Bug#246818: Deleting conffiles in maintainer scripts and the need to resurrect them later...
Florent Rougon <f.rougon@free.fr> schrieb:
> Hi,
>
> Frank Küster <frank@debian.org> wrote:
>
>> - Do not remove those files any more, just let them lie around,
>> cluttering the directory. Ugly, but we are on the safe side should we
>> ever need one again
>
> OK. Perhaps one should modify the version shipped in the package in this
> case to add a comment saying the file is not used anymore. Yes, this
> will cause a modified conffile prompt, but what is the more annoying
> behavior?
dpkg will only prompt when the user has changed the file. In this case
it is a good thing to prompt him or her, because likely the changes need
to be inserted elsewhere.
>> - Move the files to some other directory (like
>> /etc/texmf/dvips/obsoletemaps/), being able to copy them back in
>> future postinst, should we need them. Complicated and perhaps
>> error-prone, but it gives a cleaner /etc/texmf/dvips.
>
> Mmm, perhaps I didn't understand you. You mean, moving a conffile in a
> maintainer script?
Yes, moving instead of removing - and we can move it back in postinst if
we need it again after an upgrade.
>> Of course this does not apply for the generated map files and for the
>> config* files.
>
> Well, since they should be automatically generated and should therefore
> not be manually modified under "normal" conditions, one could argue they
> shouldn't be visible under /etc either... Just like updmap.cfg that is
> found under /var/lib/texmf/web2c/ and *not* under /etc.
Yes. But then there's people who argue that "You can read this is a
configuration file everywhere!" and demand files to be moved to
/etc. Like it was with texmf.cnf.
In the case of texmf.cnf I agreed, because I think it is true that many
texts advice to edit this file. With generated map files, I don't agree,
and from a personal point of view I would be glad to move it to var. I
haven't thought about possible technical solutions, because, before
doing this, we should consider the "social problems": Do we want to
argue with people annoying us with bug reports?
This is related to #213310.
Regards, Frank
--
Frank Küster, Biozentrum der Univ. Basel
Abt. Biophysikalische Chemie
Reply to: