| 1 | Revision history for Perl extension Coro. |
1 | Revision history for Perl extension Coro. |
| 2 | |
2 | |
| 3 | TODO: better (earlier) error messages on broken transfer's |
3 | TODO: should explore PerlIO::via::CoroCede and PerlIO::via::CoroAIO. |
|
|
4 | TODO: maybe implement a default message channel, very much like Erlang |
|
|
5 | (which is cool in a lot of important aspects (failures!), |
|
|
6 | but very lacking in others (higher level ipc)). |
| 4 | |
7 | |
|
|
8 | - enable/disable tracing from a new coroutine, not a pooled one. |
|
|
9 | - fix a memleak in Coro::Event. |
|
|
10 | - removed killall call from fork_eval. |
|
|
11 | - made sure store_fd is already loaded so that fork_eval |
|
|
12 | does not have to parse autoload in each subprocess. |
|
|
13 | - if the event model is EV, use EV::DNS to resolve |
|
|
14 | stuff in Coro::Util. |
|
|
15 | - don't get confused by multiple event notifications in Coro::Handle. |
|
|
16 | - initial support for EV (libevent interface). |
|
|
17 | |
|
|
18 | 4.13 Wed Oct 24 07:26:45 CEST 2007 |
|
|
19 | - add Coro::Storable::blocking_thaw. |
|
|
20 | - use a vastly more complicated technique to localise |
|
|
21 | $SIG{__WARN/DIE__} that also works on perls <= 5.8.8. |
|
|
22 | - use a coroutine for the idle callback Coro::Event, |
|
|
23 | instead of running Event in the current coroutine context. |
|
|
24 | This also catches recursive invocations. |
|
|
25 | - actually report fork errors in gethostbyname and inet_aton. |
|
|
26 | |
|
|
27 | 4.11 Thu Oct 11 02:40:24 CEST 2007 |
|
|
28 | - port to threaded perls. |
|
|
29 | |
|
|
30 | 4.1 Thu Oct 11 02:38:16 CEST 2007 |
|
|
31 | - incompatible change: $SIG{__DIE__} and $SIG{__WARN__} will now |
|
|
32 | be local to each coro (see Coro::State). |
|
|
33 | - incompatible change: for very deep reasons, cede and cede_notself |
|
|
34 | cannot return anything, so nothing will be returned. |
|
|
35 | - possibly bring back 5.10 compatibility (untested). |
|
|
36 | - work around stupid (and wrong) warning on 5.10 :(. |
|
|
37 | - overlay the saved state over the context stack. This saves |
|
|
38 | a few hundred bytes per coroutine on average and also |
|
|
39 | speeds up context switching a bit. |
|
|
40 | - further tune default stack sizes. |
|
|
41 | - (more) correctly calculate stack usage in coro_rss. |
|
|
42 | - Coro::Storable::blocking_* did not properly lock |
|
|
43 | resulting in races between coroutines. |
|
|
44 | - added Coro::Storable::guard. |
|
|
45 | - stopping to trace a coroutine could destroy the cctx of |
|
|
46 | an unrelated coroutine. |
|
|
47 | - explain the relationship between Perl and C coroutines in |
|
|
48 | more detail in Coro::State. |
|
|
49 | - Coro::Util::inet_aton did not short-circuit dotted quad forms, |
|
|
50 | causing a fork per resolve. This also affected Coro::Socket. |
|
|
51 | - switch to a separate stack in $coro->call/eval to avoid |
|
|
52 | invalidating pointers. |
|
|
53 | |
|
|
54 | 4.03 Sat Oct 6 21:24:00 CEST 2007 |
|
|
55 | - added Coro::throw method. |
|
|
56 | - minor code cleanups. |
|
|
57 | |
|
|
58 | 4.02 Sat Oct 6 02:36:47 CEST 2007 |
|
|
59 | - fix a very minor per-coroutine memleak (a single codereference). |
|
|
60 | - fixed a bug where the currently in-use c context would be freed |
|
|
61 | prematurely (can happen only when programs change the stacksize |
|
|
62 | or use tracing). |
|
|
63 | - tracing can no longer keep a coro alive after it terminated. |
|
|
64 | - do static branch prediction in the common path for gcc. gives |
|
|
65 | about 2-5% speed improvement here. |
|
|
66 | |
|
|
67 | 4.01 Fri Oct 5 22:10:49 CEST 2007 |
|
|
68 | - instead of recreating *a* standard output handle we simply |
|
|
69 | use STDOUT, which is faster and hopefully more robust. |
|
|
70 | |
|
|
71 | 4.0 Fri Oct 5 12:56:00 CEST 2007 |
|
|
72 | - incompatibly changed Coro::Storable::freeze. |
|
|
73 | - major new feature: added Coro::Debug, for interactive coroutine |
|
|
74 | debugging, tracing and much more. |
|
|
75 | - major bug fix: unbelievable, but true: $_, $/ and many other |
|
|
76 | "saved" variables actually weren't being saved. This has been fixed, |
|
|
77 | of course, while increasing performance while losing all the save |
|
|
78 | flags. |
|
|
79 | - save flags are gone, and all the api functions dealing with them. |
|
|
80 | - added Coro::Semaphore::adjust. |
|
|
81 | - added Coro::Util::fork_eval. |
|
|
82 | - added Coro::Storable::{nfreeze,blocking_{freeze,nfreeze}}. |
|
|
83 | - added Coro::killall. |
|
|
84 | - reduce initial stack sizes to allow for "micro-coroutines". |
|
|
85 | - better async_pool resource management, moved parts of async_pool |
|
|
86 | handling to XS (major speed improvement). |
|
|
87 | - actually croak before modifying important data structures. |
|
|
88 | - refuse to transfer while compiling. |
|
|
89 | - possibly support eval EXPR better now. |
|
|
90 | - enable assembly per default on linux+bsd x86+amd64. |
|
|
91 | - all internal members were renamed _something for easier subclassing. |
|
|
92 | - many minor tweaks. |
|
|
93 | |
|
|
94 | 3.63 Wed May 16 14:10:06 CEST 2007 |
|
|
95 | - implement handcoded assembly for x86/amd64 SVR ABI. |
|
|
96 | |
|
|
97 | 3.62 Fri Apr 27 21:36:06 CEST 2007 |
|
|
98 | - upgrade libcoro (which might set unwind info correctly). |
|
|
99 | - change default on linux to setjmp/longjmp. |
|
|
100 | |
|
|
101 | 3.61 Thu Apr 19 12:36:18 CEST 2007 |
|
|
102 | - Coro::Storable caused an endless loop when thawing invalid |
|
|
103 | pst's sometimes. |
|
|
104 | - use a Semaphore in Coro::Storable, as Storable doesn't |
|
|
105 | seem to be reentrant (although it is documented to |
|
|
106 | be threadsafe...). |
|
|
107 | - fix Coro::Signal to bring back the original unreliable |
|
|
108 | but stateful semantics. |
|
|
109 | - fixed a lot of typos in Coro.pm (patch submitted by David |
|
|
110 | Steinbrunner, which applied flawlessly). |
|
|
111 | |
|
|
112 | 3.6 Sat Apr 14 17:13:31 CEST 2007 |
|
|
113 | - added some bugfixes to get eg/myhttpd working again. |
|
|
114 | - added Coro::Storable for often-cede'ing freeze/thaw. |
|
|
115 | - try to do a clean exit when a coroutine calls exit |
|
|
116 | (EXPERIMENTAL). |
|
|
117 | - got rid of indirect call through _coro_init. |
|
|
118 | - updated the partly antique examples in eg/ to |
|
|
119 | work again and be a bit less magic, too. |
|
|
120 | - fixed Coro::Signal semantics to work as documented again. |
|
|
121 | |
|
|
122 | 3.55 Sun Mar 25 01:20:47 CET 2007 |
|
|
123 | - add SAVE_DEFFH to save the default filehandle and enable |
|
|
124 | it by default. |
|
|
125 | - finally move socket-operations from Coro::Socket to Coro::Handle |
|
|
126 | to be able to unblock foreign sockets. |
|
|
127 | - add Coro::State::save_also and guarded_save. |
|
|
128 | - add count accessor to Coro::Semaphore. |
|
|
129 | - add Coro::State::cctx_stacksize. |
|
|
130 | - just for the fun of it, do not rely on implicit context, |
|
|
131 | which can dramatically improve performance, but people |
|
|
132 | using windows-process-emulation perls do not care much about |
|
|
133 | performance. |
|
|
134 | |
|
|
135 | 3.51 Sun Mar 4 14:18:04 CET 2007 |
|
|
136 | - fixed a problem when you weakened references to Coro::State's |
|
|
137 | (patch by Yuval Kogman). |
|
|
138 | |
|
|
139 | 3.501 Wed Feb 28 12:44:07 CET 2007 |
|
|
140 | - rename some global symbols as macosx from hell redefines |
|
|
141 | them without asking. |
|
|
142 | |
|
|
143 | 3.5 Tue Feb 13 20:22:53 CET 2007 |
|
|
144 | - do AnyEvent model detection earlier, avoiding problems |
|
|
145 | caused by first using AnyEvent and later Coro::Event. |
|
|
146 | - implement and document Coro::Event event objects. |
|
|
147 | - fix a potential problem in Coro::Event causing crashes. |
|
|
148 | - initialise PL_comppad when creating a new coroutine, |
|
|
149 | avoids crashes on early coro destruction. |
|
|
150 | |
|
|
151 | 3.41 Mon Jan 22 19:19:49 CET 2007 |
|
|
152 | - readline on Coro::Handle did not support undefined $/, |
|
|
153 | nor did it deliver partial lines on EOF or error. |
|
|
154 | - implement malloc fallback for stack allocation because |
|
|
155 | stupid broken idiotic OSX has a stupid broken |
|
|
156 | idiotic fits-the-whole-os mmap "implementation" and |
|
|
157 | my dick feels longer if Coro is portable even to |
|
|
158 | obsolete platforms. |
|
|
159 | |
|
|
160 | 3.4 Fri Jan 19 21:52:54 CET 2007 |
|
|
161 | - remove t/09_timer.t, as it isn't really testing much |
|
|
162 | but was rather flaky in practise. |
|
|
163 | - async_pool coro would keep arguments and callback alive until |
|
|
164 | it was reused. |
|
|
165 | - cancellation of a coroutine could cause spurious idle calls |
|
|
166 | in cede_notself. |
|
|
167 | |
|
|
168 | 3.3 Sat Jan 6 03:45:00 CET 2007 |
|
|
169 | - implement $coro->on_destroy. |
|
|
170 | - Coro::Event blocking semantics have been changed, |
|
|
171 | documented and - hopefully - improved. |
| 5 | - fix nice adding, not subtracting, from priority. |
172 | - fix nice adding, not subtracting, from priority. |
| 6 | - fix ->prio and api_is_ready (patch by Mark Hinds). |
173 | - fix ->prio and api_is_ready (patch by Mark Hinds). |
|
|
174 | - fixed an assert ("... == prev__cctx->idle_te") |
|
|
175 | that could errronously trigger. |
|
|
176 | - fix various large and small memleaks. |
|
|
177 | - use a (hopefully) more stable cancel implementation |
|
|
178 | that immediately frees the coroutine data. |
|
|
179 | - cede/cede_notself return a status now. |
|
|
180 | - added Coro::guard function. |
|
|
181 | - added a global coroutine pool for jobs (on my machine, |
|
|
182 | I can create and execute 48k simple coros/s with async, |
|
|
183 | and 128k coros with async_pool). |
|
|
184 | - Coro::AIO now uses the coroutine priority as io priority. |
| 7 | |
185 | |
| 8 | 3.2 Fri Dec 22 05:07:09 CET 2006 |
186 | 3.2 Fri Dec 22 05:07:09 CET 2006 |
| 9 | - improve portability to slightly older perls. |
187 | - improve portability to slightly older perls. |
| 10 | - use cleaner coroutine destruction. |
188 | - use cleaner coroutine destruction. |
| 11 | - simplify configuration for users. |
189 | - simplify configuration for users. |
