Re: Pathological diff behaviour with mutually-recursive symlinks

[Paul Eggert]

Colin Watson <address@hidden> writes:

> My gut feel is that the best way to fix this would be to check for
> symlinks pointing up in the directory tree but not pointing to a direct
> parent, and skip them if they're underneath a directory which we intend
> to compare recursively anyway.

I don't quite follow this proposal; it sounds like it might generate
incorrect output (i.e., omit non-duplicated diffs) in some cases.

For hairy cases like this, wouldn't it be simpler to add an option to
not follow symlinks?