> > > > Yeah. Killing the_index is just the first small step (didn't look that > > small when I started). Now it's all about the_repository ;-) and we > > have like 400 references of it just in library code. > > I suspect it is much worse than that, even. There are many spots that > likely are relying on global data that _should_ be in the repository > struct but aren't yet. I don't think there's even an easy way to count > those. ;)
This is a very interesting part of the discussion, please note the series "RFC Bring the_repository into cmd_foo"[1] as that proposes one way how to deal with this, exposing the repository/index changes into the existing test suite. [1] https://public-inbox.org/git/20181018183758.81186-1-sbel...@google.com/ I'll read on.
