I've just commenced the work implementing a git reader for Buckminster. Buckminster has a fairly specific use-case that
I'd like your insight on.
When Buckminster provisions a workspace, we often start with a root such as a feature and traverse the dependencies of
that root in order to find all viable projects to check-out. During this traversal, we do a very moderate discovery on a
few files in the remote repository. We typically read files like the feature.xml or the manifest.mf. The important thing
is that we don't need to copy more data then necessary during this phase since it's actually here we determine what
projects we eventually will check-out.
How is this best done with jgit/egit? Is there a small code example somewhere that does something similar?