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

Comparing Coro/Changes (file contents):
Revision 1.548 by root, Sun Feb 13 22:06:18 2011 UTC vs.
Revision 1.603 by root, Wed Dec 5 01:09:12 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: unready_all
5TODO: myhttpd header parsing
6TODO: channel->maxsize(newsize)? 4TODO: channel->maxsize(newsize)?
7TODO: http://www.microsoft.com/msj/archive/s2ce.aspx 5TODO: http://www.microsoft.com/msj/archive/s2ce.aspx
6TODO: __GCC_HAVE_DWARF2_CFI_ASM
8 7
8TODO: add peerhost/peerport from IO::Socket::INET etc.
9
10 - implement 5.17 compatibility by almost blindly applying a
11 good-looking patch by Father Chrysostomos.
12 - use experimental fiber implementation on native windows
13 perls.
14 - support magic values as timed_io_once args.
15 - recommend AnyEvent 7+ or EV 4+, also require EV
16 version 4 or newer for Coro::EV.
17
186.10 Tue Oct 9 01:14:27 CEST 2012
19 - updated ecb.h, it had a typo that caused it to not compile on many
20 big endian systems (reported by many people).
21 - disable memory fences in ecb.h to improve portability.
22
236.09 Sat Oct 6 23:25:02 CEST 2012
24 - Coro::EV I/O watchers were not interruptible by exceptions
25 (Coro::State::throw) (testcase by sten).
26 - ->throw now puts threads into the ready queue, as this seems to
27 be expected by existing code, and code that doesn't cope with spurious
28 wakeups needs fixing anyway.
29 - use fd -1 in mmap.
30 - cast I32 to int in error message printf.
31 - warn about broken so-called "hardened" kernels.
32
336.08 Fri Apr 13 12:05:47 CEST 2012
34 - be more aggressive about exiting like perl does - formerly,
35 exiting from the non-main thread would not execute END blocks.
36
376.07 Fri Nov 11 21:21:48 CET 2011
38 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
39 - update ecb.h.
40
416.06 Mon Aug 8 23:59:48 CEST 2011
42 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
43 backend, which is an order of magnitude slower. unfortunately, cygwins
44 pthread implementation isn't very complete either, so allocate the stack
45 twice just to be sure.
46 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
47 in its ithreaded perl - best recompile cygwin and use the 'w'indows
48 backend for much better performance. also disable ithreads for
49 even better performance...).
50
516.05 Thu Aug 4 21:36:36 CEST 2011
52 - blush, condvar values would not be propagated from send to recv anymore
53 (reported by Chip Salzenberg).
54 - use exponential increase for the readline buffer length in
55 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
56
576.04 Wed Aug 3 17:15:45 CEST 2011
58 - use even more efficient and more compatible condvars for
59 compatibility to AnyEvent 6.x :)
60 - more inconsequential ecb.h updates.
61
626.03 Wed Aug 3 11:41:30 CEST 2011
63 - change how Coro patches AnyEvent condvars for compatibility to
64 AnyEvent 6.x.
65 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
66 hasn't been included.
67
686.02 Wed Jul 13 04:35:19 CEST 2011
69 - "improve portability to Gentoo" - gentoo manages to put perl variables
70 in memory areas that are farther than 2gb apart, which the jit couldn't
71 handle and barfed. now it's just a bit slower on gentoo and similar
72 systems.
73
746.01 Sun Jul 3 12:31:14 CEST 2011
75 - workarounds are good, but the test for whether pthreads are used
76 was not good. thisone should be better.
77 - check differently whether gcc generates cfi instructions itself.
78
796.0 Wed Jun 29 19:43:35 CEST 2011
80 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
81 destroyed and cleaned up automatically (e.g. async { schedule }).
82 - implement a JIT compiler for part of the thread switch code,
83 which gives a 50% speed improvement on threaded perls, and
84 about 4% on non-thraeded perls (so threaded perls now finally
85 reach about half the speed of non-threaded perls).
86 - slightly modernise Coro::Intro, add section about rouse functions.
87 - avoid DEFSV and ERRSV, giving another 10% improvement
88 in thread switching.
89 - Coro::State->is_destroyed is now called is_zombie.
90 - implement a Coro->safe_cancel method that might fail, but
91 cancels in a "safer" way if it succeeds.
92 - add preliminary support for DEBUGGING perls.
93 - get rid of two hash-accesses when initialising a new Coro - this
94 speeds up coro creation by almost a factor of two.
95 - croak when a coro that is being cancelled tries to block
96 (e.g. while executing a guard block), instead of crashing or
97 deadlocking.
98 - use a more robust and also faster method to identify Coro::State
99 objects - speeds up everything a bit.
100 - implement Coro->cancel in XS for a 20% speed improvement, and to
101 be able to implement mutual cancellation.
102 - speed up context switches by a percent or two by more efficiently
103 allocating context stack entries.
104 - implement Coro->join and Coro->on_destroy in XS for a speedup and
105 a reduction in memory use.
106 - cancelling a coro while it itself is cancelling another coro is
107 now supported and working, instead of triggering an assertion.
108 - be a bit more crash-resistant when calling (buggy) on_destroy
109 callbacks (best effort).
110 - move on_destroy into the slf_frame, to allow extension slf
111 functions to have destructors.
112 - get rid if coro refcounting - simply crash in other interpreter
113 threads by nulling the pointers on clone.
114 - simplify warn/die hook handling when loading Coro - the convoluted
115 logic seems to be no longer neccessary.
116 - use libecb instead of our own home-grown gcc hacks.
117 - document alternatives to Coro::LWP. Please use them :)
118 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
119 sigaltstack. Really. wine suffers from it, erlang suffers from it,
120 and it's known since at least 2006.
121
1225.372 Wed Feb 23 06:14:30 CET 2011
123 - apparently mingw doesn't provide a working gettimeofday, try to
124 work around that by relying on Time::HiRes (indirectly brought to
125 my attention by Max Maischein).
126 - fix some portability issues when Time::HiRes was used.
127
1285.371 Mon Feb 21 14:36:08 CET 2011
129 - backport to windows process emulation code again.
130
1315.37 Sat Feb 19 07:49:44 CET 2011
132 - add a big "Coro thread life cycle" section to "man Coro".
9 - try a tentative workaround against the breakage that 5.13 has 133 - try a tentative workaround against the breakage that 5.13 has
10 introduced without depreciation period. sigh. 134 introduced without depreciation period. sigh.
135 - no longer use Time::HiRes if gettimeofday is available, which
136 saves quite a lot of memory.
11 137
125.36 Sun Feb 13 05:33:41 CET 2011 1385.36 Sun Feb 13 05:33:41 CET 2011
13 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore, 139 - automatically load Coro::Channel, Coro::RWLock, Coro::Semaphore,
14 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new" 140 Coro::SemaphoreSet, Coro::Signal and Coro::Specific on first "new"
15 method call. 141 method call.

Diff Legend

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