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

Re: Regularly updating data files

* Ole Streicher [2014-04-04 15:33:22 +0200]:
> for an astronomy library package, there is a data library needed to be
> downloaded. This data files are updated ~weekly; however I have the
> feeling that most updates are really minor.

I'd argue against forcing updates on users: even if most of them will want
the latest version most of the time, there will be exceptions. Is there at
least a mechanism for users to override the default data library location
on a run-by-run basis?

> The question is now how to model this in Debian. I see three ways here:
> * provide a package that contains the data and update this on a regular
> base. Disadvantage is that the package is not updated on stable and
> other releases (Ubuntu), so it will be outdated probably already when
> the release is published

Indeed. However, it may still be useful as a fallback. A pure data package
with no dependencies would also be rather easy to upgrade (or downgrade)
from an alternate repository (stable-backports, an Ubuntu PPA, etc).

> * provide a shell script that does the download (this is already
> included in the package). Disadvantage is that it requires some action
> by the user and may be forgotten so that he stays with old data.
> * install a cronjob that runs the script every week or so. Disadvantage
> is that this may be considered as privacy breach.

You should let debconf ask for permission to enable it.

> * create an extra "-data-downloader" package that installs the cronjob
> and states this clearly in its description. Suggest this in the main
> package.

I think that's only needed if the dowloader infrastructure takes up
significant additional space; otherwise the debconf approach is just
as good.

Reply to: