Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!decwrl!sun!imagen!atari!portal!uunet!mcvax!ukc!warwick!mirk
From: mirk@warwick.UUCP (Mike Taylor)
Newsgroups: comp.unix.questions
Subject: Re: simple question about mv
Message-ID: <1063@ubu.warwick.UUCP>
Date: 9 Feb 89 13:59:11 GMT
References: <18231@adm.BRL.MIL> <4093@omepd.UUCP> <626@jonlab.UUCP>
Sender: news@warwick.UUCP
Reply-To: mirk@diamond.UUCP (Mike Taylor)
Organization: Computer Science, Warwick University, UK
Lines: 36

In article <626@jonlab.UUCP> jon@jonlab.UUCP (Jon H. LaBadie) writes:
>Randal and others (this is not intended to pick on Randal's suggestion)
>have proposed work-arounds to UNIX's command line interpreters (or to
>mv's) deficiency in wild card matching.
[He is referring to "wrong" behaviour of commands such as "mv *.flip *.flop"]

Hmmm, you'd like to blame this one on UNIX's wild-card handling, would
you?  Well, think about it.  In UNIX, the shell (whichever you choose
to use) does *all* wildcard expansion, and passed the result to
whatever commands are invoked.  There is, therefore, a single, well-
defined and well-understood patern of behaviour.  The alternative,
and this is what MSDOS does, is for each command to include its own
pattern-matching code, so that, firstly, the wheel is re-invented over
and over, and secondly, and IMHBDO much more seriously, different
commands may expand wildcards in different ways.

What you suggest is an example of this: The "*" in the mv(1) command
is required to act differently from that in, for instance, the command
"cp file1 file2 file3 dir", or "cp file* dir".  This sort of
incosistency, while it makes things more immediately intuitive for the
novice, means that there is no real understanding of how such things
will work in a general case - surely not a good state of affairs.

Sorry if this sounds like flamage, you do make some good points in
your article, (eg: that all the proposed solutions seem to involve
invoking a new process other than the mv(1)s themselves)  But it is
important to have consistency, and I think that UNIX's approach to
pattern matching is optimal.  Let's all remember Henry Spencer's
signature, shall we?

	"Those who do not understand UNIX are condemned to
	 re-invent it, poorly."
______________________________________________________________________________
Mike Taylor - {Christ,M{athemat,us}ic}ian ...  Email to: mirk@uk.ac.warwick.cs
Unkle Mirk sez: "You fritter and waste the hours in an offhand waistcoat." :-)
------------------------------------------------------------------------------