, broken pipes...). 764 perl bug (broken fork, broken pipes...).
432 - work around perl on windows bugs where perl returns undocumented 765 - work around perl on windows bugs where perl returns undocumented
433 error codes for sysread, syswrite etc. by taking advantage 766 error codes for sysread, syswrite etc. by taking advantage
… …
437 - use unix domain sockets in testsuite to work around 770 - use unix domain sockets in testsuite to work around
438 common perl implementation bugs on widows (they are emulated by 771 common perl implementation bugs on widows (they are emulated by
439 tcp sockets on windows. ugh.) 772 tcp sockets on windows. ugh.)
440 773
4414.71 Sat May 24 20:01:27 CEST 2008 7744.71 Sat May 24 20:01:27 CEST 2008
442 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 775 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
443 - take advantage of async name resolution of AnyEvent::Util. 776 - take advantage of async name resolution of AnyEvent::Util.
444 - work around brutal inet_aton override in Coro::LWP. 777 - work around brutal inet_aton override in Coro::LWP.
445 - take advantage of the readyhook in Coro::EV, for smoother 778 - take advantage of the readyhook in Coro::EV, for smoother
446 scheduling. 779 scheduling.
447 780
4484.7 Sun May 11 00:32:19 CEST 2008 7814.7 Sun May 11 00:32:19 CEST 2008
449 - completely reworked the Coro manpage. 782 - completely reworked the Coro manpage.
450 - added Coro::AnyEvent, generic event loop integration. 783 - added Coro::AnyEvent, generic event loop integration.
451 - implement cancel, ready and kill commands in Coro::Debug. 784 - implement cancel, ready and kill commands in Coro::Debug.
452 - document find_coro in Coro::Debug. 785 - document find_coro in Coro::Debug.
453 - incompatible change: rename has_stack to has_cctx. 786 - incompatible change: rename has_stack to has_cctx.
454 - Coro::AIO and Coro::BDB no longer force event model detection, 787 - Coro::AIO and Coro::BDB no longer force event model detection,
455 use AnyEvent::AIO and AnyEvent::BDB. 788 use AnyEvent::AIO and AnyEvent::BDB.
… …
579 912
5804.11 Thu Oct 11 02:40:24 CEST 2007 9134.11 Thu Oct 11 02:40:24 CEST 2007
581 - port to threaded perls. 914 - port to threaded perls.
582 915
5834.1 Thu Oct 11 02:38:16 CEST 2007 9164.1 Thu Oct 11 02:38:16 CEST 2007
584 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 917 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
585 be local to each coro (see Coro::State). 918 be local to each coro (see Coro::State).
586 - incompatible change: for very deep reasons, cede and cede_notself 919 - incompatible change: for very deep reasons, cede and cede_notself
587 cannot return anything, so nothing will be returned. 920 cannot return anything, so nothing will be returned.
588 - possibly bring back 5.10 compatibility (untested). 921 - possibly bring back 5.10 compatibility (untested).
589 - work around stupid (and wrong) warning on 5.10 :(. 922 - work around stupid (and wrong) warning on 5.10 :(.
… …
661 but stateful semantics. 994 but stateful semantics.
662 - fixed a lot of typos in Coro.pm (patch submitted by David 995 - fixed a lot of typos in Coro.pm (patch submitted by David
663 Steinbrunner, which applied flawlessly). 996 Steinbrunner, which applied flawlessly).
664 997
6653.6 Sat Apr 14 17:13:31 CEST 2007 9983.6 Sat Apr 14 17:13:31 CEST 2007
666 - added some bugfixes to get eg/myhttpd working again. 999 - added some bugfixes to get eg/myhttpd working again.
667 - added Coro::Storable for often-cede'ing freeze/thaw. 1000 - added Coro::Storable for often-cede'ing freeze/thaw.
668 - try to do a clean exit when a coroutine calls exit 1001 - try to do a clean exit when a coroutine calls exit
669 (EXPERIMENTAL). 1002 (EXPERIMENTAL).
670 - got rid of indirect call through _coro_init. 1003 - got rid of indirect call through _coro_init.
671 - updated the partly antique examples in eg/ to 1004 - updated the partly antique examples in eg/ to

Diff Legend

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