Thanks to all who answered my questions. So, to summarize (and ensure I understood all answers correctly):
- For packages which don't have intermediate files (things that don't need to be built, like scripts, static data, configuration files, or whatever), it's perfectly fine to have a single upstream branch, and feed the pristine-tar branch with the raw pristine-tar command - For the rest, it's recommended, but not mandatory, to have two distinct upstream branches, one tracking upstream git, and the other managed by gbp import-orig; using a single upstream branch is feasible, but would increase the size of delta files in the pristine-tar branch if the tarball's content differs from the upstream branch's one - Avoiding to "pollute" the upstream branch with "Imported upstream version..." messages isn't possible yet with gbp import-orig (no one answered about this: would it make sense to file a wishlist bug against gbp to ask for such an option ? The goal being to automate tarball download and import, without touching the upstream branch at all, using a single "gbp import-orig --uscan" instead of both "uscan" first and then "pristine-tar <tarball> <upstream branch>) - In any case, if the branches layout differs from the classic "master/upstream/pristine-tar" model, it should be documented in a README.source file Did I get all of your answers right ? Regards, -- Raphaël Halimi
signature.asc
Description: OpenPGP digital signature