Re: volatile data in packages
On Wed, Aug 04, 2004 at 03:52:53PM +1000, Russell Coker wrote:
> Several times there have been suggestions for exclusions to the policy
> regarding new packages in stable for certain types of data. One example is
> virus definitions for scanners. Another example that comes to mind is the
> configuration files for postgrey (which contain data about the behaviour that
> has been discovered in mail servers on the Internet - an upgrade or policy
> change to a server could require changes to the configuration files).
>
> Maybe it would be a good idea to have a special category of package for such
> data. Such packages would have only data files (no scripts or binaries) and
> only the very simplest pre/postinst scripts (such as restarting the daemon in
> question).
>
> Of course such package splits would be convenient even if we don't have any
> special treatment (it would make it easier for back-ports repositories, and
> reduce the amount of data that changes on stable systems). But I think that
> it would be good if such config files could be changed in stable.
>
I agree. They require a specific source/binary package and probably some kind of
categorization (volatile/optional?). For a better management source should be
separated from the rest of the package (e.g. the antivirus engine).
--
Francesco P. Lovergine
Reply to: