Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!iuvax!bsu-cs!dhesi
From: dhesi@bsu-cs.UUCP (Rahul Dhesi)
Newsgroups: comp.unix.questions
Subject: Re: Computational complexity of rm & ls
Message-ID: <6124@bsu-cs.UUCP>
Date: 14 Mar 89 23:34:43 GMT
References: <9000012@m.cs.uiuc.edu> <4461@pt.cs.cmu.edu> <1541@zen.UUCP> <871@Portia.Stanford.EDU> <TALE.89Mar13170500@imagine.pawl.rpi.edu> <1280@itm.UUCP>
Reply-To: dhesi@bsu-cs.UUCP (Rahul Dhesi)
Organization: CS Dept, Ball St U, Muncie, Indiana
Lines: 34

In article <1280@itm.UUCP> danny@itm.UUCP (Danny) writes
[about cleaning up runaway directories]:

>    The best "stragety" (thanks, Bugs) I've ever heard of is:
>        use clri to zap the directory,
>        and fsck to clean up the mess.

VAX/VMS has a consistent command structure and the switch /nodirectory
always, always means "not a directory".

Therein lies a tale.

There is a command "set file" that does many things to files, and it
accepts many options while doing those things.

The manual documented "set file/nodirectory" to mean "act on these
files only if they aren't directories", but the command interpreter
understood it to mean "make sure these aren't directories."  So if you
did

     $ set file/nodirectory *.*

all subdirectories matching *.* suddenly became regular files.  In
effect, we had a command that was (a) accessible to ordinary users and
(b) documented to be harmless but (c) did a near-clri on specified
directories.

Unfortunately, the same unprivileged users couldn't do the equivalent
of fsck (analyze/disk or something like that).

I think they fixed it.  I didn't try it again.
-- 
Rahul Dhesi         UUCP:  <backbones>!{iuvax,pur-ee}!bsu-cs!dhesi
                    ARPA:  dhesi@bsu-cs.bsu.edu