Brandon Williams <bmwill@xxxxxxxxxx> writes: > That makes sense. While at it, would it make sense to ensure that the > 'struct index_state *' which is stored in 'the_repository.index' be > '&the_index'? I was imagining (read: speculating one possible future, without thinking things through to judge if it makes sense at all) that conceptually most of the current API that works on things in the current repository to be implicitly working on the_repository singleton instance in the "repository object" world. When that happens, the_index can become a CPP macro that translates to "the_repository.index", I would think.
