Hello Andreas,
On Wed, Jul 10, 2013 at 3:12 PM, Andreas Tille <[email protected]> wrote: > > I did not found the time to dive into theses details but I think there > is no harm done to add an extra column to the importer. If it turns out > helpful (which sounds probable to me when considering this discussion) > that's fine. If we find an even better solution afterwards it is cheap > to drop it again in worst case. > > So feel free to add the patch to the importer and I'll test and implement > it. I have attached the patches. For each package check you perform I just added an extra check using the "provides" column. This check gets active once the initial check fails, for example: Let's take the case where you check for the 'djvu-viewer' package (which is virtual) First you do: * EXECUTE blend_check_existing_package('djvu-viewer') This execute call won't return anything. Now instead of moving in the next : "blend_check_package_in_new" it goes to: * EXECUTE blend_check_existing_package_provides('%djvu-viewer%') --( I use "like" condition in where clause because "provides" might contain comma separated values). This now returns a result so we set the contain_provides = 'true', if it fails(for another package) then it moves on to the next "blend_check_package_in_new" then goes to "blend_check_package_in_new_provides" and so on. It might look a little dirty but I think it will work. Kind regards Emmanouil
blends_metadata_gatherer.py.patch
Description: Binary data
setup.sql.patch
Description: Binary data
