Re: [PATCH] implement a more efficient queue file format

[Alan Jenkins <alan-jenkins@tuffmail.co.uk>]

Kay Sievers wrote:
> On Thu, May 21, 2009 at 20:27, Alan Jenkins <alan-jenkins@tuffmail.co.uk> wrote:
>   
>> Instead of creating a file for each event, append their details to a
>> log file.  The file is periodically rebuilt (garbage-collected) to
>> prevent it from growing indefinitely.
>>     
>
> Sounds cool. Boots fine here. :)
>
> I've switched a few int to unsigned int, to silent some gcc warnings,
> merged the few entries in queue-private.h just into libudev-private.h,
> moved the file format comments to the C file, renamed the
> seqnums_are_finished() to seqnum_sequence_is_finished() and exported
> it, bumped the minor version of libudev, and comitted it.
>
> Would be great, if you can check, that I didn't mess it up. :)
>
> Thanks a lot,
> Kay
>   

Wow, thanks!  I checked the delta.  I don't see any problems in these 
changes.


I only see one warning myself

lib/libudev-queue-export.c:143: warning: comparison between signed and 
unsigned

142 n = seqnum - udev_queue_export->seqnum_max;
143 if (n >= devpaths->devpaths_size)
144        goto read_error;

Arguably it _might_ be better addressed in this one place, by casting 
devpaths_size to unsigned.  It seems to be a matter of personal taste 
though, and I'm not complaining.


I have an unverifiable memory of reading "unsigned int i is evil" on 
LKML.  Half the internet says unsigned has relatively special semantics 
(intended for e.g. bit manipulation), you should avoid it unless that's 
what you actually need, and size_t should never have been defined as 
unsigned.  Of course, the other half advocate using unsigned liberally 
to exclude negative values when passing array indexes around :-).

Alan