On 13/04/2020 17:02, Drew Parsons wrote: > Thanks for the analysis, Mattia. > > That processing output certainly is hard to parse. Does it mean those > packages > listed for s390x at the end of the dataset are the ones which are making the > problem?
Yes. Those packages would become uninstallable if the set listed before migrated to testing. Some notes to consider as well: - s390x was not the only problem, just an example. britney picks a random architecture to do the first tests, and if they fail it stops there, but if they pass it will check every architecture to ensure that nothing breaks - arch:all packages are only checked on amd64 and i386 atm so there likely were some more uninstallable packages that were not reported there. Emilio
