We sat down with Reinhard Max together with Michael (thanks to Edith) to review the database model we were using and why every decision was made. After a long discussion and review, we realized lot of parts were we have to try different approaches and do tradeoffs between complexity/speed/disk space. Brief summary: 1.- We have a new attribute model, that will be used to store dynamically in the database everythng that is not a column itself but a enum'ed string, like relations, languages. This makes the conversion of types between zypp classes and the db obsolete (hard coded enums). So old code can read new values and viceversa. 2.- We will use the attribute model to store all text attributes. This allows to avoid changes in the database schema when new attributes appear. Allows us to name attributes using strings instead of hard coded columns, and makes easy to insert translations and also to search and retrieve. The amount of insert statements to create the cache will be higher but lets see how much. The first version of the API to store the texts is in svn. I am working on the query api so the sources can retrieve the values on demand. I want to do it so you can reuse this query object from any code. again, thanks Reinhard for the help. I havent done much with the source handling, as we havent sat to discuss a concept yet. Duncan -- To unsubscribe, e-mail: zypp-devel+unsubscribe@opensuse.org For additional commands, e-mail: zypp-devel+help@opensuse.org