Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!hao!ames!aurora!labrea!decwrl!pyramid!prls!philabs!micomvax!musocs!mcgill-vision!mouse
From: mouse@mcgill-vision.UUCP (der Mouse)
Newsgroups: comp.unix.wizards
Subject: Re: suggest solutions for "rm *"
Message-ID: <911@mcgill-vision.UUCP>
Date: Thu, 19-Nov-87 05:07:11 EST
Article-I.D.: mcgill-v.911
Posted: Thu Nov 19 05:07:11 1987
Date-Received: Fri, 27-Nov-87 00:05:57 EST
References: <5715@j.cc.purdue.edu>
Distribution: na
Organization: McGill University, Montreal
Lines: 29
Keywords: human factors, interfaces, shells

In article <5715@j.cc.purdue.edu>, ksb@j.cc.purdue.edu (Kevin Braunsdorf) writes:
> Can anyone suggest a solution to the novice user's classic error message:
>
> 	% rm * .o
> 	rm: .o nonexistent

Sure, use rm -f, that way it won't complain :-) :-)

There's the one we use here.  The rm program reads the current
directory and figures out what * would expand to.  If its argument list
contains this list of files, and there are more than four or so files
involved, it asks the user "are you sure you mean to remove all those
files?" before doing so.  There is an option to suppress this behavior,
for use by shell scripts and people who know they said rm * and they
meant it.

This avoids the familiarity/fatigue factor of aliasing rm to rm -i
(there are tales of people who type "rm *" and answer "yes" for every
file on the list because they are so used to doing so).  The question
is asked only when appropriate.

The check for "four or more files" is to avoid questions when removing
just the last file in a directory, for example.

Not perfect (eg, "rm foo/*>o") but it goes a long way.

					der Mouse

				(mouse@mcgill-vision.uViViV