| 1 | Revision history for Perl extension Coro. |
1 | Revision history for Perl extension Coro. |
| 2 | |
2 | |
| 3 | TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO. |
3 | TODO: should explore PerlIO::coroaio (perl leaks like hell). |
| 4 | TODO: maybe implement a default message channel, very much like Erlang's |
4 | TODO: maybe implement a default message channel, very much like Erlang's |
| 5 | actor model (which is cool in a lot of important aspects (failures!), |
5 | actor model (which is cool in a lot of important aspects (failures!), |
| 6 | but very lacking in others (higher level ipc)). |
6 | but very lacking in others (higher level ipc)). |
|
|
7 | |
|
|
8 | 5.0 |
|
|
9 | - NEW ARCHITECTURE: use 4.912 for the stable version. |
|
|
10 | - bump API version to 7 - all dependents must be recompiled. |
|
|
11 | - directly patch the entersub opcode calling ssl functions (cede, |
|
|
12 | transfer and so on). this does speed up context switching by |
|
|
13 | up to 10%. More importanly, it frees us from the hardcoded |
|
|
14 | behaviour of entersub, so we might actually be able to return |
|
|
15 | something from those functions. |
|
|
16 | - take advanatge of __builtin_frame_address on gcc. |
|
|
17 | - expose THX in coroapi (not sure whether this was a wise decision). |
|
|
18 | |
|
|
19 | 4.912 Thu Nov 13 18:31:23 CET 2008 |
|
|
20 | - minor cleanups. |
|
|
21 | - use much larger stacks on linux and perl < 5.8.8. |
|
|
22 | - Coro::Debug::new_unix_server did not unlink the socket |
|
|
23 | when destroyed. |
|
|
24 | |
|
|
25 | 4.911 Tue Nov 11 04:26:17 CET 2008 |
|
|
26 | - "port" to threaded perls. |
|
|
27 | |
|
|
28 | 4.91 Mon Nov 10 05:36:38 CET 2008 |
|
|
29 | - the ->throw exception object no longer leaks. |
|
|
30 | - creating a new cctx leaked a scopestack entry (memleak). |
|
|
31 | - new coroutines didn't get created with a zero flags field |
|
|
32 | (unknown impact). |
|
|
33 | - calling ->throw on a not-yet-started coroutine should now work |
|
|
34 | instead of being ignored. |
|
|
35 | - ->throw is now supported on Coro::State objects. |
|
|
36 | - clean up cctx creation code a bit. |
|
|
37 | - entersub is actually a UNOP, not a LOGOP (not a bugfix). |
|
|
38 | |
|
|
39 | 4.9 Sat Nov 8 17:45:27 CET 2008 |
|
|
40 | - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). |
|
|
41 | - (libcoro) no longer leak threads in the experimental pthread backend, |
|
|
42 | also speed it up considerably. |
|
|
43 | - (libcoro) do not rely on makecontext passing void *'s unscathed. |
|
|
44 | - fix compiletime dependencies on libcoro in the Makefile. |
|
|
45 | - cctx_count wasn't always updated properly. |
|
|
46 | - Coro::State::cctx_stacksize wasn't applied correctly. |
|
|
47 | - new function Coro::State::cctx_max_idle. |
|
|
48 | - the default max number of idle C contexts is now 4. |
|
|
49 | - (libcoro) try harder to get _setjmp/_longjmp. |
|
|
50 | - (libcoro) cleanup and extend the libcoro API to officially |
|
|
51 | allow the creation of empty source contexts. |
|
|
52 | - very experimental workaround for the totally broken netbsd platform. |
|
|
53 | - tried to hack around openbsd bugs. |
|
|
54 | |
|
|
55 | 4.804 Wed Nov 5 16:36:00 CET 2008 |
|
|
56 | - Coro::Debug::new_unix_server would not create a non-blocking listening |
|
|
57 | socket, sometimes causing freezes. |
|
|
58 | - (libcoro) fix misaligned stack points for setjmp and assembly |
|
|
59 | methods, which can cause crashes on x86/x86_64 with a sufficiently |
|
|
60 | aggressive compiler. |
|
|
61 | - new function: Coro::Debug::new_tcp_server. |
|
|
62 | - move ->throw into the Coro class because it only works on coro objects. |
|
|
63 | |
|
|
64 | 4.803 Mon Nov 3 17:16:12 CET 2008 |
|
|
65 | - (libcoro) use a global asm statement to become independent of gcc |
|
|
66 | otpimisations for CORO_ASM (thanks to pippijn for the idea). |
|
|
67 | - try to workaround yet another broken bsd, this time dragonfly. |
|
|
68 | |
|
|
69 | 4.802 Thu Oct 30 10:56:12 CET 2008 |
|
|
70 | - support -fno-omit-frame-pointer on x86 with the assembly method. |
|
|
71 | - tune 01_unblock.t tests a bit. |
|
|
72 | |
|
|
73 | 4.801 Wed Oct 22 18:33:37 CEST 2008 |
|
|
74 | - improve readline speed for very long "lines". |
|
|
75 | - backport to 5.8.8. |
|
|
76 | |
|
|
77 | 4.8 Thu Oct 2 13:34:40 CEST 2008 |
|
|
78 | - new function Coro::AIO::aio_wait. |
|
|
79 | - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent". |
|
|
80 | - greatly speed up and reduce memory usage of Coro::AIO requests. |
|
|
81 | - implement some other ยต-optimisations. |
|
|
82 | |
|
|
83 | 4.749 Mon Sep 29 14:40:12 CEST 2008 |
|
|
84 | - port to slow and broken pseudo-threaded perls. (courtesy pippijn). |
|
|
85 | |
|
|
86 | 4.748 Sat Sep 27 14:03:19 CEST 2008 |
|
|
87 | - implement, but do not document, PerlIO::cede(granularity). |
|
|
88 | - Coro::Storable forgot to wrap Storable::pstore. |
|
|
89 | - work around the multitude of leaks and memory corruption |
|
|
90 | bugs in PerlIO::via by using our own C-level perliol. As a side |
|
|
91 | effect, Coro::Storable is now much, much, much faster. |
|
|
92 | |
|
|
93 | 4.747 Tue Sep 23 01:59:41 CEST 2008 |
|
|
94 | - fix a per-cv memleak (one empty array was leaked per |
|
|
95 | code reference). |
|
|
96 | - avoid a crash in coro->call|rss when the coroutine was already |
|
|
97 | destroyed (most noticably when using Coro::Debug::ps :) |
|
|
98 | - also protect *Storable::FILE. |
|
|
99 | - push up default storable granularity to 20ms. |
|
|
100 | |
|
|
101 | 4.746 Sun Sep 21 03:22:20 CEST 2008 |
|
|
102 | - be more insistent on locking Storable against reentrancy |
|
|
103 | in Coro::Storable. |
|
|
104 | - move swap_def?v and throw to Coro::State, as documented. |
|
|
105 | |
|
|
106 | 4.745 Thu Jul 24 00:14:38 CEST 2008 |
|
|
107 | - remove debugging code related to MgPV_nolen_const, also try to |
|
|
108 | make it compile with perl 5.8.6 (yes, apple apparently loves |
|
|
109 | outdated software). Reported by John S. |
|
|
110 | |
|
|
111 | 4.744 Tue Jul 8 22:06:35 CEST 2008 |
|
|
112 | - correctly provide default DIE/WARN handlers as documented. |
|
|
113 | - also overwrite PL_vtbl_sigelem.svt_clear, even though current |
|
|
114 | implementations inside perl work fine for us. |
|
|
115 | |
|
|
116 | 4.743 Mon Jun 16 00:21:57 CEST 2008 |
|
|
117 | - when using Coro::EV without running EV::loop it could |
|
|
118 | result in busy-waiting for events, this has been fixed. |
|
|
119 | - reduce codesize and improve performance by using EV_DEFAULT_UC. |
|
|
120 | |
|
|
121 | 4.742 Sat May 31 14:10:21 CEST 2008 |
|
|
122 | - implement a workaround for (some) perl <5.8.8 versions. |
|
|
123 | - require EV 3.3+. |
| 7 | |
124 | |
| 8 | 4.741 Fri May 30 23:33:09 CEST 2008 |
125 | 4.741 Fri May 30 23:33:09 CEST 2008 |
| 9 | - tell netbsd how utterly broken their imitation of an OS is |
126 | - tell netbsd how utterly broken their imitation of an OS is |
| 10 | and refuse to build by default if pthreads are in use. |
127 | and refuse to build by default if pthreads are in use. |
| 11 | - switch to "s" method on all bsd's by default, as their ucontext |
128 | - switch to "s" method on all bsd's by default, as their ucontext |
