On Sat, Sep 18 2021, David Aguilar wrote: > The "dirlinks" and "growing" repositories should not outlive the > tests that use them. Why not? The pattern of not bothering to clean up the work scratch are is fine as long as it doesn't leave crap around for other tests. See the referenec to "repo" at[1]. I.e. is this needed for later tests that change in this series, or just cleanup in case a future test ever cares that there's an untracked "dirlinks" and "growing" directory at the top-level? 1. https://lore.kernel.org/git/87y27veeyj.fsf@xxxxxxxxxxxxxxxxxxx/
