On Aug 3, 11:36am, James Graham (Captech) wrote:
} Subject: Re: Symlink ownership
} The idea of stuffing a symlink in as a "new kind of directory entry"
} will serve only to muddy the current directory structure, and for things
} which don't need stats and don't even necessarily need the object name
} of the symlink, it will only add performance overhead.  Minimal, but still,
} why do it?

The performance degradation may be more than minimal.  Picture a directory
with a few thousand near maximal length (1K) symlinks.  Any operation
that wants to follow the last symlink in this directory has to first
scan over n-1 symlink names *and their contents* before it gets to the
symlink it wants.  It would seem to me that having to scan the contents
of a postively humungous directory but have quick access to some inline
data at the end would be slower than scanning a merely large directory
and doing a couple of extra seeks and reads.

			---  Truck