On 15 December 2014 at 13:09, Hedayat Vatankhah <hedayat.fwd(a)gmail.com> wrote:
1. If PK really needs its own *copy* of the cache, that's OK
(well, not OK
but acceptable), but IMHO it should not download it independently too.
But yum/dnf only download the files it needs for the single operation,
and not all the "matching" files, unless you're using "dnf
makecache"
or something like that.
preferably using hardlinks if possible
Yes, hardlinks help for the space-on-disk problem.
2. I believe that the use should know, and more importantly be able
to
control WHEN the repo data is being updated. At the very least, he should be
able to specify if the updates are automatic or not using a very user
friendly method (probably during/after the installation; or per network
connection).
At the moment the PK front-ends only download when on wifi or wired.
Do you have an actual use-case for per-network configuration?
3. I think the repository data management backend should be separate
from
the frontends (including PK, and dnf cli).
PK isn't a front-end, gnome-packagekit and gnome-software are. PK is
just the mechanism.
Richard.