| … | |
… | |
| 3 | TODO: maybe add mincore support? available on at least darwin, solaris, linux, freebsd |
3 | TODO: maybe add mincore support? available on at least darwin, solaris, linux, freebsd |
| 4 | TODO: openbsd requires stdint.h for intptr_t - why posix? |
4 | TODO: openbsd requires stdint.h for intptr_t - why posix? |
| 5 | |
5 | |
| 6 | TODO: make mtouch/readdir maybe others cancellable in-request |
6 | TODO: make mtouch/readdir maybe others cancellable in-request |
| 7 | TODO: fadvise request |
7 | TODO: fadvise request |
|
|
8 | TODO: fdopendir/utimensat |
|
|
9 | TODO: maybe work around 3.996gb barrier in pread/pwrite as well, maybe readahead etc.? |
| 8 | 1.0 |
10 | 1.0 |
|
|
11 | - for simple request api, initialise result/errorno to -1/ECANCELED. |
|
|
12 | - fix a deadlock where a wakeup signal could be missed when |
|
|
13 | a timeout occured at the same time. |
| 9 | - use nonstandard but maybe-working-on-bsd fork technique. |
14 | - use nonstandard but maybe-working-on-bsd fork technique. |
|
|
15 | - use fewer time() syscalls when waiting for new requests. |
|
|
16 | - fix a path-memory-leak in readdir when using the wrappers |
|
|
17 | (reported by Thomas L. Shinnick). |
| 10 | - support a max_idle value of 0. |
18 | - support a max_idle value of 0. |
| 11 | - support setting of idle timeout value (eio_set_idle_timeout). |
19 | - support setting of idle timeout value (eio_set_idle_timeout). |
| 12 | - readdir: correctly handle malloc failures. |
20 | - readdir: correctly handle malloc failures. |
| 13 | - readdir: new flags argument, can return inode |
21 | - readdir: new flags argument, can return inode |
| 14 | and possibly filetype, can sort in various ways. |
22 | and possibly filetype, can sort in various ways. |
| … | |
… | |
| 44 | - fix eio_custom prototype. |
52 | - fix eio_custom prototype. |
| 45 | - work around a Linux (and likely FreeBSD and other kernels) bug |
53 | - work around a Linux (and likely FreeBSD and other kernels) bug |
| 46 | where sendfile would not transfer all the requested bytes on |
54 | where sendfile would not transfer all the requested bytes on |
| 47 | large transfers, using a heuristic. |
55 | large transfers, using a heuristic. |
| 48 | - use libecb, and apply lots of minor space optimisations. |
56 | - use libecb, and apply lots of minor space optimisations. |
|
|
57 | - disable sendfile on darwin, broken as everything else. |
|
|
58 | - add realpath request and implementation. |
|
|
59 | - cancelled requests will still invoke their request callbacks. |
|
|
60 | - add fallocate. |
|
|
61 | - do not acquire any locks when forking. |
|
|
62 | - incorporated some mingw32 changes by traviscline. |
|
|
63 | - added syncfs support, using direct syscall. |
|
|
64 | - set thread name on linux (ps -L/Hcx, top, gdb). |
|
|
65 | - remove useless use of volatile variables. |
|
|
66 | - fix memory leak when reaping threads. |
|
|
67 | - use utime now uses nanosecond resolution on posix 2008 systems. |
|
|
68 | - allow taking advantage of posix 2008 xxxat functions and fdopendir |
|
|
69 | by implementing a working directory abstraction. |
|
|
70 | - make readahead emulation behave more like actual readahead by never failing. |
|
|
71 | - added EIO_LSEEK (untested). |
|
|
72 | - added EIO_FALLOC_FL_PUNCH_HOLE. |
|
|
73 | - wtf. etp_proc returned 0, and no compiler ever complained. |
|
|
74 | - remove pread/pwrite emulation, as the only system that lacked them |
|
|
75 | (cygwin) provides them for a while now. |
|
|
76 | - provide pread/pwrite implementations for win32. |
|
|
77 | - implement aio_realpath for windows. |
|
|
78 | |
