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

Comparing Coro/Changes (file contents):
Revision 1.555 by root, Fri Apr 29 15:43:26 2011 UTC vs.
Revision 1.610 by root, Wed Dec 19 23:36:16 2012 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: cancel calling another coro that cancels ourselves => crash (CF_DELAYCANCEL?) -> canceltest
7 6
7 - improved Coro::State documentation a bit.
8 - Coro::Debug::command now flushes the output.
9 - add hack detection code for x32 abi, because the braindead slugs
10 who designed that made it look exactly like x86_64 without
11 providing proper compile time symbols to test for it. as a result,
12 this detection cannot work reliably.
13
146.23 Fri Dec 7 23:36:37 CET 2012
15 - use experimental fiber implementation on native windows
16 perls.
17 - use sizeof (void *) as multiplication factor for stack sizes,
18 to accomodate the totally braindamaged microsoft 64 bit "os".
19 - changed verifier host from win2k-ap510-32 to win7-sp516-32/64.
20 activeperl 5.16 crashes when PerlIO_define_layer is called due
21 to some bug in the perl dll, strawberry perl at least passes
22 the testsuite.
23 - implement Coro::Handle->peeraddr/host/port, for slightly
24 improved compatibility with LWP.
25 - implement 5.17 compatibility by almost blindly applying a
26 good-looking patch by Father Chrysostomos.
27 - move stack management functions into libcoro 3.
28 - libcoro version 3 "released".
29 - support magic values as timed_io_once args.
30 - recommend AnyEvent 7+ or EV 4+, also require EV
31 version 4 or newer for Coro::EV.
32
336.10 Tue Oct 9 01:14:27 CEST 2012
34 - updated ecb.h, it had a typo that caused it to not compile on many
35 big endian systems (reported by many people).
36 - disable memory fences in ecb.h to improve portability.
37
386.09 Sat Oct 6 23:25:02 CEST 2012
39 - Coro::EV I/O watchers were not interruptible by exceptions
40 (Coro::State::throw) (testcase by sten).
41 - ->throw now puts threads into the ready queue, as this seems to
42 be expected by existing code, and code that doesn't cope with spurious
43 wakeups needs fixing anyway.
44 - use fd -1 in mmap.
45 - cast I32 to int in error message printf.
46 - warn about broken so-called "hardened" kernels.
47
486.08 Fri Apr 13 12:05:47 CEST 2012
49 - be more aggressive about exiting like perl does - formerly,
50 exiting from the non-main thread would not execute END blocks.
51
526.07 Fri Nov 11 21:21:48 CET 2011
53 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
54 - update ecb.h.
55
566.06 Mon Aug 8 23:59:48 CEST 2011
57 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
58 backend, which is an order of magnitude slower. unfortunately, cygwins
59 pthread implementation isn't very complete either, so allocate the stack
60 twice just to be sure.
61 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
62 in its ithreaded perl - best recompile cygwin and use the 'w'indows
63 backend for much better performance. also disable ithreads for
64 even better performance...).
65
666.05 Thu Aug 4 21:36:36 CEST 2011
67 - blush, condvar values would not be propagated from send to recv anymore
68 (reported by Chip Salzenberg).
69 - use exponential increase for the readline buffer length in
70 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
71
726.04 Wed Aug 3 17:15:45 CEST 2011
73 - use even more efficient and more compatible condvars for
74 compatibility to AnyEvent 6.x :)
75 - more inconsequential ecb.h updates.
76
776.03 Wed Aug 3 11:41:30 CEST 2011
78 - change how Coro patches AnyEvent condvars for compatibility to
79 AnyEvent 6.x.
80 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
81 hasn't been included.
82
836.02 Wed Jul 13 04:35:19 CEST 2011
84 - "improve portability to Gentoo" - gentoo manages to put perl variables
85 in memory areas that are farther than 2gb apart, which the jit couldn't
86 handle and barfed. now it's just a bit slower on gentoo and similar
87 systems.
88
896.01 Sun Jul 3 12:31:14 CEST 2011
90 - workarounds are good, but the test for whether pthreads are used
91 was not good. thisone should be better.
92 - check differently whether gcc generates cfi instructions itself.
93
946.0 Wed Jun 29 19:43:35 CEST 2011
95 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
96 destroyed and cleaned up automatically (e.g. async { schedule }).
97 - implement a JIT compiler for part of the thread switch code,
98 which gives a 50% speed improvement on threaded perls, and
99 about 4% on non-thraeded perls (so threaded perls now finally
100 reach about half the speed of non-threaded perls).
101 - slightly modernise Coro::Intro, add section about rouse functions.
102 - avoid DEFSV and ERRSV, giving another 10% improvement
103 in thread switching.
104 - Coro::State->is_destroyed is now called is_zombie.
105 - implement a Coro->safe_cancel method that might fail, but
106 cancels in a "safer" way if it succeeds.
107 - add preliminary support for DEBUGGING perls.
108 - get rid of two hash-accesses when initialising a new Coro - this
109 speeds up coro creation by almost a factor of two.
110 - croak when a coro that is being cancelled tries to block
111 (e.g. while executing a guard block), instead of crashing or
112 deadlocking.
113 - use a more robust and also faster method to identify Coro::State
114 objects - speeds up everything a bit.
8 - implement Coro->cancel in XS for a 20% speed improvement, and to 115 - implement Coro->cancel in XS for a 20% speed improvement, and to
9 be able to implement mutual cancellation. 116 be able to implement mutual cancellation.
117 - speed up context switches by a percent or two by more efficiently
118 allocating context stack entries.
119 - implement Coro->join and Coro->on_destroy in XS for a speedup and
120 a reduction in memory use.
121 - cancelling a coro while it itself is cancelling another coro is
122 now supported and working, instead of triggering an assertion.
10 - be a bit more crash-resistant when calling (buggy) on_destroy 123 - be a bit more crash-resistant when calling (buggy) on_destroy
11 callbacks (best effort). 124 callbacks (best effort).
125 - move on_destroy into the slf_frame, to allow extension slf
126 functions to have destructors.
127 - get rid if coro refcounting - simply crash in other interpreter
128 threads by nulling the pointers on clone.
129 - simplify warn/die hook handling when loading Coro - the convoluted
130 logic seems to be no longer neccessary.
131 - use libecb instead of our own home-grown gcc hacks.
132 - document alternatives to Coro::LWP. Please use them :)
133 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
134 sigaltstack. Really. wine suffers from it, erlang suffers from it,
135 and it's known since at least 2006.
12 136
135.372 Wed Feb 23 06:14:30 CET 2011 1375.372 Wed Feb 23 06:14:30 CET 2011
14 - apparently mingw doesn't provide a working gettimeofday, try to 138 - apparently mingw doesn't provide a working gettimeofday, try to
15 work around that by relying on Time::HiRes (indirectly brought to 139 work around that by relying on Time::HiRes (indirectly brought to
16 my attention by Max Maischein). 140 my attention by Max Maischein).

Diff Legend

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