Fujii Masao <masao.fujii@gmail.com> writes:
> On the second thought, I'm thinking that it might be overkill to add
> such not simple code for that small benefit.
Yeah --- there's a limit to how much code we should expend on detecting
this type of error. It's not like the case is all that plausible.
One idea that I don't think got discussed is stat'ing the two directories
and verifying that their dev/inode numbers are different. I don't know
how portable that is though.
regards, tom lane
