[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#379089: Configuration file shadowed?



Christoph Berg <myon@debian.org> wrote:

> Re: Frank Küster 2006-07-21 <86slku93ym.fsf@alhambra.kuesterei.ch>
>> I believe that we need to rephrase the TeX Policy.  But this requires
>> not just to specifiy that each "cfg" file must be in /etc.  Instead, I
>> think we need to find a distinction between
>> 
>> - files that can be used to modify the behavior of programs, and/or
>>   files that make sense to customize site-wide behavior on a multiuser
>>   system (I just cannot find an example of a file that would only
>>   fulfill the second half of the sentence)
>> 
>>   => must go to /etc
>
> I remeber that a tetex update last year (maybe even longer ago) had
> prompted me and several others for about a dozen changed conffiles I
> had never heard of before and I surely had never touched. Similarly, I
> have seen some debconf prompts that asked me about tex settings I
> wouldn't ever use. I might have run at the wrong debconf priority
> there, but the point is: imho the vast majority of tetex users will
> never touch any of the settings and be happy with the defaults, so
> please make sure that any new conf(ig) file you introduce does not
> prompt the user. 

Exactly these things caused us to make the move.  I admit, it was also
caused by design errors we (or even former maintainers in the old times)
made, and it might be possible to have a package with 300 configuration
files in /etc.  I don't expect that we're going to manage it properly,
though, especially as long as ucf's functionality isn't integrated with
dpkg.  

> Likewise, if you are going to overwrite a file
> outside of /etc on upgrade that one out of 1000 users might have
> touched, please don't prompt, but tell the users in README.Debian to
> use dpkg-divert. Imho, in this case, customizing for the
> might-have-changed case doesn't serve the (other) users who are just
> confused by the questions.

In this case, as has been pointed out, we were wrong in treating the
file like any TeX input file, and it should have been a conffile.
Probably unchanged by 99.99% of the users.  

Regards, Frank
-- 
Frank Küster
Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich
Debian Developer (teTeX/TeXLive)



Reply to: