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

Comparing Coro/Changes (file contents):
Revision 1.559 by root, Fri May 6 21:15:17 2011 UTC vs.
Revision 1.589 by root, Sun Feb 19 11:14:45 2012 UTC

… …
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: http://www.microsoft.com/msj/archive/s2ce.aspx
6 6
7 - croak when a coro being cancelled tries to block (e.g. while 7 - be more aggressive about exiting like perl does - formerly,
8 executing a guard block). 8 exiting from the non-main thread would not execute END blocks.
9
106.07 Fri Nov 11 21:21:48 CET 2011
11 - work around a bug in PerlIO (setting $SIG{__WARN__} to a PVCV).
12 - update ecb.h.
13
146.06 Mon Aug 8 23:59:48 CEST 2011
15 - cygwin unfortunately patches the stack at runtime, so we use the pthreads
16 backend, which is an order of magnitude slower. unfortunately, cygwins
17 pthread implementation isn't very complete either, so allocate the stack
18 twice just to be sure.
19 (note: cygwin also enables mymalloc, which is NOT THREADSAFE ON WINDOWS,
20 in its ithreaded perl - best recompile cygwin and use the 'w'indows
21 backend for much better performance. also disable ithreads for
22 even better performance...).
23
246.05 Thu Aug 4 21:36:36 CEST 2011
25 - blush, condvar values would not be propagated from send to recv anymore
26 (reported by Chip Salzenberg).
27 - use exponential increase for the readline buffer length in
28 Coro::Handle. also reduce initial allocation to 1020 from 4096 bytes.
29
306.04 Wed Aug 3 17:15:45 CEST 2011
31 - use even more efficient and more compatible condvars for
32 compatibility to AnyEvent 6.x :)
33 - more inconsequential ecb.h updates.
34
356.03 Wed Aug 3 11:41:30 CEST 2011
36 - change how Coro patches AnyEvent condvars for compatibility to
37 AnyEvent 6.x.
38 - update ecb.h, to no longer include <pthread.h> in case WinNT.h
39 hasn't been included.
40
416.02 Wed Jul 13 04:35:19 CEST 2011
42 - "improve portability to Gentoo" - gentoo manages to put perl variables
43 in memory areas that are farther than 2gb apart, which the jit couldn't
44 handle and barfed. now it's just a bit slower on gentoo and similar
45 systems.
46
476.01 Sun Jul 3 12:31:14 CEST 2011
48 - workarounds are good, but the test for whether pthreads are used
49 was not good. thisone should be better.
50 - check differently whether gcc generates cfi instructions itself.
51
526.0 Wed Jun 29 19:43:35 CEST 2011
53 - INCOMPATIBLE CHANGE: unreferenced coro objects will now be
54 destroyed and cleaned up automatically (e.g. async { schedule }).
55 - implement a JIT compiler for part of the thread switch code,
56 which gives a 50% speed improvement on threaded perls, and
57 about 4% on non-thraeded perls (so threaded perls now finally
58 reach about half the speed of non-threaded perls).
59 - slightly modernise Coro::Intro, add section about rouse functions.
60 - avoid DEFSV and ERRSV, giving another 10% improvement
61 in thread switching.
62 - Coro::State->is_destroyed is now called is_zombie.
9 - implement a Coro::State->safe_cancel method that might fail, but 63 - implement a Coro->safe_cancel method that might fail, but
10 cancels in a "safer" way. 64 cancels in a "safer" way if it succeeds.
65 - add preliminary support for DEBUGGING perls.
66 - get rid of two hash-accesses when initialising a new Coro - this
67 speeds up coro creation by almost a factor of two.
68 - croak when a coro that is being cancelled tries to block
69 (e.g. while executing a guard block), instead of crashing or
70 deadlocking.
71 - use a more robust and also faster method to identify Coro::State
72 objects - speeds up everything a bit.
11 - implement Coro->cancel in XS for a 20% speed improvement, and to 73 - implement Coro->cancel in XS for a 20% speed improvement, and to
12 be able to implement mutual cancellation. 74 be able to implement mutual cancellation.
75 - speed up context switches by a percent or two by more efficiently
76 allocating context stack entries.
77 - implement Coro->join and Coro->on_destroy in XS for a speedup and
78 a reduction in memory use.
13 - cancelling a coro while it itself is cancelling another coro is 79 - cancelling a coro while it itself is cancelling another coro is
14 now supported and working, instead of triggering an assertion. 80 now supported and working, instead of triggering an assertion.
15 - be a bit more crash-resistant when calling (buggy) on_destroy 81 - be a bit more crash-resistant when calling (buggy) on_destroy
16 callbacks (best effort). 82 callbacks (best effort).
17 - move on_destroy into the slf_frame, to allow extension slf 83 - move on_destroy into the slf_frame, to allow extension slf
18 functions to have destructors. 84 functions to have destructors.
85 - get rid if coro refcounting - simply crash in other interpreter
86 threads by nulling the pointers on clone.
87 - simplify warn/die hook handling when loading Coro - the convoluted
88 logic seems to be no longer neccessary.
89 - use libecb instead of our own home-grown gcc hacks.
90 - document alternatives to Coro::LWP. Please use them :)
91 - work around another mindless idiotic NEEDLESS bug in openbsd/mirbsds
92 sigaltstack. Really. wine suffers from it, erlang suffers from it,
93 and it's known since at least 2006.
19 94
205.372 Wed Feb 23 06:14:30 CET 2011 955.372 Wed Feb 23 06:14:30 CET 2011
21 - apparently mingw doesn't provide a working gettimeofday, try to 96 - apparently mingw doesn't provide a working gettimeofday, try to
22 work around that by relying on Time::HiRes (indirectly brought to 97 work around that by relying on Time::HiRes (indirectly brought to
23 my attention by Max Maischein). 98 my attention by Max Maischein).

Diff Legend

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