Re: [Acl-devel] [ANNOUNCE] attr 2.4.47

[Kamil Dudka]

On Tuesday 21 May 2013 14:33:36 Christoph Anton Mitterer wrote:
> Hi Kamil.
> 
> On Tue, 2013-05-21 at 10:22 +0200, Kamil Dudka wrote:
> > Sorry, I was not aware of your bug report at all.  It was supposed to
> > fix https://bugzilla.redhat.com/660619 and I believe the bug is fixed.
> 
> No worries... :)
> 
> > > While now you use an exit status, it doesn't seem to be a reasonable
> > > one.
> >
> > It is quite common to return a non-zero exit code if an error occur,
> > isn't it?
> 
> Sure... I just wouldn't classify a missing xattr a classical error... it
> personally consider it more like a "Are you living on the northern
> hemisphere? Yes/No" Q&A.... neither of the two answers would be an
> error.
> An error would be the "answer" "sorry I don't understand you and
> therefore cannot answer your question".
> 
> > > 3) That an attribute is missing, is however not real error, is it?
> >
> > Definitely not something we should silently ignore I guess.
> 
> Sure... I'd just have used different error codes, with specifying what
> each of them means like
> 1: general error (e.g. everything not yet caught, or wrong command line
> options or no xattr support)
> 2: the attribute is simply not set
> 3: ...
> 
> At least this would make scripting a lot easier in many cases ;)

Providing more fine-grained error codes sounds useful.  Is there any similar 
utility that uses the error codes above?  It would be good to make the error 
codes compatible with other utilities that use them already.

Kamil