I think the general consensus is that symbolic links don't have owners
or modes, per se as per the 4.4 BSD decision (pardon the misnomer)
for symlinks not to have inodes; therefore (under the current paradigm)
it would be moot to treat symlinks differently; symlinks are already
followed.

Not that I necessarily agree with this standpoint as symlinks could
be used for extra permissions.

That, and the fact that symlinks inherit the inode of the parent directory
have tossed extra code into the getcwd()/getwd() routines (I forget which
is used more prevalently) to do a stat() instead of just scanning the
directory for a matching entry.

At least that's my take on it; I assume someone will (gently) correct me
if I'm wrong.

I certainly don't see the current paradigm changing any time soon as most
folks seem to see symlinks as a necessary evil which should not be given
any more resources than absolutely necessary to carry out their function.

				--*greywolf;