Re: /etc/adjtime, /etc/timezone, etc.
On Mon, 5 Oct 1998, Ian Jackson wrote:
> Santiago Vila writes ("Re: /etc/adjtime, /etc/timezone, etc."):
> ...
> > Of course it is possible. But the reason policy says some files should not
> > be conffiles is the following: "Doing this will lead to dpkg giving the
> > user confusing and possibly dangerous options for conffile update when the
> > package is upgraded.", which is completely false for conffiles that never
> > change (I mean: conffiles for which the package provided version is always
> > the same). Therefore, if we want to encourage the creation of certain
> > files in the postinst (as opposed to being them conffiles), we need a
> > better rationale (see my bug report, #26402).
>
> Firstly, in situations where a conffile file is created by a script
> before dpkg gets to it, dpkg will still prompt.
Yes, but I was talking about creation of certain files in the postinst "as
opposed to being them conffiles".
> Secondly, you can't guarantee that the file in the package will never
> change.
I can't guarantee anything, because free software comes with no
warranty :-), but if I never change the file in the package, then the file
would never change (as it is happening with /etc/timezone so far).
> Why can't you just handle this in the postinst, without involving dpkg
> ?
Of course I can, I've already done this (in base-files_2.0.1), but I still
think policy needs a little bit more of rationale. I see it is better,
you see it is better, and Manoj sees it is better, but policy does not
explain well enough (IMHO) why it is better, it does not contemplate all
possible cases.
BTW: Please, Cc: me when replying if you do it one month after my
last post, I almost missed this message :-)
--
"68340b7f29bea02be84b717c47a6c0a1" (a truly random sig)
Reply to: