| 1 | Revision history for IO::AIO |
1 | Revision history for IO::AIO |
| 2 | |
2 | |
| 3 | TODO: better autoconf.pm that can return LIBS etc. |
3 | TODO: better autoconf.pm that can return LIBS etc. |
| 4 | TODO: aio_cptree/mvtree |
4 | TODO: aio_cptree/mvtree |
| 5 | TODO: reduce condvar fairness: schedule hot-cache-threads first? |
5 | TODO: reduce condvar fairness: schedule hot-cache-threads first? |
| 6 | TODO: splice/tee/vmsplice/sync_file_range? (http://kerneltrap.org/node/6505 http://lwn.net/Articles/178199/) |
6 | TODO: splice/tee/vmsplice? (http://kerneltrap.org/node/6505 http://lwn.net/Articles/178199/) |
| 7 | TODO: aio_fcntl, at least for file-locking |
7 | TODO: aio_fcntl, at least for file-locking |
|
|
8 | TODO: fallocate, if it wouldn't be so useless. |
| 8 | |
9 | |
| 9 | 3.03 |
10 | - use common::sense. |
|
|
11 | - use common schmorp.h header. |
|
|
12 | - allow integers (file descriptors) in addition to file handles. |
|
|
13 | |
|
|
14 | 3.261 Wed Jul 1 10:11:51 CEST 2009 |
|
|
15 | - more OS X workarounds (patch by Tokuhiro Matsuno). |
|
|
16 | |
|
|
17 | 3.26 Tue Jun 30 09:33:26 CEST 2009 |
|
|
18 | - os x of course claims posix 2008 but lacks posix_fadvise. try to work |
|
|
19 | around this horribly broken OS in a somewhat hackish way. might help |
|
|
20 | other os's too. |
|
|
21 | |
|
|
22 | 3.25 Sat Jun 27 05:18:26 CEST 2009 |
|
|
23 | - added IO::AIO::fadvise and IO::AIO::sendfile. |
|
|
24 | - (libeio) replaced quicksort+insertion sort by a tuned radix |
|
|
25 | sort + insertion sort, resulting in comparable runtime (usually |
|
|
26 | faster) to the old version, but without any ill side effects on |
|
|
27 | degenerated (for quicksort) data. |
|
|
28 | - (libeio) correctly sort dirs by inodes if we have perfect type |
|
|
29 | knowledge and DIRS_FIRST is used. |
|
|
30 | - (libeio) internally the wrong DT_* constants were used, but no |
|
|
31 | known system has differing EIO_DT_*/DT_* values, so not an issue. |
|
|
32 | - removed a (harmless) assert that was left over on the code |
|
|
33 | but should not have been. |
|
|
34 | - use more correct types (change IO length from IV to VAL64, |
|
|
35 | which makes a difference on systems without 64 bit perls). |
|
|
36 | |
|
|
37 | 3.23 Sat Jun 13 16:57:58 CEST 2009 |
|
|
38 | - fix off-by-one bug in aio_readdir that was introduced in the |
|
|
39 | rushed 3.22 release. |
|
|
40 | |
|
|
41 | 3.22 Sat Jun 13 15:32:40 CEST 2009 |
|
|
42 | - speed up readdirx sort algorithm slightly. |
|
|
43 | - bin/treescan was missing from distro tarball. |
|
|
44 | |
|
|
45 | 3.21 Fri Jun 12 18:45:53 CEST 2009 |
|
|
46 | - new options --dirs and --files for treescan. |
|
|
47 | - install bin/treescan by default. |
|
|
48 | - (libeio) aio_readdir can now be cancelled while executing. |
|
|
49 | - fix a printf format string for 64 bit systems (could lead |
|
|
50 | to problems on big endian 64 bit systems). |
|
|
51 | - do not use qsort() but our own algorithm: glibc initialises |
|
|
52 | the whole locale and I/O subsystem inside qsort, causing |
|
|
53 | a stack overflow on 32 bit machines. The new sort uses much less |
|
|
54 | stack and is more than twice as fast in typical situations. |
|
|
55 | |
|
|
56 | 3.2 Sun Jun 7 20:30:05 CEST 2009 |
|
|
57 | - (libeio) pwrite emulation was even more flawed than intended and did |
|
|
58 | not restore the file offset. |
|
|
59 | - add aio_readdirx, which can return inode and filetype and sort |
|
|
60 | the names in various ways. |
|
|
61 | - unfortunately, utime, chmod, chown on an open file that has just |
|
|
62 | been written can easily block, which caused aio_copy to block the |
|
|
63 | process. no more! |
|
|
64 | - no longer rely on dst path in aio_copy when futime is available. |
|
|
65 | |
|
|
66 | 3.19 Tue Apr 21 22:05:21 CEST 2009 |
|
|
67 | - more perl 5.10 workarounds for aio_read and write. |
|
|
68 | - aio_write no longer modifies the sv (if possible). |
|
|
69 | - aio_read now works correctly with magic values. |
|
|
70 | |
|
|
71 | 3.18 Sun Apr 19 21:17:32 CEST 2009 |
|
|
72 | - better diagnostics when some aio_* functions get passed an illegal fh. |
|
|
73 | - try to avoid crashes due to incompatible 5.10 API changes. grmbl. |
|
|
74 | |
|
|
75 | 3.17 Thu Nov 20 08:45:36 CET 2008 |
|
|
76 | - (libeio) added aio_sync_file_range (untested). |
|
|
77 | - add aio_busy to @IO::AIO::AIO_REQ. |
|
|
78 | |
|
|
79 | 3.16 Wed Oct 22 18:28:01 CEST 2008 |
|
|
80 | - use SvREFCNT_inc instead of SvREFCNT_inc_NN in a non-speed critical |
|
|
81 | part to improve portability to perl 5.8 (reported by szymon). |
|
|
82 | |
|
|
83 | 3.15 Mon Oct 13 00:39:55 CEST 2008 |
|
|
84 | - automatic removal of feeders was broken. |
|
|
85 | - (libeio) use a more robust method to detect whether a feeder |
|
|
86 | has added something to the group or not. |
|
|
87 | |
|
|
88 | 3.1 Thu Oct 2 13:34:40 CEST 2008 |
|
|
89 | - pre-deref the passed callback object, for increased speed |
|
|
90 | and decreased memory usage. |
|
|
91 | - call on_next_submit callback even for aio_group - impact unknown, |
|
|
92 | but seems more correct. |
|
|
93 | - $req->cb now returns the original callback. |
|
|
94 | - (libeio) pass EIO_GROUP/aio_group requests through the queue without |
|
|
95 | requiring a thread switch. |
|
|
96 | - (libeio) status code of aio_futime and aio_utime was always passed 0. |
|
|
97 | - do some other ยต-optimisations. |
|
|
98 | |
|
|
99 | 3.07 Sat Aug 2 16:06:13 CEST 2008 |
|
|
100 | - do not include LIBS in autoconf tests. |
|
|
101 | |
|
|
102 | 3.06 Tue Jul 15 12:41:32 CEST 2008 |
|
|
103 | - move preadwritelock definition before it's first reference, |
|
|
104 | to make it compile again on broken platforms (they still exist) |
|
|
105 | or badly cofngiured perls (redhat...), reported by Rob Bloodgood. |
|
|
106 | |
|
|
107 | 3.05 Thu Jun 19 23:23:52 CEST 2008 |
|
|
108 | - work around perl overriding readdir etc. with thread-unsafe |
|
|
109 | (sic!) versions. |
|
|
110 | |
|
|
111 | 3.04 Wed Jun 18 01:35:38 CEST 2008 |
|
|
112 | - (libeio) fix eio_mknod, which confused it's arguments. |
|
|
113 | - (libeio) do not use readdir_r, as Ulrich Drepper pointed out that this |
|
|
114 | is stupid. |
|
|
115 | - (libeio) fix eio__readahead prototype, patch by Jost Krieger. |
|
|
116 | - (libeio) fix a bug that could potentially cause IO::AIO |
|
|
117 | not to get initialised properly. |
|
|
118 | |
|
|
119 | 3.03 Thu May 29 05:33:30 CEST 2008 |
| 10 | - (libeio) correctly call pthread_attr_destroy (). |
120 | - (libeio) correctly call pthread_attr_destroy (). |
| 11 | - (libeio) work around broken bsd headers once more. |
121 | - (libeio) work around broken bsd headers once more. |
| 12 | - reduce shared library size again by not including wrappers. |
122 | - reduce shared library size again by not including wrappers. |
| 13 | - max_outstanding could cause poll to enter a busy-waiting loop. |
123 | - max_outstanding could cause poll to enter a busy-waiting loop. |
| 14 | - document the new IO::AIO::poll_cb result value. |
124 | - document the new IO::AIO::poll_cb result value. |
