| 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 | |
7 | |
| 8 | 4.748 |
8 | TODO: testcancel |
|
|
9 | TODO: guard as SLF - to avoid deadlocks. |
|
|
10 | - avoid deadlock in Coro::Channel if maxsize == 1 (Richard Hundt). |
|
|
11 | |
|
|
12 | 5.0 |
|
|
13 | - NEW ARCHITECTURE: use the latest 4.x version if you experience |
|
|
14 | stability issues. |
|
|
15 | - bump API version to 7 - all dependents must be recompiled. |
|
|
16 | - Coro::Semaphore->adjust didn't correctly wake up enough waiters. |
|
|
17 | - async_pool did free a scalar value twice |
|
|
18 | ("Attempt to unreference..."). |
|
|
19 | - fix a longstanding bug where calling terminate on a coro that |
|
|
20 | was waiting for a semaphore that was just becoming available |
|
|
21 | would cause a deadlock (semaphore would get into a state where |
|
|
22 | it was available but waiters were waiting(. |
|
|
23 | - perl's process emulation is now not even theoretically supported |
|
|
24 | anymore. |
|
|
25 | - use named constants in Coro::Channel (Richard Hundt). |
|
|
26 | - directly patch the entersub opcode calling SLF functions (cede, |
|
|
27 | transfer and so on). this does speed up context switching, but |
|
|
28 | more importanly, it frees us from the hardcoded behaviour of |
|
|
29 | entersub, so we might actually be able to return something from |
|
|
30 | those functions and atcually create new ones. |
|
|
31 | - take advantage of __builtin_frame_address on gcc. |
|
|
32 | - expose THX in coroapi (not sure whether this was a wise decision, |
|
|
33 | as "threaded" perls are running at half speed anyways). |
|
|
34 | - implement execute_slf (schedule-like-function) interface that makes |
|
|
35 | it possible to implement schedule-like-functions in XS. |
|
|
36 | - use new SLF interface to massively speed up Coro::EV by roughly a |
|
|
37 | factor of two. |
|
|
38 | - removed timed_down and timed_guard functions - they were not being |
|
|
39 | used anyways and should be replaced by a more generic mechanism - |
|
|
40 | and were annoying to support anyways :) |
|
|
41 | - used new SLF interface to massively speed up Coro::Semaphore by a |
|
|
42 | factor of three. |
|
|
43 | - used new SLF interface to speed up Coro::AIO by roughly a factor of |
|
|
44 | four and reduce its memory usage considerably. |
|
|
45 | - forgot to include Coro/libcoro/README in the dist for all these years. |
|
|
46 | - work around a freebsd pthreads bug (manual testcancel is required as |
|
|
47 | pthread_cond_wait isn't a cancellation point on freebsd). |
|
|
48 | |
|
|
49 | 4.913 Sat Nov 15 07:58:28 CET 2008 |
|
|
50 | - async_pool did free a scalar value twice |
|
|
51 | ("Attempt to unreference..."). |
|
|
52 | |
|
|
53 | 4.912 Thu Nov 13 18:31:23 CET 2008 |
|
|
54 | - minor cleanups. |
|
|
55 | - use much larger stacks on linux and perl < 5.8.8. |
|
|
56 | - Coro::Debug::new_unix_server did not unlink the socket |
|
|
57 | when destroyed. |
|
|
58 | |
|
|
59 | 4.911 Tue Nov 11 04:26:17 CET 2008 |
|
|
60 | - "port" to threaded perls. |
|
|
61 | |
|
|
62 | 4.91 Mon Nov 10 05:36:38 CET 2008 |
|
|
63 | - the ->throw exception object no longer leaks. |
|
|
64 | - creating a new cctx leaked a scopestack entry (memleak). |
|
|
65 | - new coroutines didn't get created with a zero flags field |
|
|
66 | (unknown impact). |
|
|
67 | - calling ->throw on a not-yet-started coroutine should now work |
|
|
68 | instead of being ignored. |
|
|
69 | - ->throw is now supported on Coro::State objects. |
|
|
70 | - clean up cctx creation code a bit. |
|
|
71 | - entersub is actually a UNOP, not a LOGOP (not a bugfix). |
|
|
72 | |
|
|
73 | 4.9 Sat Nov 8 17:45:27 CET 2008 |
|
|
74 | - (libcoro) did not preserve rbp with CORO_ASM (we are getting there). |
|
|
75 | - (libcoro) no longer leak threads in the experimental pthread backend, |
|
|
76 | also speed it up considerably. |
|
|
77 | - (libcoro) do not rely on makecontext passing void *'s unscathed. |
|
|
78 | - fix compiletime dependencies on libcoro in the Makefile. |
|
|
79 | - cctx_count wasn't always updated properly. |
|
|
80 | - Coro::State::cctx_stacksize wasn't applied correctly. |
|
|
81 | - new function Coro::State::cctx_max_idle. |
|
|
82 | - the default max number of idle C contexts is now 4. |
|
|
83 | - (libcoro) try harder to get _setjmp/_longjmp. |
|
|
84 | - (libcoro) cleanup and extend the libcoro API to officially |
|
|
85 | allow the creation of empty source contexts. |
|
|
86 | - very experimental workaround for the totally broken netbsd platform. |
|
|
87 | - tried to hack around openbsd bugs. |
|
|
88 | |
|
|
89 | 4.804 Wed Nov 5 16:36:00 CET 2008 |
|
|
90 | - Coro::Debug::new_unix_server would not create a non-blocking listening |
|
|
91 | socket, sometimes causing freezes. |
|
|
92 | - (libcoro) fix misaligned stack points for setjmp and assembly |
|
|
93 | methods, which can cause crashes on x86/x86_64 with a sufficiently |
|
|
94 | aggressive compiler. |
|
|
95 | - new function: Coro::Debug::new_tcp_server. |
|
|
96 | - move ->throw into the Coro class because it only works on coro objects. |
|
|
97 | |
|
|
98 | 4.803 Mon Nov 3 17:16:12 CET 2008 |
|
|
99 | - (libcoro) use a global asm statement to become independent of gcc |
|
|
100 | otpimisations for CORO_ASM (thanks to pippijn for the idea). |
|
|
101 | - try to workaround yet another broken bsd, this time dragonfly. |
|
|
102 | |
|
|
103 | 4.802 Thu Oct 30 10:56:12 CET 2008 |
|
|
104 | - support -fno-omit-frame-pointer on x86 with the assembly method. |
|
|
105 | - tune 01_unblock.t tests a bit. |
|
|
106 | |
|
|
107 | 4.801 Wed Oct 22 18:33:37 CEST 2008 |
|
|
108 | - improve readline speed for very long "lines". |
|
|
109 | - backport to 5.8.8. |
|
|
110 | |
|
|
111 | 4.8 Thu Oct 2 13:34:40 CEST 2008 |
|
|
112 | - new function Coro::AIO::aio_wait. |
|
|
113 | - Coro.:AIO and Coro::BDB now "use Coro::AnyEvent". |
|
|
114 | - greatly speed up and reduce memory usage of Coro::AIO requests. |
|
|
115 | - implement some other ยต-optimisations. |
|
|
116 | |
|
|
117 | 4.749 Mon Sep 29 14:40:12 CEST 2008 |
|
|
118 | - port to slow and broken pseudo-threaded perls. (courtesy pippijn). |
|
|
119 | |
|
|
120 | 4.748 Sat Sep 27 14:03:19 CEST 2008 |
|
|
121 | - implement, but do not document, PerlIO::cede(granularity). |
| 9 | - Coro::Storable forgot to wrap Storable::pstore. |
122 | - Coro::Storable forgot to wrap Storable::pstore. |
| 10 | - work around a memory leak in perl in Coro::Storable |
123 | - work around the multitude of leaks and memory corruption |
| 11 | (and note that perl 5.10.0 has a double-free bug |
124 | bugs in PerlIO::via by using our own C-level perliol. As a side |
| 12 | that is not being workarounded). |
125 | effect, Coro::Storable is now much, much, much faster. |
| 13 | |
126 | |
| 14 | 4.747 Tue Sep 23 01:59:41 CEST 2008 |
127 | 4.747 Tue Sep 23 01:59:41 CEST 2008 |
| 15 | - fix a per-cv memleak (one empty array was leaked per |
128 | - fix a per-cv memleak (one empty array was leaked per |
| 16 | code reference). |
129 | code reference). |
| 17 | - avoid a crash in coro->call|rss when the coroutine was already |
130 | - avoid a crash in coro->call|rss when the coroutine was already |
