14 Mar
2025
14 Mar
'25
1:15 p.m.
On Fri Mar 14 13:09:19 2025 +0000, Jinoh Kang wrote:
On second thought, actually such rename would still be no-op even if we got to the the codepath below.
Sorry, I was mistaken, it wouldn't be a no-op.
To be honest I'm not sure how frequent such scenario would be in practice. I also wonder what the semantics for bind mounts should be in practice--perhaps like directory junctions?
In any case it is not immediately clear `if (!is_dir)` check is correct, or serves as a performance advantage. Making the directory comparison unconditional should simplify the logic too.