Re: fakeroot problem

[Roland McGrath]

> Can you think of any specific failure case I could test, or is it just a
> vague feeling?  

I can't think of any case reasonable other than the translated underlying
node itself that would hit the new code.  Maybe firmlink, but that would
only come up if there was a firmlink to a fakeroot virtual node, rather
than a firmlink in the underlying filesystem--and that should not be
possible.  So right now I can't think of anything other than a filesystem
intentionally playing games to confuse fakeroot that would be relevant.

> Also, do you want me to test the changes individually?  I have only
> checked both at once so far.

Only if there appears to be a problem, then to narrow it down.