| … | |
… | |
| 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: fix setsig |
6 | TODO: fix setsig |
| 7 | TODO: splice/tee/vmsplice/sync_file_range? (http://kerneltrap.org/node/6505 http://lwn.net/Articles/178199/) |
7 | TODO: splice/tee/vmsplice/sync_file_range? (http://kerneltrap.org/node/6505 http://lwn.net/Articles/178199/) |
| 8 | |
8 | |
| 9 | 2.4 |
9 | 2.51 Sat Oct 6 16:04:54 CEST 2007 |
| 10 | - fix aio_busy w.r.t. fractional delays. |
10 | - perlio isn't generally threadsafe, so aio_close could not work reliably. |
| 11 | - add aio_chmod, aio_chown, aio_utime (linux successfully demonstrated |
11 | aio_close now tries a safe hack that might not be that asynchronous |
|
|
12 | (see the manpage for details). |
|
|
13 | - discard callback return values as to not grow the stack endlessly |
|
|
14 | when poll_cb handles many requests without returning. |
|
|
15 | - minor code cleanups. |
|
|
16 | |
|
|
17 | 2.5 Thu Oct 4 14:49:08 CEST 2007 |
|
|
18 | - replaced _fd2fh with faster xs code. |
|
|
19 | - aio_close will now try to do "the right thing" and thus might |
|
|
20 | work sensibly for the very first time. |
|
|
21 | |
|
|
22 | 2.41 Mon Sep 24 21:28:21 CEST 2007 |
|
|
23 | - after fork, invest some work to make sure that the poll_fileno |
|
|
24 | stays stable (by dup2'ing the new fd over the old one), to make |
|
|
25 | it easier for programs/libs that don't control their forking |
|
|
26 | behaviour and cannot recreate their watchers. |
|
|
27 | |
|
|
28 | 2.4 Sun Aug 5 18:44:22 CEST 2007 |
|
|
29 | - add aio_truncate, aio_chmod, aio_chown, aio_utime (linux |
| 12 | that you can block on futimes...). |
30 | successfully demonstrated that you can block on futimes...). |
| 13 | - allow undef as fileoffset for aio_read/write and use read/write(2) |
31 | - allow undef as fileoffset for aio_read/write and use read/write(2) |
| 14 | internally (useful for sockets or O_APPEND handles). |
32 | internally (useful for sockets or O_APPEND handles). |
|
|
33 | - allow undef for length in aio_write. |
|
|
34 | - negative dataoffsets work as expected now in aio_read/aio_write. |
| 15 | - use NV instead of UV for 32 bit perls and file offsets, as NVs |
35 | - use NV instead of UV for 32 bit perls and file offsets, as NVs |
| 16 | have a larger range then. |
36 | have a larger range then. |
| 17 | - shared code between BDB and IO::AIO. |
37 | - shared code between BDB and IO::AIO. |
|
|
38 | - aio_busy was completely broken. now fixed. |
|
|
39 | - readahead emulation now returns something. |
| 18 | - correctly set errno to ENOSYS on unimplemented functions |
40 | - correctly set errno to ENOSYS on unimplemented functions |
| 19 | (should never happen, though). |
41 | (should never happen, though). |
|
|
42 | - large changes to make it partially compile and run on win32, |
|
|
43 | but win32 lacks too much functionality, and perl overrides way |
|
|
44 | too many functions with crashing versions. |
| 20 | |
45 | |
| 21 | 2.33 Tue Jan 23 23:55:41 CET 2007 |
46 | 2.33 Tue Jan 23 23:55:41 CET 2007 |
| 22 | - fix install path (Andreas J. Koenig). |
47 | - fix install path (Andreas J. Koenig). |
| 23 | |
48 | |
| 24 | 2.32 Mon Jan 22 16:56:23 CET 2007 |
49 | 2.32 Mon Jan 22 16:56:23 CET 2007 |
