[asterisk-dev] [Code Review] inotify(7) support for pbx_spool
Sean Bright
sean.bright at gmail.com
Tue Feb 17 18:27:41 CST 2009
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://reviewboard.digium.com/r/137/#review456
-----------------------------------------------------------
/trunk/configure.ac
<http://reviewboard.digium.com/r/137/#comment1089>
I know that technically this functionality is in libc, but I think making the argument to configure be --with-inotify would be less confusing (right now users will see --with-c). The last argument can just be changed from [c] to [inotify] and everything else will just work.
/trunk/pbx/pbx_spool.c
<http://reviewboard.digium.com/r/137/#comment1090>
We might want to alloca/malloc this since 256 is an arbitrary (but for most cases sufficient) limit. Either is better than using PATH_MAX.
/trunk/pbx/pbx_spool.c
<http://reviewboard.digium.com/r/137/#comment1091>
Extracting the ternary part of this into a local might help de-verbosify this statement a bit.
- Sean
On 2009-02-12 19:06:24, Tilghman Lesher wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://reviewboard.digium.com/r/137/
> -----------------------------------------------------------
>
> (Updated 2009-02-12 19:06:24)
>
>
> Review request for Asterisk Developers.
>
>
> Summary
> -------
>
> Instead of scanning the directory and trying to keep tabs on the mtime of the directory, an approach which has already been shown to have an issue with directory entry races, use the kernel builtin method of inotify(7) to identify every file as it gets created in the spool directory.
>
>
> Diffs
> -----
>
> /trunk/configure 175312
> /trunk/configure.ac 175312
> /trunk/include/asterisk/autoconfig.h.in 175312
> /trunk/pbx/pbx_spool.c 175312
>
> Diff: http://reviewboard.digium.com/r/137/diff
>
>
> Testing
> -------
>
> Compile-time testing only.
>
>
> Thanks,
>
> Tilghman
>
>
More information about the asterisk-dev
mailing list