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

Comparing Coro/Changes (file contents):
Revision 1.515 by root, Tue Jul 21 03:39:58 2009 UTC vs.
Revision 1.624 by root, Sat Feb 22 14:12:45 2014 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: maybe implement a default message channel, very much like Erlang's
5 actor model (which is cool in a lot of important aspects (failures!),
6 but very lacking in others (higher level ipc)).
7TODO: unready_all
8TODO: myhttpd header parsing
9TODO: channel->maxsize(newsize)? 4TODO: channel->maxsize(newsize)?
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
10 8
9TODO: warnhook reports from wrong line number now, when arg is Math::GMP number.
10 - FATAL: $Coro::IDLE blocked itself will now use Carp::confess
11 to report a full stacktrace, which should help find out
12 where the actual call is.
13 - disable ecb.h libm dependency, to compile on ancient systems
14 or under adverse conditions.
15
166.33 Mon Nov 18 11:26:27 CET 2013
17 - do not crash when freeing padlists with holes (in 5.18).
18 - tentative SVt_BIND 5.19 port/fix.
19
206.32 Tue Nov 5 15:35:35 CET 2013
21 - use a new algorithm to derive padlists for perl 5.18. The old
22 one could lead to 0-pointer accesses inside perl (reported
23 by Darin McBride).
24
256.31 Thu May 9 07:39:48 CEST 2013
26 - Coro::AIO requests would crash if the thread was ready'd
27 while the request was ongoing.
28
296.29 Wed May 8 02:55:18 CEST 2013
30 - when an on_destroy handler destructs the coro currently being
31 destructed a perl scalar could be accessed after being freed,
32 likely causing a crash.
33
346.28 Wed Mar 6 06:58:02 CET 2013
35 - clean remnants of existing __DIE__ and __WARN__ handlers so
36 they lose their magic and will not cause segfaults later
37 (testcase by Andrey Sagulin).
38 - improved Coro::State documentation a bit.
39 - Coro::Debug::command now flushes the output.
40 - add hack detection code for x32 abi, because the braindead slugs
41 who designed that made it look exactly like x86_64 without
42 providing proper compile time symbols to test for it. as a result,
43 this detection cannot work reliably.
44 - valgrind stack registering was broken.
45 - do not rely on Time::HiRes anymore in Coro::Debug.
46
476.23 Fri Dec 7 23:36:37 CET 2012
48 - use experimental fiber implementation on native windows
49 perls.
50 - use sizeof (void *) as multiplication factor for stack sizes,
51 to accomodate the totally braindamaged microsoft 64 bit "os".
52 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
53 activeperl 5.16 crashes when PerlIO_define_layer is called due
54 to some bug in the perl dll, strawberry perl at least passes
55 the testsuite.
56 - implement Coro::Handle->peeraddr/host/port, for slightly
57 improved compatibility with LWP.
58 - implement 5.17 compatibility by almost blindly applying a
59 good-looking patch by Father Chrysostomos.
60 - move stack management functions into libcoro 3.
61 - libcoro version 3 "released".
62 - support magic values as timed_io_once args.
63 - recommend AnyEvent 7+ or EV 4+, also require EV
64 version 4 or newer for Coro::EV.
65
666.10 Tue Oct 9 01:14:27 CEST 2012
67 - updated ecb.h, it had a typo that caused it to not compile on many
68 big endian systems (reported by many people).
69 - disable memory fences in ecb.h to improve portability.
70
716.09 Sat Oct 6 23:25:02 CEST 2012
72 - Coro::EV I/O watchers were not interruptible by exceptions
73 (Coro::State::throw) (testcase by sten).
74 - ->throw now puts threads into the ready queue, as this seems to
75 be expected by existing code, and code that doesn't cope with spurious
76 wakeups needs fixing anyway.
77 - use fd -1 in mmap.
78 - cast I32 to int in error message printf.
79 - warn about broken so-called "hardened" kernels.
80
816.08 Fri Apr 13 12:05:47 CEST 2012
82 - be more aggressive about exiting like perl does - formerly,
83 exiting from the non-main thread would not execute END blocks.
84
856.07 Fri Nov 11 21:21:48 CET 2011
86 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
87 - update ecb.h.
88
896.06 Mon Aug 8 23:59:48 CEST 2011
90 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
91 backend, which is an order of magnitude slower. unfortunately, cygwins
92 pthread implementation isn't very complete either, so allocate the stack
93 twice just to be sure.
94 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
95 in its ithreaded perl - best recompile cygwin and use the 'w'indows
96 backend for much better performance. also disable ithreads for
97 even better performance...).
98
996.05 Thu Aug 4 21:36:36 CEST 2011
100 - blush, condvar values would not be propagated from send to recv anymore
101 (reported by Chip Salzenberg).
102 - use exponential increase for the readline buffer length in
103 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
104
1056.04 Wed Aug 3 17:15:45 CEST 2011
106 - use even more efficient and more compatible condvars for
107 compatibility to AnyEvent 6.x :)
108 - more inconsequential ecb.h updates.
109
1106.03 Wed Aug 3 11:41:30 CEST 2011
111 - change how Coro patches AnyEvent condvars for compatibility to
112 AnyEvent 6.x.
113 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
114 hasn't been included.
115
1166.02 Wed Jul 13 04:35:19 CEST 2011
117 - "improve portability to Gentoo" - gentoo manages to put perl variables
118 in memory areas that are farther than 2gb apart, which the jit couldn't
119 handle and barfed. now it's just a bit slower on gentoo and similar
120 systems.
121
1226.01 Sun Jul 3 12:31:14 CEST 2011
123 - workarounds are good, but the test for whether pthreads are used
124 was not good. thisone should be better.
125 - check differently whether gcc generates cfi instructions itself.
126
1276.0 Wed Jun 29 19:43:35 CEST 2011
128 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
129 destroyed and cleaned up automatically (e.g. async { schedule }).
130 - implement a JIT compiler for part of the thread switch code,
131 which gives a 50% speed improvement on threaded perls, and
132 about 4% on non-thraeded perls (so threaded perls now finally
133 reach about half the speed of non-threaded perls).
134 - slightly modernise Coro::Intro, add section about rouse functions.
135 - avoid DEFSV and ERRSV, giving another 10% improvement
136 in thread switching.
137 - Coro::State->is_destroyed is now called is_zombie.
138 - implement a Coro->safe_cancel method that might fail, but
139 cancels in a "safer" way if it succeeds.
140 - add preliminary support for DEBUGGING perls.
141 - get rid of two hash-accesses when initialising a new Coro - this
142 speeds up coro creation by almost a factor of two.
143 - croak when a coro that is being cancelled tries to block
144 (e.g. while executing a guard block), instead of crashing or
145 deadlocking.
146 - use a more robust and also faster method to identify Coro::State
147 objects - speeds up everything a bit.
148 - implement Coro->cancel in XS for a 20% speed improvement, and to
149 be able to implement mutual cancellation.
150 - speed up context switches by a percent or two by more efficiently
151 allocating context stack entries.
152 - implement Coro->join and Coro->on_destroy in XS for a speedup and
153 a reduction in memory use.
154 - cancelling a coro while it itself is cancelling another coro is
155 now supported and working, instead of triggering an assertion.
156 - be a bit more crash-resistant when calling (buggy) on_destroy
157 callbacks (best effort).
158 - move on_destroy into the slf_frame, to allow extension slf
159 functions to have destructors.
160 - get rid if coro refcounting - simply crash in other interpreter
161 threads by nulling the pointers on clone.
162 - simplify warn/die hook handling when loading Coro - the convoluted
163 logic seems to be no longer neccessary.
164 - use libecb instead of our own home-grown gcc hacks.
165 - document alternatives to Coro::LWP. Please use them :)
166 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
167 sigaltstack. Really. wine suffers from it, erlang suffers from it,
168 and it's known since at least 2006.
169
1705.372 Wed Feb 23 06:14:30 CET 2011
171 - apparently mingw doesn't provide a working gettimeofday, try to
172 work around that by relying on Time::HiRes (indirectly brought to
173 my attention by Max Maischein).
174 - fix some portability issues when Time::HiRes was used.
175
1765.371 Mon Feb 21 14:36:08 CET 2011
177 - backport to windows process emulation code again.
178
1795.37 Sat Feb 19 07:49:44 CET 2011
180 - add a big "Coro thread life cycle" section to "man Coro".
181 - try a tentative workaround against the breakage that 5.13 has
182 introduced without depreciation period. sigh.
183 - no longer use Time::HiRes if gettimeofday is available, which
184 saves quite a lot of memory.
185
1865.36 Sun Feb 13 05:33:41 CET 2011
187 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
188 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
189 method call.
190 - undocument Coro::Timer::sleep and obsolete whole module.
191 - optimise Coro::Timer::timeout memory and cpu usage.
192 - slightly updated Coro::Intro for recent changes.
193 - do not initialise PL_dirty anymore.
194
1955.25 Thu Nov 11 01:08:39 CET 2010
196 - try a different approach on netbsd - netbsd 5 finally has marginally
197 working pthreads, but still broken ucontext/sigaltstack.
198 - openbsd 4.8 finally got their act together, Coro works out of the box
199 with asm, setjmp and pthreads (no change, just informational).
200
2015.24 Sat Oct 23 11:27:12 CEST 2010
202 - port to the EV 4.0 API.
203 - work around bugs in mingw32, making strawberry perl work
204 out of the box.
205 - correctly modify Coro::AIO function prototypes
206 so that they reflect the "no optional parameters" rule.
207 - "ported" libcoro to C++.
208
2095.23 Mon May 17 18:50:42 CEST 2010
210 - be more resistant to ordering changes when initialising
211 Coro::AnyEvent, Coro::EV and Coro::Event (reported by Matthias
212 Waldorf).
213 - document that perl 5.12 deliberately removed support for cloning.
214
2155.22 Wed Apr 14 03:55:35 CEST 2010
216 - correctly return udnef on errors in Coro::Handle::read/write
217 (testcase by Marc Mims).
218 - convert Coro::Util into a "perl compatibility wrapper" - the functions
219 are less useful now, but are drop-in replacements for existing
220 functions, listing better alternatives in the documentation. This also
221 fixes a bug in Coro::LWP which naively substituted Socket::inet_aton
222 with Coro::Util::inet_aton.
223 - do not override $Coro::idle unconditionally in Coro.pm, as other
224 modules could have provided their own idle coro already
225 (for exmaple, Coro::AnyEvent).
226 - fix Coro::Util::gethost* functions.
227 - Coro::Timer corretcly exports it's symbols (reported by Hideki Yamamura).
228
2295.21 Wed Dec 16 07:19:51 CET 2009
230 - automatically load Coro::AnyEvent when AnyEvent and Coro are used
231 together.
232 - add some examples on how to combine other event loops with Coro in
233 Coro::AnyEvent, and how to run it (and not to block). Seems to be
234 the most common source of confusion.
235 - try to catch people naively blocking in an event callback.
236 - work around the perl filehandle bug issue in conjunction with
237 older common::sense (as indirectly pointed out by ZSystem).
238 - clarify the "not from signal handlers" section.
239
2405.2 Sun Oct 4 14:54:24 CEST 2009
241 - Coro::Storable destroyed the prototypes of the functions it wrapped.
242 - export rouse_cb and rouse_wait by default now.
243 - fix various prototype mismatches in Coro::AnyEvent and Coro::Handle.
244 - new method $state->swap_sv.
245 - added section on "windows process emulation" to the manpage,
246 after a not-so-fruitful (nor-friendly) "discussion" with chip
247 salzenberg (discussion implies arguments, but his only arguments
248 were ad-hominems, one wonders why he started it in the first
249 place). I hope this explains it well enough for him to understand,
250 and maybe well enough for others to understand.
251 - use common::sense everywhere now.
252 - idle callbacks are no longer supported, use idle coros instead.
253 - print a thread listing when a deadlock is detected.
254
2555.17 Sat Aug 22 23:09:31 CEST 2009
256 - work around a bug in the perl debugger causing crashes
257 when running under the debugger by marking _pool_handler as nodebug.
258 - speed up Coro::async considerably.
259 - try some hacks to get netbsd to work "more often" - their broken
260 setjmp/longjmp, ucontext *and* phtreads are really hard on Coro.
261 - convert Coro to AE 5.0 API.
262
2635.162 Tue Jul 28 04:04:03 CEST 2009
264 - perl 5.8.2 is now minimum requirement.
265 - skip t/19_handle.t on broken windows perls.
266
2675.161 Wed Jul 22 04:47:38 CEST 2009
268 - Coro::AnyEvent::poll could have a different prototype when EV was
269 used as backend (analyzed by Tatsuhiko Miyagawa).
11 - Coro::AnyEvent errornously initialised the event loop when loaded, 270 - Coro::AnyEvent errornously initialised the event loop when loaded,
12 not on demand. 271 not on demand.
272 - try to workaround rare */t/01_unblock.t failures.
13 273
145.16 Tue Jul 21 01:44:37 CEST 2009 2745.16 Tue Jul 21 01:44:37 CEST 2009
15 - Coro::AnyEvent failed to hook into the event loop 275 - Coro::AnyEvent failed to hook into the event loop
16 when no threads had been readied between detecting 276 when no threads had been readied between detecting
17 the event loop and actually running it. 277 the event loop and actually running it.
… …
231 - ->throw is now supported on Coro::State objects. 491 - ->throw is now supported on Coro::State objects.
232 - clean up cctx creation code a bit. 492 - clean up cctx creation code a bit.
233 - entersub is actually a UNOP, not a LOGOP (not a bugfix). 493 - entersub is actually a UNOP, not a LOGOP (not a bugfix).
234 494
2354.9 Sat Nov 8 17:45:27 CET 2008 4954.9 Sat Nov 8 17:45:27 CET 2008
236 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). 496 - (libcoro) did not preserve rbp with CORO_ASM (we are getting there).
237 - (libcoro) no longer leak threads in the experimental pthread backend, 497 - (libcoro) no longer leak threads in the experimental pthread backend,
238 also speed it up considerably. 498 also speed it up considerably.
239 - (libcoro) do not rely on makecontext passing void *'s unscathed. 499 - (libcoro) do not rely on makecontext passing void *'s unscathed.
240 - fix compiletime dependencies on libcoro in the Makefile. 500 - fix compiletime dependencies on libcoro in the Makefile.
241 - cctx_count wasn't always updated properly. 501 - cctx_count wasn't always updated properly.
… …
303 - remove debugging code related to MgPV_nolen_const, also try to 563 - remove debugging code related to MgPV_nolen_const, also try to
304 make it compile with perl 5.8.6 (yes, apple apparently loves 564 make it compile with perl 5.8.6 (yes, apple apparently loves
305 outdated software). Reported by John S. 565 outdated software). Reported by John S.
306 566
3074.744 Tue Jul 8 22:06:35 CEST 2008 5674.744 Tue Jul 8 22:06:35 CEST 2008
308 - correctly provide default DIE/WARN handlers as documented. 568 - correctly provide default DIE/WARN handlers as documented.
309 - also overwrite PL_vtbl_sigelem.svt_clear, even though current 569 - also overwrite PL_vtbl_sigelem.svt_clear, even though current
310 implementations inside perl work fine for us. 570 implementations inside perl work fine for us.
311 571
3124.743 Mon Jun 16 00:21:57 CEST 2008 5724.743 Mon Jun 16 00:21:57 CEST 2008
313 - when using Coro::EV without running EV::loop it could 573 - when using Coro::EV without running EV::loop it could
… …
335 - sprinkle "no warnings" freely over everything, also suppress 595 - sprinkle "no warnings" freely over everything, also suppress
336 warnings for some other modules. 596 warnings for some other modules.
337 - fix typo in WSAEWOULDBLOCK. 597 - fix typo in WSAEWOULDBLOCK.
338 598
3394.72 Sun May 25 05:14:36 CEST 2008 5994.72 Sun May 25 05:14:36 CEST 2008
340 - tweak META.yaml a bit, unfortunately, there is no documented way 600 - tweak META.yaml a bit, unfortunately, there is no documented way
341 to have optional dependencies with CPAN. doh :( 601 to have optional dependencies with CPAN. doh :(
342 - avoid running some tests on windows because they would fail due to 602 - avoid running some tests on windows because they would fail due to
343 perl bug (broken fork, broken pipes...). 603 perl bug (broken fork, broken pipes...).
344 - work around perl on windows bugs where perl returns undocumented 604 - work around perl on windows bugs where perl returns undocumented
345 error codes for sysread, syswrite etc. by taking advantage 605 error codes for sysread, syswrite etc. by taking advantage
… …
349 - use unix domain sockets in testsuite to work around 609 - use unix domain sockets in testsuite to work around
350 common perl implementation bugs on widows (they are emulated by 610 common perl implementation bugs on widows (they are emulated by
351 tcp sockets on windows. ugh.) 611 tcp sockets on windows. ugh.)
352 612
3534.71 Sat May 24 20:01:27 CEST 2008 6134.71 Sat May 24 20:01:27 CEST 2008
354 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()"). 614 - fix a bug in Coro::AnyEvent ("Usage: Coro::AnyEvent::_schedule()").
355 - take advantage of async name resolution of AnyEvent::Util. 615 - take advantage of async name resolution of AnyEvent::Util.
356 - work around brutal inet_aton override in Coro::LWP. 616 - work around brutal inet_aton override in Coro::LWP.
357 - take advantage of the readyhook in Coro::EV, for smoother 617 - take advantage of the readyhook in Coro::EV, for smoother
358 scheduling. 618 scheduling.
359 619
3604.7 Sun May 11 00:32:19 CEST 2008 6204.7 Sun May 11 00:32:19 CEST 2008
361 - completely reworked the Coro manpage. 621 - completely reworked the Coro manpage.
362 - added Coro::AnyEvent, generic event loop integration. 622 - added Coro::AnyEvent, generic event loop integration.
363 - implement cancel, ready and kill commands in Coro::Debug. 623 - implement cancel, ready and kill commands in Coro::Debug.
364 - document find_coro in Coro::Debug. 624 - document find_coro in Coro::Debug.
365 - incompatible change: rename has_stack to has_cctx. 625 - incompatible change: rename has_stack to has_cctx.
366 - Coro::AIO and Coro::BDB no longer force event model detection, 626 - Coro::AIO and Coro::BDB no longer force event model detection,
367 use AnyEvent::AIO and AnyEvent::BDB. 627 use AnyEvent::AIO and AnyEvent::BDB.
… …
491 751
4924.11 Thu Oct 11 02:40:24 CEST 2007 7524.11 Thu Oct 11 02:40:24 CEST 2007
493 - port to threaded perls. 753 - port to threaded perls.
494 754
4954.1 Thu Oct 11 02:38:16 CEST 2007 7554.1 Thu Oct 11 02:38:16 CEST 2007
496 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now 756 - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now
497 be local to each coro (see Coro::State). 757 be local to each coro (see Coro::State).
498 - incompatible change: for very deep reasons, cede and cede_notself 758 - incompatible change: for very deep reasons, cede and cede_notself
499 cannot return anything, so nothing will be returned. 759 cannot return anything, so nothing will be returned.
500 - possibly bring back 5.10 compatibility (untested). 760 - possibly bring back 5.10 compatibility (untested).
501 - work around stupid (and wrong) warning on 5.10 :(. 761 - work around stupid (and wrong) warning on 5.10 :(.
… …
573 but stateful semantics. 833 but stateful semantics.
574 - fixed a lot of typos in Coro.pm (patch submitted by David 834 - fixed a lot of typos in Coro.pm (patch submitted by David
575 Steinbrunner, which applied flawlessly). 835 Steinbrunner, which applied flawlessly).
576 836
5773.6 Sat Apr 14 17:13:31 CEST 2007 8373.6 Sat Apr 14 17:13:31 CEST 2007
578 - added some bugfixes to get eg/myhttpd working again. 838 - added some bugfixes to get eg/myhttpd working again.
579 - added Coro::Storable for often-cede'ing freeze/thaw. 839 - added Coro::Storable for often-cede'ing freeze/thaw.
580 - try to do a clean exit when a coroutine calls exit 840 - try to do a clean exit when a coroutine calls exit
581 (EXPERIMENTAL). 841 (EXPERIMENTAL).
582 - got rid of indirect call through _coro_init. 842 - got rid of indirect call through _coro_init.
583 - updated the partly antique examples in eg/ to 843 - updated the partly antique examples in eg/ to

Diff Legend

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