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

Bug#2831: timezone conffiles vs. postinst



alegre@mars.superlink.net writes ("Bug#2831: timezone conffiles vs. postinst"):
> In article <[🔎] m0uFhJ7-0002YGC@chiark.chu.cam.ac.uk>, you wrote:
> >If you insist that you're not going to change this then obviously the
> >Guidelines are not clear enough and I shall have to edit them to make
> >what you're doing clearly forbidden and then reopen the bug report.
>
> Please do it if you think it is necessary. But please explain not only
> what they are not for but also what they _are_ for.

The dpkg conffiles mechanism is for the large number of files that the
user might want to edit and which shouldn't therefore be simply
overwritten during upgrades, but for which it's perfectly sensible for
most users just to ship and install a single standard version and have
it upgraded automatically.

If the file *needs* local configuration then you can if you must leave
it to the local admin - distributing a template using the dpkg
mechanism.  This is not a good solution, but can get a package
released - better than no package.

Better is to do almost what you're doing, using the postinst script to
create an appropriate file.  However, if you do this then you should
*not* include a copy of the file in the package and *not* use dpkg's
conffiles mechanism.  If you do you'll almost always find that dpkg's
mechanism and your postinst will interfere with each other.

Do you follow me ?

> >If you'd like to explain to me some reasons why I shouldn't do this
> >please go ahead.  I'm certainly willing to listen, but I'm afraid I'm
> >not going to accept `matter of taste' because I don't think it is an
> >answer to my comments, which are based not on taste but on the
> >intended use and actual effect of the dpkg conffiles mechanism.
>
> Could you please explain why you think that /etc/passwd from the base package
> and /etc/timezone from the timezone package should behave differently?

/etc/passwd in the base package is mishandled at the moment.
See Bug#2524.

Ian.


Reply to: