From: bsa@kf8nh.wariat.org (Brandon S. Allbery) Subject: Re: links (was Re: anonymous ftp 0.99pl1) Date: Thu, 2 Sep 1993 01:40:07 GMT
In article <CCp75G.HCF@cs690-3.erie.ge.com> teffta@cs690-3.erie.ge.com (Andrew R. Tefft) writes:
>Both symlinks and hard links have their place. Given the choice
>I almost always choose symlinks for reason #1. I take it you choose
>hard links. Why the prejudice?
I can't speak for Jon, but I prefer hard links because
(1) if you remove the "wrong" end of a hard link, you can put it back; hard
links have no "wrong" ends. Remove the wrong end of a symlink and you're left
with a dangling symlink and no real file. Why should I have to check, when
hard links do the "right" thing already?
(2) different programs treat symlinked directories in different ways,
resulting in LOTS of confusion. Yes, some shells which do this can be
configured. But does *every* program that cares about pathnames *really* have
to check for symlinks and for configuration information on how to treat them?
My experience says they must, or else Bad Things can happen; my sanity, and my
preference for "clean" implementations, say "Gack!!!".
(3) Similar to #1: try symlinking to more than whatever depth the kernel's
been configured for and watch things break. And sometimes you can't rearrange
the symlinks to fix it, if you insist on using them. Frankly, I think
loopback NFS mounts are a cleaner way to handle the cases that have led to
this in my experience.
++Brandon
-- Brandon S. Allbery kf8nh@kf8nh.ampr.org bsa@kf8nh.wariat.org "MSDOS didn't get as bad as it is overnight -- it took over ten years of careful development." ---dmeggins@aix1.uottawa.ca