Automatically creating repackaged upstream tarballs for Debian

One of the less exciting points in the day of a Debian Developer is the moment they realize they have to create a repackaged upstream source tarball.

This is often a process that they have to repeat on each new upstream release too.

Wouldn't it be useful to:

  • Scan all the existing repackaged upstream source tarballs and diff them against the real tarballs to catalog the things that have to be removed and spot patterns?
  • Operate a system that automatically produces repackaged upstream source tarballs for all tags in the upstream source repository or all new tarballs in the upstream download directory? Then the DD can take any of them and package them when he wants to with less manual effort.
  • Apply any insights from this process to detect non-free content in the rest of the Debian archive and when somebody is early in the process of evaluating a new upstream project?

Google Summer of Code is back

One of the Google Summer of Code projects this year involves recursively building Java projects from their source. Some parts of the project, such as repackaged upstream tarballs, can be generalized for things other than Java. Web projects including minified JavaScript are a common example.

Andrew Schurman, based near Vancouver, is the student selected for this project. Over the next couple of weeks, I'll be starting to discuss the ideas in more depth with him. I keep on stumbling on situations where repackaged upstream tarballs are necessary and so I'm hoping that this is one area the community will be keen to collaborate on.