Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!rutgers!gatech!ulysses!hector!ekrell
From: ekrell@hector.UUCP
Newsgroups: comp.unix.wizards
Subject: Re: symbolic links are a botch
Message-ID: <2676@ulysses.homer.nj.att.com>
Date: Sat, 20-Jun-87 21:32:34 EDT
Article-I.D.: ulysses.2676
Posted: Sat Jun 20 21:32:34 1987
Date-Received: Sun, 21-Jun-87 12:03:39 EDT
References: <7879@brl-adm.ARPA> <2211@bunker.UUCP>
Sender: daemon@ulysses.homer.nj.att.com
Reply-To: ekrell@hector (Eduardo Krell)
Organization: AT&T Bell Labs, Murray Hill
Lines: 29
Keywords: HUH?

In article <2211@bunker.UUCP> zink@bunker.UUCP (David Zink) writes:
>I'm So confused.  What's wrong about more than eight symbolic loops?
>What's the difference between :
>mkdir foo
>ln -s `pwd` foo/foo
>followed by reference to a pathname with infinite foo/'s in it as compared
>with a pathname with infinite ../'s in it?

How can you have an infinite number of ".."s in a filename unless you
encounter a loop somewhere (either symbolic or hard link)?.

>What's the point of outlawing some loops and not others? And what's more
>important, why invalidate otherwise valid pathnames?

All loops are invalidated (ie, catched). Someone decided that 8 was large
enough for the maximum number of symbolic links encountered during the expansion
of a particular pathname.

>(or is it just the ATT crowd saying lets do the ATT botch that's not like the
>UCB botch?)

It isn't an ATT vs BSD battle. The point is that some people considered
the semantics of symbolic links as implemented in BSD to be a botch. If
symbolic links could be added to System V, should they be broken too
(just to keep BSD compatibility), or should they be done the right way?
    
    Eduardo Krell                   AT&T Bell Laboratories, Murray Hill

    {ihnp4,seismo,ucbvax}!ulysses!ekrell