| … | |
… | |
| 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? (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. |
|
|
9 | TODO: aio_mincore? |
|
|
10 | TODO: getxattr etc.? |
|
|
11 | TODO: MS_xxx, posix_fadvise, posix_madvise |
|
|
12 | TODO: realpath |
|
|
13 | |
|
|
14 | 3.92 Wed Jun 29 14:45:41 CEST 2011 |
|
|
15 | - ecb.h was missing. |
|
|
16 | |
|
|
17 | 3.91 Wed Jun 29 13:24:42 CEST 2011 |
|
|
18 | - (libeio) work around a Linux (and likely FreeBSD and other |
|
|
19 | kernels) bug where sendfile would not transfer all the requested |
|
|
20 | bytes on large transfers, using a heuristic. |
|
|
21 | - buggy sendfile caused aio_move/copy to sometimes fail for big |
|
|
22 | files (fortunately it checks that the whole file has been |
|
|
23 | transferred...) |
|
|
24 | - use libecb for higher performance and higher portability. |
|
|
25 | - (libeio) disable sendfile on darwin, it's too broken. |
|
|
26 | - disable fork tests on !linux, as only linux supports mixing |
|
|
27 | pthread and fork in perl. |
|
|
28 | - document the fact that fork doesn't work anymore when using |
|
|
29 | this module, on anything !GNU/Linux. |
|
|
30 | - increase timeout in t/04_fork.t, as too many CPAN-tester setups |
|
|
31 | run on a 0.1MHz cpu. Or so. |
|
|
32 | |
|
|
33 | 3.9 Fri May 27 02:43:47 CEST 2011 |
|
|
34 | - (libeio) fix memory corruption in aio_readdirx for the flags |
|
|
35 | combination READDIR_STAT_ORDER | READDIR_DIRS_FIRST. |
|
|
36 | - add lots of newre POSIX and GNU/Linux-specific open |
|
|
37 | flags. |
|
|
38 | |
|
|
39 | 3.8 Sun Mar 27 12:25:33 CEST 2011 |
|
|
40 | - use nonstandard but maybe-working-on-bsd fork technique. |
|
|
41 | - support a max_idle value of 0. |
|
|
42 | - support setting of idle timeout value (IO::AIO::idle_timeout). |
|
|
43 | |
|
|
44 | 3.72 Fri Feb 11 04:25:38 CET 2011 |
|
|
45 | - use _POSIX_MEMLOCK_RANGE to detect mlock/munlock. |
|
|
46 | - aio_mknod always used a dev_t value of 0. |
|
|
47 | - new treescan option: --grep. |
|
|
48 | - add more S_IF macros, and major/minor/makedev "macros". |
|
|
49 | |
|
|
50 | 3.71 Thu Dec 30 08:18:46 CET 2010 |
|
|
51 | - the numerical result value passed to callbacks did not stringify |
|
|
52 | correctly, due to internal reuse and failure to reset the sv flags. |
|
|
53 | - actually test for posix_[mf]advise, as at least uClibc defines |
|
|
54 | _POSIX_ADVISORY_INFO without actually having any of the required |
|
|
55 | funcitonality. ugh. |
|
|
56 | |
|
|
57 | 3.7 Mon Nov 1 23:00:34 CET 2010 |
|
|
58 | - implement/add madvise, mmap, munmap, aio_mlockall, munlockall, |
|
|
59 | aio_mlock, munlock, msync and mtouch to @EXPORT_OK. |
|
|
60 | - document the sad state of affairs w.r.t. pthread on many bsds. |
|
|
61 | - do not enable mmap on systems without _POSIX_MAPPED_FILES |
|
|
62 | (openbsd 4.8). |
|
|
63 | - do not leak memory in IO::AIO::mmap when the scalar already |
|
|
64 | had string-data. |
|
|
65 | - add O_RDWR, O_APPEND and O_EXCL symbols and semi-document |
|
|
66 | them. |
|
|
67 | - cache the result IV, for a minor speedup in the common case. |
|
|
68 | - croak when an mmapped-scalar changes location, to detect |
|
|
69 | user-errors better. |
|
|
70 | - fix aio_readlink prototype. |
|
|
71 | |
|
|
72 | 3.65 Wed Mar 31 02:45:05 CEST 2010 |
|
|
73 | - actually use PATH_MAX instead of NAME_MAX for readlink, as to not |
|
|
74 | cut off long pathnames in aio_readlink (based on patch by |
|
|
75 | Rasmus Andersson). |
|
|
76 | - a double fork partially killed the event pipe (great testcase |
|
|
77 | by dormando). affects IO::AIO, BDB and Async::Interrupt. |
|
|
78 | - suck steve's dick and rename our symbols to avoid colliding |
|
|
79 | with nonstandard identifier spam on the broken os x pseudo-os. |
|
|
80 | affects IO::AIO and BDB. |
|
|
81 | |
|
|
82 | 3.6 Mon Jan 11 00:43:39 CET 2010 |
|
|
83 | - (libeio) more fixes for the freebsd/apple sendfile - broken by |
|
|
84 | manpage standards, but apparently correct accoridng to actual |
|
|
85 | kernel sources. |
|
|
86 | - add IO::AIO::mmap/munmap support. |
|
|
87 | - add IO::AIO::m(un)lockall support. |
|
|
88 | - clean up manpage. |
|
|
89 | |
|
|
90 | 3.5 Thu Jan 7 21:25:04 CET 2010 |
|
|
91 | - (liebio) fix freebsd sendfile (Vanilla Hsu). |
|
|
92 | - (libeio) also fix darwin sendfile that suffered from a similar bug. |
|
|
93 | - add aio_statvfs as interface to statvfs/fstatvfs. |
|
|
94 | - work around buggy symlink() on cygwin in the testsuite. |
|
|
95 | - wtf. freebsd-8.0 actually passes the testsuite! I knew |
|
|
96 | they could implement fork and semaphores one day! cheers! |
|
|
97 | |
|
|
98 | 3.4 Sat Jan 2 15:13:04 CET 2010 |
|
|
99 | - (libeio) max_poll_time was not properly converted to ticks. |
|
|
100 | - clarify cancel_subs description. |
|
|
101 | - IO::AIO::sendfile did not actually return the return value |
|
|
102 | from the sendfile call. |
|
|
103 | - implement aio_msync, aio_mtouch. |
|
|
104 | - (libeio) tentatively support darwin in sendfile. |
|
|
105 | |
|
|
106 | 3.31 Thu Nov 12 02:14:29 CET 2009 |
|
|
107 | - fix result status documentation of aio_copy, aio_move. |
|
|
108 | - speed up object creation considerably by avoiding |
|
|
109 | hash lookups. |
|
|
110 | |
|
|
111 | 3.3 Wed Aug 5 13:52:58 CEST 2009 |
|
|
112 | - use common::sense. |
|
|
113 | - use common schmorp.h header. |
|
|
114 | - allow integers (file descriptors) in addition to file handles. |
|
|
115 | - take advantage of linux' eventfd (but it seems slower on smp?). |
|
|
116 | - use poll not select on posix platforms. |
|
|
117 | |
|
|
118 | 3.261 Wed Jul 1 10:11:51 CEST 2009 |
|
|
119 | - more 0S X workarounds (patch by Tokuhiro Matsuno). |
|
|
120 | |
|
|
121 | 3.26 Tue Jun 30 09:33:26 CEST 2009 |
|
|
122 | - 0S X of course claims to be posix 2008 but lacks posix_fadvise. |
|
|
123 | try to work around this horribly broken OS in a somewhat hackish |
|
|
124 | way. might help other os's too. |
|
|
125 | |
|
|
126 | 3.25 Sat Jun 27 05:18:26 CEST 2009 |
|
|
127 | - added IO::AIO::fadvise and IO::AIO::sendfile. |
|
|
128 | - (libeio) replaced quicksort+insertion sort by a tuned radix |
|
|
129 | sort + insertion sort, resulting in comparable runtime (usually |
|
|
130 | faster) to the old version, but without any ill side effects on |
|
|
131 | degenerated (for quicksort) data. |
|
|
132 | - (libeio) correctly sort dirs by inodes if we have perfect type |
|
|
133 | knowledge and DIRS_FIRST is used. |
|
|
134 | - (libeio) internally the wrong DT_* constants were used, but no |
|
|
135 | known system has differing EIO_DT_*/DT_* values, so not an issue. |
|
|
136 | - removed a (harmless) assert that was left over on the code |
|
|
137 | but should not have been. |
|
|
138 | - use more correct types (change IO length from IV to VAL64, |
|
|
139 | which makes a difference on systems without 64 bit perls). |
|
|
140 | |
|
|
141 | 3.23 Sat Jun 13 16:57:58 CEST 2009 |
|
|
142 | - fix off-by-one bug in aio_readdir that was introduced in the |
|
|
143 | rushed 3.22 release. |
|
|
144 | |
|
|
145 | 3.22 Sat Jun 13 15:32:40 CEST 2009 |
|
|
146 | - speed up readdirx sort algorithm slightly. |
|
|
147 | - bin/treescan was missing from distro tarball. |
|
|
148 | |
|
|
149 | 3.21 Fri Jun 12 18:45:53 CEST 2009 |
|
|
150 | - new options --dirs and --files for treescan. |
|
|
151 | - install bin/treescan by default. |
|
|
152 | - (libeio) aio_readdir can now be cancelled while executing. |
|
|
153 | - fix a printf format string for 64 bit systems (could lead |
|
|
154 | to problems on big endian 64 bit systems). |
|
|
155 | - do not use qsort() but our own algorithm: glibc initialises |
|
|
156 | the whole locale and I/O subsystem inside qsort, causing |
|
|
157 | a stack overflow on 32 bit machines. The new sort uses much less |
|
|
158 | stack and is more than twice as fast in typical situations. |
| 8 | |
159 | |
| 9 | 3.2 Sun Jun 7 20:30:05 CEST 2009 |
160 | 3.2 Sun Jun 7 20:30:05 CEST 2009 |
| 10 | - (libeio) pwrite emulation was even more flawed than intended and did |
161 | - (libeio) pwrite emulation was even more flawed than intended and did |
| 11 | not restore the file offset. |
162 | not restore the file offset. |
| 12 | - add aio_readdirx, which can return inode and filetype and sort |
163 | - add aio_readdirx, which can return inode and filetype and sort |
