Hi Gijs, Benda,
I still don't see why this is a big problem yet. What I have seen is
that the only thing that is changed are the leap seconds and the
geomagnetic field. Leap seconds come every 1-2 years or so and are well
known before, and the geomagnetic field data seem to be updated (from
the differences in the time stamp) every few months (twice a year or
I would prefer a date based versioning, since this allows a simple
update. Otherwise, we would have large difficulties to check if the
package got a newer version.
Gijs Molenaar <firstname.lastname@example.org> writes:
> I don't think that works much better than just date based versioning.
> I'm not 100% sure what the update script does, but most of the action is
> in FTP'in to FTP servers around the world and downloading all kind of
> data files. I think a date based versioning schema is better, since then
> you can see how old the data is :)
> But other problems are:
> * how do we test if the data is valid?
> * What do we do if some of the data is missing or corrupted?
> * How can we be sure the data retrieved from all the FTP servers is up
> to date?
> * What do we do when the data is not valid?
I think this all is left to the packager (and should be not too
difficult to solve manually on each new version)
I would however add as potential problem:
* what is the license of the data? (DFSG compliance is required)