Tian Yuchen [off-list ref] writes:
I must admit I hadn't considered this edge case at all. Thank you for
pointing it out :]
It is easy to see, if you run the tests, though ;-)
I will:
- squash your diff into my patch
- rename the error code to `READ_GITFILE_ERR_STAT_MISSING`
I think MISSING is more appropriate than STAT_MISSING. Our stat(2)
call positively identified that the given path does not exist on the
filesystem.
- combine this with the commit message refinements and test cleanups
suggested by Karthik in the previous thread.
Yeah, cleaning after yourselves in the tests is a good idea.