| 1 |
Revision history for IO::AIO |
| 2 |
|
| 3 |
TODO: better autoconf.pm that can return LIBS etc. |
| 4 |
TODO: aio_cp/mvtree/rmtree |
| 5 |
TODO: aio_utime, aio_chmod, aio_chown? |
| 6 |
|
| 7 |
2.1 Sun Oct 29 02:01:57 CET 2006 |
| 8 |
- INCOMPATIBLE CHANGE: replace IO::AIO::poll_some by |
| 9 |
IO::AIO::max_poll_reqs and IO::AIO::max_poll_time. |
| 10 |
- fix a bug in aio_move where it would fail everytime |
| 11 |
it would have to copy files manually. |
| 12 |
- doc updates. |
| 13 |
- start threads only on (very low) demand. |
| 14 |
- end superfluous threads automatically after 10s idling. |
| 15 |
- add IO::AIO::max_idle to control this. |
| 16 |
- some tuning applied to random places. |
| 17 |
- add aio_mknod. |
| 18 |
|
| 19 |
2.0 Thu Oct 26 18:27:58 CEST 2006 |
| 20 |
- minor incompatibility: max_outstanding semantics have changed. |
| 21 |
- major rewrite, consider all functionality EXPERIMENTAL. |
| 22 |
- aio requests can be cancelled. |
| 23 |
- callbacks can be (re-)set later. |
| 24 |
- aio requests can choose between 9 priorities. |
| 25 |
- add aio_group: aio requests can be grouped |
| 26 |
into composite requests. |
| 27 |
- generator interface for groups. |
| 28 |
- add IO::AIO::poll_some. |
| 29 |
- major documentation improvements. |
| 30 |
- very minor bugfixes. |
| 31 |
- add aio_nop for dummy requests. |
| 32 |
- add aio_busy mainly for benchmarking and debugging. |
| 33 |
- use "fast" mutexes on linux. |
| 34 |
- use dynamic (and larger) buffers instead of |
| 35 |
putting them onto the stack. |
| 36 |
- optimise x86 and amd64 a bit. |
| 37 |
- better error checking when ENOMEM. |
| 38 |
- hopefully fix mem and dirp leaks on fork. |
| 39 |
- warn about broken perl malloc. |
| 40 |
- compiles on cygwin (but pread is broken on cygwin |
| 41 |
which is not detected by autoconf, so you have to |
| 42 |
edit autoconf/config.h yourself to enable emulation). |
| 43 |
|
| 44 |
1.8 Sat Jun 24 22:20:44 CEST 2006 |
| 45 |
- add and/or document aio_rename, aio_link, aio_symlink and aio_move. |
| 46 |
|
| 47 |
1.73 Wed Mar 1 22:49:32 CET 2006 |
| 48 |
- codename "jost" |
| 49 |
- AIX/Solaris(?) aren't being helpful again, hardcode a value |
| 50 |
for NAME_MAX and drive with your eyes closed (reported by |
| 51 |
wcooley@nakedape.cc). |
| 52 |
|
| 53 |
1.72 Thu Feb 2 00:40:44 CET 2006 |
| 54 |
- another minor bug in aio_scandir, again, the callback was called |
| 55 |
twice when the directory can't be stat'ed. |
| 56 |
|
| 57 |
1.71 Mon Dec 26 20:20:03 CET 2005 |
| 58 |
- grr, another minor bug in aio_scandir, this time resulting |
| 59 |
in a runtime error. |
| 60 |
|
| 61 |
1.7 Mon Dec 26 19:29:48 CET 2005 |
| 62 |
- aio_scandir errornously called callback twice in some cases. |
| 63 |
- added AnyEvent example. |
| 64 |
|
| 65 |
1.61 Wed Sep 7 19:40:42 CEST 2005 |
| 66 |
- no longer include $Config{libs}, as distros include too much |
| 67 |
garbage in there that is not installed on typical systems |
| 68 |
(for example, debian links perl against -lgdbm and many more |
| 69 |
libraries despite perl not needing it, slowing down startup and |
| 70 |
wasting memory). |
| 71 |
|
| 72 |
1.6 Tue Aug 30 17:44:44 CEST 2005 |
| 73 |
- added aio_readdir, aio_scandir. |
| 74 |
- added aio_sendfile, including sendfile emulation when sendfile |
| 75 |
isn't available. Linux, FreeBSD and HP-UX might work (sendfilev |
| 76 |
support is unavailable on my solaris 8 machine, but preliminary |
| 77 |
code is there. What's lacking is configury stuff and testing...). |
| 78 |
|
| 79 |
1.5 Sat Aug 20 02:32:19 CEST 2005 |
| 80 |
- create the aio threads on demand only (good after forking). |
| 81 |
- somebody nagged about C89 compatibility, so this release should |
| 82 |
be C89 compatible, but no guarantees for future ones. |
| 83 |
- abort when the configure script fails. |
| 84 |
|
| 85 |
1.4 Wed Aug 17 08:07:27 CEST 2005 |
| 86 |
- forgot to recreate the pipe after forking, this could cause |
| 87 |
deadlocks. |
| 88 |
|
| 89 |
1.3 Wed Aug 17 07:25:54 CEST 2005 |
| 90 |
- properly propagate exceptions from callbacks, instead of silently |
| 91 |
eating them. |
| 92 |
- use a different fork algorithm that avoids recursive callback |
| 93 |
invocations and reduces fork latency in busy processes. |
| 94 |
- fix a bug where the buffer scalar in aio_read got assigned the wrong |
| 95 |
length after reading. |
| 96 |
- fix a bug where the child would process some outstanding requests |
| 97 |
from the parent. |
| 98 |
- make a copy of the callback scalar, to avoid some very uncommon |
| 99 |
(but valid) usages where the callback scalar changes. |
| 100 |
- the callback was never freed in aio_read/aio_write. |
| 101 |
- aio_read/aio_write will now set the readonly flag on the data |
| 102 |
scalar for better error reporting. |
| 103 |
|
| 104 |
1.2 Wed Aug 17 01:22:58 CEST 2005 |
| 105 |
- verified on cygwin, linux/amd64+ia32+alpha, solaris. |
| 106 |
- use SvPVbyte for filenames now. You *need* to encode your strings to |
| 107 |
the external filename encoding first. |
| 108 |
- fix the bug where filehandles could leak (again). |
| 109 |
- clarify freebsd situation: make it work and tell me how, or go away |
| 110 |
with your underdocumented, broken and outdated OS. |
| 111 |
- fork'ing should now be officially supported (where linux and |
| 112 |
posix contradict each other, I followed posix). |
| 113 |
- only call read() once per poll, not twice (speedup). |
| 114 |
- new functions: aio_rmdir, aio_symlink. |
| 115 |
- force callback arguments to be undef or a reference, for better |
| 116 |
error catching. |
| 117 |
|
| 118 |
1.1 Sun Aug 7 05:27:23 CEST 2005 |
| 119 |
- added configure script to detect availability of some calls. |
| 120 |
(imperfect, as we cannot detect things like -lrt due to missing |
| 121 |
feedback yet). |
| 122 |
- emulate fdatasync via fsync, pread/pwrite via unsafe read/write |
| 123 |
and readahead via manual reading on platforms lacking those. |
| 124 |
- ported t/03_errors.t+framework by Brad Fitzpatrick |
| 125 |
from Linux::AIO. |
| 126 |
- removed XOPEN_SOURCE because stupid solaris 9 doesn't |
| 127 |
compile with it enabled. Probably breaks others. Oh well. |
| 128 |
- try to be more careful with errno on non-threaded perls. |
| 129 |
- cygwins perl insisted on a matching open mode when morphing |
| 130 |
the fd to an fh. |
| 131 |
- convert sv's to "byte" encoding in aio_read/aio_write. Might |
| 132 |
do that for pathnames, too, in the future, so make sure you |
| 133 |
properly encode your pathnames. |
| 134 |
|
| 135 |
1.0 Fri Jul 22 10:43:30 CEST 2005 |
| 136 |
- aio_open did not reliably deliver (open) filehandles. |
| 137 |
- all aio_*-functions keep a reference to the filehandle so |
| 138 |
it doesn't go away while the aio requests is working on it. |
| 139 |
- disable warnings in AIO.pm. |
| 140 |
|
| 141 |
0.9 Wed Jul 20 23:55:04 CEST 2005 |
| 142 |
- allow recursive calls to poll_cb, use less locking calls, too. |
| 143 |
- fix the Gtk2-example in the SYNOPSIS. |
| 144 |
- use pread/pwrite not pread64/pwrite64. this was simply a leftover |
| 145 |
from Linux::AIO. |
| 146 |
(found by compiling on cygwin. Unfortunately, cygwin declares but |
| 147 |
does not implement pread/pwrite). |
| 148 |
- only include <sys/syscall.h> on linux. |
| 149 |
|
| 150 |
0.5 Tue Jul 12 13:03:09 CEST 2005 |
| 151 |
- removed #include <endian.h>, it's not portable, and we don't |
| 152 |
need it anyway. |
| 153 |
- link against -lrt, as it might contain goodies such as fdatasync, |
| 154 |
and often aio_read, which we do not use yet, but it shouldn't hurt. |
| 155 |
|
| 156 |
0.4 Mon Jul 11 05:28:25 CEST 2005 |
| 157 |
- added IO::AIO::flush, IO::AIO::poll. |
| 158 |
- make callback argument optional for all calls. |
| 159 |
|
| 160 |
0.3 Mon Jul 11 01:42:37 CEST 2005 |
| 161 |
- restructured manpage. |
| 162 |
- added SYNOPSIS with examples for Event, Glib, Tk. |
| 163 |
- remove explicit context support so I don't have to maintain |
| 164 |
it for threaded perls (i.e. it should work now :). |
| 165 |
|
| 166 |
0.2 Sun Jul 10 22:57:12 CEST 2005 |
| 167 |
- return immediately from poll_wait if nreqs==0. |
| 168 |
- get rid of the request pipe and use faster thread-thread |
| 169 |
communications for most requests and responses. |
| 170 |
|
| 171 |
0.1 |
| 172 |
- cloned from Linux::AIO, which is very similar. |
| 173 |
|
