![](https://seccdn.libravatar.org/avatar/e1d3f322373a851c42e21dd2264df481.jpg?s=120&d=mm&r=g)
On Mon, Nov 10, 2008 at 01:50:32 +0100, Carlos E. R. wrote:
It should do the same as Yum, IMO: download metadata + packages, and when that fails at any point, you'd refresh the local metadata and start over, and once you have a consistent state on your system (including packages) _then_ start the actual update.
That requires double disk space.
With some intelligence it could actually be possible to delete cached packages that are known to be obsolete already, before their new versions are downloaded, no?
And there is no guarantee that we won't have to restart the download many times.
No guarantuee, but this is function of the frequency of factory replacement, and your download speed. Obviously, when factory is replaced every 5 minutes, this is never going to work, unless a lot of snapshots are kept around. So there needs to be some piece and time for updates. The same happens for the buildservice - it's always possible that some repository is being updated while you access it. By the time you press 'y' it could be have been replaced, so you need to start again. The difference to the Factory is that the second attempt is very likely successfull. While with Factory the second attemt is very likely unsuccessful ;) Peter -- "WARNING: This bug is visible to non-employees. Please be respectful!" SUSE LINUX Products GmbH Research & Development