Hi, On Thu, 17 Apr 2008, Duncan Mac-Vicar Prett wrote:
- does repo2solv parse the right languages? -> should the logic be there or should the caller specify which ones?
I'm undecided here. What will libzypp download? All available packages.* files, or only a subset of them? My concern is that if I simply convert all files that libzypp happened to download (are in the raw/ directory) it might include stale packages.* files, and will take up much space. So it would be kind of nice if repo2solv gets a list of packages from libzypp that it thinks are the current ones (and that it made sure are also current on disk). Another thing we need to do for this to become nice, is to split the solv files into multiple ones. That is already implemented and needs only a little bit of support code in libzypp (I'll do that). But it also would be nice to reorganize where we store these .solv files (and subfiles). I would suggest storing them into a directory per repo. Currently we have (let $N be the alias of the repo): raw/$N/* : downloaded files $N.cookie : generated from download $N.solv : " I would like to reorganize this to: raw/$N/* : as before $N/solv : main solv file for $N $N/cookie : you guessed it $N/solv.DU : disk usage data $N/solv.$LANG : the several translated strings $N/solv.FL : filelist (exists only sometimes) $N/solv.misc : (optional) could contain the non-solvable attributes With the correct callback in libzypp the actual split of attributes over different files will not matter, it won't even notice (they are loaded on demand). It doesn't even need to know the names of the subfiles. Ciao, Michael. -- To unsubscribe, e-mail: zypp-devel+unsubscribe@opensuse.org For additional commands, e-mail: zypp-devel+help@opensuse.org